studies.jsonl

{"study_id": "1", "original_paper_pdf": "1/input/original_paper.pdf", "initial_details": "[CLAIM]\nAn interquartile increase in support for Trump (I.Q.R. = 20.3%) resulted in a 4.1 percentage point decrease in social distancing (95% C.I. = 3.0–5.2) [ p < 0.001].\n\n[HYPOTHESES]\nAt the level of U.S. counties, support for Donald Trump in the 2016 presidential election will be negatively associated with social distancing behavior.", "replication_data_files": ["1/input/replication_data/county_variables.csv", "1/input/replication_data/kavanagh_analysis.R", "1/input/replication_data/transportation.csv"], "human_preregistration": "1/gt/human_preregistration.pdf", "human_report": "1/gt/human_report.pdf", "expected_post_registration": {"original_study": {"claim": {"hypothesis": "At the level of U.S.counties,support for Donald Trump in the 2016 presidential election will be negatively associated with social distancing behavior.", "hypothesis_location": "The hypothesis is not stated explicitly in the text but the relationship is mentioned in the abstract, introduction and results sections.", "statement": "An interquartile increase in support for Trump (I.Q.R. = 20.3%) resulted in a 4.1 percentage point decrease in social distancing (95% C.I. = 3.0–5.2) [ p < 0.001].", "statement_location": "Page 4, Results section; page 7, Table 1.", "study_type": "Observational"}, "data": {"source": "Unacast (for social distancing. Unacast measures county-level averages of distance traveled per person); American Community Survey (for socioeconomic status, operationalized by income per capita); MIT Election Data and Science Lab (political preferences, operationalized by the 2016 county-level vote share for President Trump); Census (for rurality).", "wave_or_subset": "Unacast: March 19–28, 2020. American Community Survey: 5-year averages from 2014–2018. MIT Election Data and Science Lab: 2016 county-level vote share. Census: 2010.", "sample_size": "3037", "unit_of_analysis": "US counties.", "access_details": "not stated; the authors thank Unacast for providing their social distancing dataset for research use, but no access details are provided.", "notes": "There is a potential for omitted-variable and ecological biases due to aggregate, cross-sectional data. The data do not sample all cell-phone users and do not reflect non-users."}, "method": {"description": "The authors used multivariable OLS regression to assess how socioeconomic conditions and political orientation were associated with COVID social distancing.", "steps": "1. Presumably the first step was to obtain the data from Unacast, American Community Survey, MIT Election Data and Science Lab, and Census. \n2. After cleaning and merging datasets, the authors had to calculate percentage point changes in average mobility.\n3. Then, they conducted bivariate analyses of per capita income and the share of voters supporting President Trump with social distancing (p. 4, Results section).\n4. Finally, they estimated associations between the degree of social distancing and socioeconomic\nand political factors using cross-sectional OLS regressions.", "models": "cross-sectional ordinary least squares regressions", "outcome_variable": "social distancing (measured as change in average mobility from March 19–28, relative to matched days of pre-COVID-19 reference week).", "independent_variables": "income (county level per capita); vote share for Donald Trump (2016 elections, county level)", "control_variables": "percentage male; percentage Black; percentage Hispanic; percentage with college degree; employment in retail; employment in transportation; employment in health, education, social services; percentage rural; age distribution (p. 3, Methods section).", "tools_software": "not stated"}, "results": {"summary": "An interquartile increase in support for Trump (20.3%) was associated with a 4.1-percentage-point decrease in social distancing (95% CI = 3.0–5.2, p < 0.001).", "numerical_results": [{"outcome_name": "social distancing", "value": "4.12", "unit": "percentage point", "effect_size": "not stated", "confidence_interval": {"lower": "3.05", "upper": "5.19", "level": "0.95"}, "p_value": "<0.001", "statistical_significance": "true", "direction": "positive"}]}, "metadata": {"original_paper_id": "https://doi.org/10.1101/2020.04.06.20055632", "original_paper_title": "Association of County-Level Socioeconomic and Political Characteristics with Engagement in Social Distancing for COVID-19.", "original_paper_code": "not stated", "original_paper_data": "not stated"}}}, "expected_post_registration_2": {"original_study": {"claim": {"hypothesis": "Higher county-level support for President Trump in the 2016 election is associated with reduced engagement in social distancing during March 19–28, 2020.", "hypothesis_location": "Abstract (page 2: 'greater Republican orientation were associated with significantly reduced social distancing among U.S. counties') and Introduction (page 3: 'socioeconomic and political determinants of engagement in social distancing among U.S. counties').", "statement": "The study finds that an interquartile increase in county-level support for President Trump in 2016 (20.3 percentage points) is associated with a 4.1 percentage point decrease in social distancing, indicating less reduction in mobility in those counties.", "statement_location": "Results section, page 4 ('An interquartile increase in support for Trump ... resulted in a 4.1 percentage point decrease in social distancing (95% C.I. = 3.0–5.2).') and Table 1 (page 7, row 'Share of Trump voters').", "study_type": "Observational (cross-sectional ecological analysis using ordinary least squares regression on county-level data)."}, "data": {"source": "County-level social distancing data from Unacast based on anonymized cell phone location records, linked with county-level socioeconomic and political data from the American Community Survey (ACS, 2014–2018) and the MIT Election Data and Science Lab.", "wave_or_subset": "Cross-sectional measurement of social distancing for March 19–28, 2020, expressed as change relative to matched days in a pre-COVID-19 reference week, across U.S. counties.", "sample_size": "3,037 U.S. counties with available mobility, socioeconomic, and political data.", "unit_of_analysis": "County (county-level percentage-point change in average mobility).", "access_details": "not stated (the paper notes that Unacast provided the social distancing dataset for research use but does not describe public access procedures).", "notes": "Social distancing is measured as percentage change in average distance traveled per person during March 19–28, 2020 relative to a pre-COVID-19 reference week, using data from 15–17 million anonymous cell phone users per day. Negative values indicate greater social distancing (larger mobility reductions). Analyses adjust for county sociodemographic and labor market characteristics, rurality, and state fixed effects."}, "method": {"description": "The authors used cross-sectional ordinary least squares regressions to estimate how county-level socioeconomic status and political preferences, including Trump 2016 vote share, are associated with the degree of social distancing, measured as changes in average mobility during March 19–28, 2020 relative to a pre-COVID-19 reference week.", "steps": ["Subset data to U.S. counties with valid Unacast mobility metrics and linked ACS and MIT Election Data and Science Lab measures.", "Compute county-level change in average mobility for March 19–28, 2020 relative to matched days of a pre-COVID-19 reference week as the outcome measure of social distancing.", "Specify main exposure variables: county per capita income and county-level share of voters supporting President Trump in the 2016 election.", "Assemble covariates: county percentages of male, Black, and Hispanic residents; age distribution; share of adults with college degrees; shares of workers in retail, transportation, and health/education/social services; and rurality, plus indicators for each state.", "Estimate multivariable ordinary least squares regressions of percentage-point change in average mobility on the main exposures and covariates, including state fixed effects and age-decade controls.", "Interpret the regression coefficient for county Trump vote share as the change in social distancing (percentage-point change in mobility) associated with an interquartile increase in support for Trump."], "models": "Cross-sectional ordinary least squares regression with state fixed effects for percentage-point change in average county mobility.", "outcome_variable": "Percentage-point change in average county mobility during March 19–28, 2020 relative to matched days in a pre-COVID-19 reference week (negative values indicate greater social distancing).", "independent_variables": "County-level share of voters supporting President Trump in the 2016 election (continuous, modeled per interquartile-range increase).", "control_variables": "Per capita income; percentages male, Black, and Hispanic; age distribution (percentage in each decade of life); percentage of adults with a college degree; county shares of employment in retail, transportation, and health/education/social services; percentage rural; and state fixed effects.", "tools_software": "not stated"}, "results": {"summary": "In multivariable models adjusting for sociodemographic characteristics, labor market composition, rurality, and state fixed effects, higher county-level support for President Trump in 2016 is significantly associated with reduced engagement in social distancing: an interquartile increase in Trump vote share corresponds to a 4.1 percentage point decrease in social distancing (less reduction in mobility).", "numerical_results": [{"outcome_name": "Percentage-point change in average county mobility (social distancing) associated with Trump vote share", "value": 4.12, "unit": "percentage-point change in average mobility per interquartile increase in county Trump vote share", "effect_size": "OLS regression coefficient = 4.12", "confidence_interval": {"lower": 3.05, "upper": 5.19, "level": 95}, "p_value": "<0.001", "statistical_significance": 1, "direction": "positive"}]}, "metadata": {"original_paper_id": "10.1101/2020.04.06.20055632", "original_paper_title": "Association of County-Level Socioeconomic and Political Characteristics with Engagement in Social Distancing for COVID-19", "original_paper_code": "not stated", "original_paper_data": "not stated"}}}, "expected_post_registration_3": null}
{"study_id": "2", "original_paper_pdf": "2/input/original_paper.pdf", "initial_details": "[CLAIM]\nUse of memorization moderated the effect of school-average ability on mathematical self-concept (-.089,p<0.001; ), with an effect size of -0.157\n\n[HYPOTHESES]\nThe interaction of the use of memorization and school-average ability will be negative in its association with mathematical self-concept", "replication_data_files": ["2/input/replication_data/PISA2012.replication.RDS", "2/input/replication_data/!!CORRECTION!!_Seaton_AmEduResJourn_2010_Blxd_final.R"], "human_preregistration": "2/gt/human_preregistration.pdf", "human_report": "2/gt/human_report.pdf", "expected_post_registration": {"original_study": {"claim": {"hypothesis": "The interaction of the use of memorization and school-average ability will be negative in its association with mathematical self-concept.", "hypothesis_location": "p. 401, Hypotheses and Research Questions Addressed by the Present Investigation section, point 6.", "statement": "Use of memorization moderated the effect of school-average ability on mathematical self-concept (-.089,p<0.001; ), with an effect size of -0.157.", "statement_location": "p. 409, Study Methods: Cognitive and Metacognitive Learning Strategies section; also Table 3 and Table 4 (for effect size).", "study_type": "Observational"}, "data": {"source": "Program for International Student Assessment (PISA) database - Organisation for Economic Cooperation and Development (OECD). (2005b).", "wave_or_subset": "year 2003", "sample_size": "265180", "unit_of_analysis": "students", "access_details": "not stated", "notes": "Students are nested within schools, and schools are nested within countries. N = 265,180 students who attended 10,221 schools in 41 countries. To enable multilevel modeling schools with 10 or fewer students were excluded from further analysis. Only students who completed math self-concept items were included. other confounding variables not considered here (e.g., school expenditure levels, teacher characteristics, and other individual student  characteristics) are likely to have an impact upon academic performance and school climate, and these likely vary greatly across the many schools included in this sample. Also, no school policy or school practice variables were included in the present investigation. "}, "method": {"description": "The authors used three-level multilevel regression analyses to examine how school-average mathematics ability influenced students’ mathematics self-concept and whether this relationship was moderated by 16 socioeconomic and academic self-regulation constructs.", "steps": "1. The authors presumably obtained the data from the OECD technical report, cleaned it, and removed students who did not complete math self-concept items, and schools with 10 or fewer students. \n2. Then they standardised the values for mathematics ability, mathematics self-concept, and all the potential moderators (including memorisation) across the entire sample. \n3. A school-average mathematics ability variable was calculated by averaging each plausible value separately within each school. \n4. This school-average mathematics ability variable was not restandardized, thus keeping all variables in the same metric as the individual test scores.\n5. Cross-products with school-average ability were created for each potential moderator (again, including memorisation) but were not restandardized.\n6. Sample weights were used to prevent biased estimates of population parameters.\n7. The authors then ran three-level multilevel regression analyses (students within schools within countries) for each of the PISA plausible value of mathematics ability.\n8. Then the authors averaged the regression results to get final parameter estimates that represented the overall pattern across those regressions.\n9. Finally, the effect sizes comparable with Cohen’s d were calculated using Tymms’s (2004) formula.", "models": "multilevel regression", "outcome_variable": "Mathematics self-concept", "independent_variables": "Individual mathematics ability (linear and quadratic), school-average mathematics ability, use of memorization (moderator), and the interaction between school-average mathematics ability and use of memorization", "control_variables": "NA", "tools_software": "not stated"}, "results": {"summary": "Memorization use significantly moderated the relationship between school-average ability and mathematical self-concept, weakening this association (–0.089, p < 0.001; effect size = –0.157).", "numerical_results": [{"outcome_name": "Mathematics self-concept", "value": "-0.089", "unit": "standard deviation (the DV was standardised)", "effect_size": "-0.157", "confidence_interval": {"lower": "not stated", "upper": "not stated", "level": "not stated"}, "p_value": "< .001", "statistical_significance": "true", "direction": "negative"}]}, "metadata": {"original_paper_id": "10.3102/0002831209350493", "original_paper_title": "Big-Fish-Little-Pond Effect: Generalizability and Moderation—Two Sides of the Same Coin.", "original_paper_code": "not stated", "original_paper_data": "not stated"}}}, "expected_post_registration_2": {"original_study": {"claim": {"hypothesis": "The interaction of the use of memorization and school-average ability will be negatively related to mathematics self-concept", "hypothesis_location": "Page 12 (Hypotheses and Research Questions Addressed\nby the Present Investigation) Hypothesis 6", "statement": "The study finds that the Memorization × School-Average Mathematics Ability interaction is statistically significant and negative (b = -0.089), indicating that students who use memorization more experience a stronger negative effect of high school-average ability on mathematical self-concept.", "statement_location": "Table 3 (page 21), row 'School-Average Ability × Moderator' under Memorization (value = -0.089), and text on p. 20 (Study Methods: Cognitive and Metacognitive Learning Strategies) stating it had a 'statistically significant negative association.'", "study_type": "Observational"}, "data": {"source": "Programme for International Student Assessment (PISA) 2003 dataset.", "wave_or_subset": "PISA 2003 mathematics assessment sample.", "sample_size": "265,180 students who attended 10,221 schools in 41 countries.", "unit_of_analysis": "Individual student (nested within schools and countries in a multilevel model).", "access_details": "not stated", "notes": "Analyses include students with non-missing data for mathematical self-concept, memorization strategy use, individual mathematics ability, and school-average mathematics ability."}, "method": {"description": "The authors used multilevel modeling to test whether memorization strategy use moderates the big-fish–little-pond effect (the negative association between school-average mathematics ability and mathematical self-concept).", "steps": ["Identify students with complete data on mathematical self-concept, memorization strategy use, individual ability, and school-average mathematics ability.", "Compute school-average mathematics ability for each school based on aggregated achievement scores.", "Specify a multilevel model with students nested within schools and countries.", "Include memorization as a student-level predictor and school-average ability as a school-level predictor.", "Add the Memorization × School-Average Ability interaction term to test moderation.", "Estimate the model and interpret the interaction coefficient from Table 3."], "models": "Multilevel linear regression (three-level hierarchical model with students nested within schools nested within countries).", "outcome_variable": "Mathematical self-concept.", "independent_variables": "Memorization strategy use, School-average mathematics ability, Memorization × School-Average Ability interaction.", "control_variables": "Individual mathematics ability; intrinsic motivation; instrumental motivation; self-efficacy; elaboration; control strategies; anxiety; competitiveness; cooperativeness; sense of belonging; teacher–student relations; gender; family SES; school-average SES; and country-level effects (three-level model structure).", "tools_software": "not stated"}, "results": {"summary": "The Memorization × School-Average Ability interaction is statistically significant and negative, indicating that memorization strengthens the negative association between attending a high-ability school and mathematical self-concept.", "numerical_results": [{"outcome_name": "Mathematical self-concept (interaction effect of Memorization × School-Average Mathematics Ability)", "value": -0.089, "unit": "unstandardized regression coefficient", "effect_size": "-0.157", "confidence_interval": {"lower": "not stated", "upper": "not stated", "level": "not stated"}, "p_value": "p < .001", "statistical_significance": 1, "direction": "negative"}]}, "metadata": {"original_paper_id": "10.3102/0002831209350493", "original_paper_title": "Big-Fish–Little-Pond Effect: Generalizability and Moderation—Two Sides of the Same Coin", "original_paper_code": "not stated", "original_paper_data": "not stated"}}}, "expected_post_registration_3": null}
{"study_id": "3", "original_paper_pdf": "3/input/original_paper.pdf", "initial_details": "[CLAIM]\nThe positive sign of democracy index indicates that more democratic countries are affected more by the disease. coefficient = 86.76467, p = 0.0001.\n\n[HYPOTHESES]\nAt the country level, the democracy index will be positively associated with the total number of confirmed infections per one million people.", "replication_data_files": ["3/input/replication_data/Hossain 2020 - Replication Analysis.do", "3/input/replication_data/COVID replication.rds", "3/input/replication_data/COVID replication.dta"], "human_preregistration": "3/gt/human_preregistration.pdf", "human_report": "3/gt/human_report.pdf", "expected_post_registration": {"original_study": {"claim": {"hypothesis": "At  thecountry level,the democracy index will be positively associated with the total number of confirmed infections per one million people.", "hypothesis_location": "p. 8, Determinants of Infectious Disease Spread section; it also mentioned in the abstract.", "statement": "The positive sign of democracy index indicates that more democratic countries are affected more by the disease. coefficient = 86.76467, p = 0.0001.", "statement_location": "Table1, Model1.", "study_type": "Observational"}, "data": {"source": "European Centre for Disease Prevention and Control (ECDC) for data on coronavirus and population; meteoblue.com for the average yearly temperature; indexmundi.com for the average precipitation; Databank of the World Bank for the openness data; Wikipedia for the scores of democracy index.", "wave_or_subset": "coronavirus total infection cases: from 31 December 2019 to 03 April 2020; population data: year 2018; Wikipedia data: year 2019.", "sample_size": "163", "unit_of_analysis": "countries", "access_details": "not stated; all sources seem to be open access.", "notes": "Some countries were excluded (it is not mentioned which ones) due to unavailability of all data. Total cases of infection were converted to cases per one million population to capture the population effect. The economic and social variables, viz. openness, democracy index and population density were measured on yearly basis. The explanatory variables are static in nature.The estimated models suffer from heteroscedasticity and violate the normality assumption. Also, the model of interest (Model 1) is not free from autocorrelation."}, "method": {"description": "The authors check linkages between the severity of COVID-19 infections and environmental, economic, and social factors across countries.", "steps": "1. The process is not described explicitly but it can be infered that: the authors collected the data from relevant sources. \n2. After cleaning they converted total infection cases to cases per one million population. \n3. Then the data was merged and the model estimated with least squares method.", "models": "least squares method", "outcome_variable": "Y = cases of infection per one million people on 03 April 2020 by countries", "independent_variables": "yearly average temperature, yearly average precipitation, openness (ratio of international trade to GDP), democracy index (proxy for social cohesion) , population density", "control_variables": "NA", "tools_software": "Excel and STATA for manipulation and transformation of data; EVIEWS for estimation."}, "results": {"summary": "The results show that higher democracy scores are associated with greater COVID-19 impact (coefficient = 86.76, p = 0.0001).", "numerical_results": [{"outcome_name": "Y", "value": "86.76467", "unit": "infection cases", "effect_size": "not stated", "confidence_interval": {"lower": "not stated", "upper": "not stated", "level": "not stated"}, "p_value": "0.0001", "statistical_significance": "true", "direction": "positive"}]}, "metadata": {"original_paper_id": "https://doi.org/10.1101/2020.04.08.20058164", "original_paper_title": "Is the spread of COVID-19 across countries influenced by environmental, economic and social factors?", "original_paper_code": "not stated", "original_paper_data": "not stated"}}}, "expected_post_registration_2": {"original_study": {"claim": {"hypothesis": "Higher yearly average temperature is associated with fewer confirmed cases of COVID-19 infection per million population across countries.", "hypothesis_location": "Data sources and characteristics (p. 9) / Modeling Approach, p. 11)", "statement": "The study finds that yearly average temperature has a statistically significant negative association with cases of COVID-19 infection per million people, such that countries with higher average temperatures tend to have fewer confirmed cases per million.", "statement_location": "Abstract (p. 1) and Estimation Results and Discussion / Table 1 (p. 14, Temperature row).", "study_type": "Observational"}, "data": {"source": "Country-level data on confirmed COVID-19 cases and population from the European Centre for Disease Prevention and Control (ECDC); yearly average temperature from meteoblue.com; average precipitation from indexmundi.com; economic openness (trade as a percentage of GDP) from the World Bank Databank; and democracy index scores from the Economist Intelligence Unit’s Democracy Index 2019, as compiled on Wikipedia.", "wave_or_subset": "Cross-sectional snapshot of total confirmed COVID-19 cases by country on 03 April 2020, converted to cases per one million population, for 163 countries.", "sample_size": "163 countries.", "unit_of_analysis": "Country.", "access_details": "not stated (the paper lists data sources but does not describe specific access procedures or restrictions for these datasets).", "notes": "Total confirmed cases of COVID-19 by country are taken from 31 December 2019 to 03 April 2020, but only cases per one million people on 03 April 2020 are used as the dependent variable. Environmental, economic, and social explanatory variables (temperature, precipitation, openness, democracy index, population density) are measured on a yearly basis. Some countries are excluded due to missing data on one or more of these variables."}, "method": {"description": "The study uses cross-sectional ordinary least squares regressions to examine whether country-level environmental, economic, and social characteristics—including yearly average temperature—are associated with confirmed COVID-19 cases per million population across countries.", "steps": ["Collect total confirmed COVID-19 cases and 2018 population by country from the ECDC and compute cases per one million population for 03 April 2020 as the dependent variable Y.", "Merge country-level yearly average temperature, average precipitation, economic openness (international trade as a percentage of GDP), democracy index scores, and population density into the dataset for the set of countries with complete data.", "Specify an initial linear regression model regressing cases of infection per million (Y) on temperature, precipitation, openness, democracy index, and population density.", "Estimate this model with least squares and identify that precipitation and population density are not statistically significant predictors.", "Re-estimate the model excluding precipitation and population density, retaining yearly average temperature, openness, and democracy index as predictors (Model 1, no lag).", "Conduct further models including a transformed lagged dependent variable via residuals to account for past infection levels (Models 2 and 3), while keeping temperature, openness, and democracy as predictors.", "Use Excel and STATA for data manipulation and EVIEWS software to estimate the regression models."], "models": "Ordinary least squares linear regression of cases of infection per million population on yearly average temperature, economic openness, and democracy index (with additional specifications including a residual-based proxy for lagged infections).", "outcome_variable": "Total confirmed cases of COVID-19 infection per one million people in each country on 03 April 2020.", "independent_variables": "Yearly average temperature of countries (X1, measured as country-level average temperature).", "control_variables": "Economic openness (international trade as a percentage of GDP); democracy index score; and, in extended lag models, the residual term v_i derived from regressing lagged cases on the explanatory variables.", "tools_software": "Excel and STATA for data manipulation and transformation; EVIEWS for estimation of the regression models."}, "results": {"summary": "Across the reported regression models, yearly average temperature is negatively and significantly associated with COVID-19 cases per million.", "numerical_results": [{"outcome_name": "Cases of COVID-19 infection per one million people (effect of yearly average temperature)", "value": -16.55022, "unit": "unstandardized OLS regression coefficient (change in cases per million per one-unit increase in yearly average temperature)", "effect_size": "unstandardized OLS regression coefficient", "confidence_interval": {"lower": "not stated", "upper": "not stated", "level": "not stated"}, "p_value": "0.0064 (Model 1, no lag)", "statistical_significance": 1, "direction": "negative"}]}, "metadata": {"original_paper_id": "10.1101/2020.04.08.20058164", "original_paper_title": "Is the spread of COVID-19 across countries influenced by environmental, economic and social factors?", "original_paper_code": "not stated", "original_paper_data": "not stated"}}}, "expected_post_registration_3": null}
{"study_id": "4", "original_paper_pdf": "4/input/original_paper.pdf", "initial_details": "", "replication_data_files": ["4/input/replication_data/Gerhold_covid_Azg9_0948_final.R", "4/input/replication_data/data_gerhold.csv"], "human_preregistration": "4/gt/human_preregistration.pdf", "human_report": "4/gt/human_report.pdf", "expected_post_registration": {"original_study": {"claim": {"hypothesis": "Women will express more concern about COVID-19 than men.", "hypothesis_location": "the abstract, also mentioned in the Summary section, p. 10.", "statement": "In detail, 62.1% agree (answering either with“strongly agree” or“agree” onthe5-pointLikert Scale) that they are worriedabout COVID-19 in general (women=68.2%, men=55.7%, p<.01).", "statement_location": "p. 5, Fear section.", "study_type": "Observational"}, "data": {"source": "only an ", "online access ISO-certified panel provider is mentioned": "", "wave_or_subset": "There is no specific subset but the data were collected from the 19th to the 23rd of March 2020.", "sample_size": "1242", "unit_of_analysis": "adult respondents", "access_details": "no explicit mention in the text but the panel provider was responsible for data collection, so the data was probably sent to the authors", "notes": "not stated"}, "method": {"description": "The authors analysed a representative online survey of German adults to examine gender, age, and regional differences in concern, fear, and coping strategies related to COVID-19.", "steps": "1. The data was collected by an external panel provider, then cleaned. \n2. All respondents who took less than seven minutes to complete the questionnaire were removed. \n3. Then descriptives were calculated and reported.", "models": "the text doesn’t explicitly mention the model, but probably it was chi-squared", "outcome_variable": "worry about COVID-19", "independent_variables": "Gender", "control_variables": "NA", "tools_software": "not stated"}, "results": {"summary": "About 62% of the respondents reported being worried about COVID-19, with women expressing more concern (68.2%) than men (55.7%) (p < .01).", "numerical_results": [{"outcome_name": "Worried about COVID-19", "value": "women=68.2%, men=55.7%", "unit": "%", "effect_size": "not stated", "confidence_interval": {"lower": "not stated", "upper": "not stated", "level": "not stated"}, "p_value": "<.01", "statistical_significance": "true", "direction": "NA"}]}, "metadata": {"original_paper_id": "not stated", "original_paper_title": "COVID-19: Risk perception and Coping strategies. Results from a survey in Germany.", "original_paper_code": "not stated", "original_paper_data": "not stated"}}}, "expected_post_registration_2": {"original_study": {"claim": {"hypothesis": "Women are more concerned about COVID-19 than men.", "hypothesis_location": "Gender differences in concern are reported in the Results section under 'Fear', page 5, where women show higher agreement that they are worried about COVID-19.", "statement": "The study finds that women report significantly higher concern about COVID-19 than men, with 68.2% of women agreeing that they are worried about COVID-19 compared to 55.7% of men (p < .01).", "statement_location": "Results section, page 5.", "study_type": "Observational"}, "data": {"source": "Online access panel survey of adults in Germany conducted by an ISO-certified panel provider.", "wave_or_subset": "Data collected from 19–23 March 2020.", "sample_size": "1242 respondents after data cleansing.", "unit_of_analysis": "Individual respondent.", "access_details": "not stated", "notes": "Respondents were quota-sampled to be online-representative by gender, age, and federal state. All survey questions were mandatory. Respondents who completed the survey in under seven minutes were removed."}, "method": {"description": "The study measured concern about COVID-19 using Likert-scale items and compared responses between women and men to assess gender differences in expressed concern.", "steps": ["Administer a mandatory-question online survey to adult respondents in Germany.", "Measure concern using the item: 'The COVID-19 worries me' on a 5-point Likert scale.", "Calculate the percentage of respondents agreeing ('strongly agree' or 'agree') by gender.", "Compare percentages between women and men and report statistical significance."], "models": "not stated (results reported as descriptive percentages with significance testing).", "outcome_variable": "Agreement with the statement 'The COVID-19 worries me' (5-point Likert scale, summarized as percentage agreeing).", "independent_variables": "Gender (women vs. men).", "control_variables": "not stated (no regression or adjusted model is reported).", "tools_software": "not stated"}, "results": {"summary": "Women express significantly more concern about COVID-19 than men, as measured by agreement with the statement 'The COVID-19 worries me.'", "numerical_results": [{"outcome_name": "Percentage agreeing they are worried about COVID-19", "value": 68.2, "unit": "percentage of women agreeing", "effect_size": "difference in proportions (68.2% vs 55.7%)", "confidence_interval": {"lower": "not stated", "upper": "not stated", "level": "not stated"}, "p_value": "< .01", "statistical_significance": 1, "direction": "positive"}]}, "metadata": {"original_paper_id": "not stated", "original_paper_title": "COVID-19: Risk perception and coping strategies. Results from a survey in Germany.", "original_paper_code": "not stated", "original_paper_data": "not stated"}}}, "expected_post_registration_3": null}
{"study_id": "5", "original_paper_pdf": "5/input/original_paper.pdf", "initial_details": "[CLAIM]\nIn  terms  of  disease  detection,  more  educated  respondents  have  a  higher probability of being diagnosed, but only conditional on being in poor general health (marginal effect for Years of Education=-0.00867, SE=0.00420, significantat 5% level).\n\n[HYPOTHESIS]\nAmong the sample of respondents in poor general health who were found to be hypertensive during a screening, the probability of being undiagnosed decreases with education", "replication_data_files": ["5/input/replication_data/Kim & Radoias 2016 - Replication Analysis.do", "5/input/replication_data/replication_data.dta"], "human_preregistration": "5/gt/human_preregistration.pdf", "human_report": "5/gt/human_report.pdf", "expected_post_registration": {"original_study": {"claim": {"hypothesis": "Among the sample of respondents in poor general health who were found to be hypertensive during a screening, the probability of being undiagnosed decreases with education.", "hypothesis_location": "the abstract, it is also discussed in sections: 1. Introduction, 2. Theoretical framework.", "statement": "In  terms  of  disease  detection,  more  educated  respondents  have  a  higher probability of being diagnosed, but only conditional on being in poor general health (marginal effect for Years of Education=-0.00867, SE=0.00420, significant at 5% level).", "statement_location": "Table 2, row: Years of Education, column: Respondents in poor health.", "study_type": "Observational"}, "data": {"source": "Indonesian Family Life Survey (IFLS) fielded by the RAND corporation in collaboration with the Center for Population and Policy Studies of the University of Gadjah Mada and Survey METER.", "wave_or_subset": "fourth wave, the data were collected between November 2007 and April 2008.", "sample_size": "4209", "unit_of_analysis": "hypertensive adults (over 45 years of age in 2007)", "access_details": "There are no details provided except for the information that the IFLS is a publicly available data set.", "notes": "Blood pressure was measured three times by trained nurses. The first reading was dropped and average of last two used to construct the hypertension variable. Under-diagnosed individuals are those hypertensives at survey but never diagnosed by a doctor. Respondents’ health variablw was constructed using self-reported health status. It was on a 1–4 scale and later dichotomized into ‘good’ vs. ‘poor’ groups. Per capita expenditures (PCE) was used as proxy for household income."}, "method": {"description": "The authors analyzed data from the fourth wave of the IFLS, focusing on adults over 45 who were found hypertensive during clinical screenings. They examined how education, time preferences, and other socio-economic factors influence the likelihood of being hypertensive but undiagnosed.", "steps": "1. The authors got access to the publicly available Indonesian Family Life Survey dataset.\n2. Then they restricted the sample to hypertensive adults aged 45 and older. \n3. Then, individuals who reported never having been diagnosed with hypertension by a doctor were coded these as under-diagnosed. \n4. The sample was split into those in good health (very healthy or somewhat healthy) and those in poor health (unhealthy or somewhat unhealthy).\n5. Adding explanatory variables including years of education, time and risk preference parameters, age, age squared, per capita household expenditures, distance to nearest health center, and sex, the authors estimated separate probit models for the two subsamples (good vs. poor general health), using the binary under-diagnosis variable as the dependent variable. \n6. Finally, marginal effects for all explanatory variables were computed and reported.", "models": "probit regression", "outcome_variable": "being hypertensive but not previously diagnosed (binary yes = 1, no = 0)", "independent_variables": "education (measured in years of formal education), age, age squared (to allow for possible non-linear effects), individual risk and time preferences, the distance from the closest health center (to proxy for the ease of access to medical care), household per capita expenditures (PCE), and a sex dummy", "control_variables": "not stated", "tools_software": "not stated"}, "results": {"summary": "More educated respondents were more likely to be diagnosed with the disease, but only among those in poor general health (marginal effect for Years of Education=-0.00867, SE=0.00420, significantat 5% level).", "numerical_results": [{"outcome_name": "Hypertension under-diagnosis", "value": "-0.00867", "unit": "NA", "effect_size": "not stated", "confidence_interval": {"lower": "not stated", "upper": "not stated", "level": "not stated"}, "p_value": "0.05", "statistical_significance": "true", "direction": "negative"}]}, "metadata": {"original_paper_id": "http://dx.doi.org/10.1016/j.socscimed.2015.11.051", "original_paper_title": "Education, individual time preferences, and asymptomatic disease\ndetection.", "original_paper_code": "not stated", "original_paper_data": "not stated"}}}, "expected_post_registration_2": {"original_study": {"claim": {"hypothesis": "For hypertensive adults in good general health, education will not significantly reduce the probability of being under-diagnosed with hypertension.", "hypothesis_location": "Section 2. Theoretical framework (page 3), where the authors state that they expect 'significantly smaller (possibly zero or even negative) effects of education for generally healthy people than for unhealthy people'.", "statement": "The study finds that among hypertensive adults who report good general health, years of education have a small and statistically nonsignificant association with the probability of being under-diagnosed, indicating that education does not meaningfully reduce under-diagnosis in this group.", "statement_location": "Section 4. Disease detection results (page 5), where the authors note that for individuals in good health 'the education level does not matter at all,' and Table 2 (page 5), column 'Respondents in good health', row 'Years of Education' (marginal effect = 0.00295, standard error = 0.00206, no significance stars).", "study_type": "Observational"}, "data": {"source": "Indonesian Family Life Survey (IFLS).", "wave_or_subset": "IFLS Wave 4 (2007-2008), restricted to adults over age 45 who were hypertensive based on survey blood pressure measurements.", "sample_size": "4,209 hypertensive adults (1,793 men and 2,416 women); disease-detection probit subsamples of 3,145 respondents in good general health and 1,064 respondents in poor general health.", "unit_of_analysis": "Individual respondent.", "access_details": "The IFLS is described as a publicly available dataset that has received IRB approval at RAND and in Indonesia.", "notes": "Hypertension status is based on nurse-measured blood pressure (average of the second and third readings), using WHO cutoffs (systolic ≥140 or diastolic ≥90). Under-diagnosed respondents are those found hypertensive in the IFLS measurements who reported never having been diagnosed with hypertension by a doctor. General health status (GHS) is self-rated on a 1-4 scale and used to split the sample into 'good' (very healthy or somewhat healthy) and 'poor' (unhealthy or somewhat unhealthy) general health groups."}, "method": {"description": "The authors estimate probit models of hypertension under-diagnosis using IFLS data, examining how education and individual time preferences relate to the probability of being under-diagnosed, separately for respondents in good versus poor general health.", "steps": ["Identify adults (over age 45) in IFLS Wave 4 who are hypertensive based on nurse-measured blood pressure following WHO cutoffs.", "Classify respondents as 'under-diagnosed' if they are hypertensive in the survey measurements but report never having been diagnosed with hypertension by a doctor.", "Obtain self-rated general health status (GHS) and split the hypertensive sample into two groups: respondents in good general health (very healthy or somewhat healthy) and respondents in poor general health (unhealthy or somewhat unhealthy).", "Construct explanatory variables including years of education, household per capita expenditures, time preference, risk preference, distance to the nearest health center, age, age squared, and sex.", "Estimate probit models of the probability of being under-diagnosed for the full sample and separately for the good-health and poor-health subsamples, and report marginal effects.", "Interpret the marginal effect of years of education on under-diagnosis within the good-health subsample to assess whether education significantly affects disease detection for generally healthy hypertensive adults."], "models": "Probit regression models of the probability of being under-diagnosed with hypertension, reporting marginal effects for the full sample and for subsamples defined by general health status.", "outcome_variable": "Indicator variable equal to 1 if a respondent is hypertensive based on IFLS blood pressure measurements but has not previously been diagnosed with hypertension by a doctor (i.e., under-diagnosed), conditional on being in good general health.", "independent_variables": "Years of formal education completed by the respondent.", "control_variables": "Household per capita expenditures (log PCE); time preference category; risk preference category; distance to the closest health center; age; age squared; and a female dummy indicator.", "tools_software": "not stated"}, "results": {"summary": "In the subsample of hypertensive adults who report good general health, years of education have statistically nonsignificant marginal effect on the probability of being under-diagnosed.", "numerical_results": [{"outcome_name": "Probability of being under-diagnosed with hypertension (good general health subsample, effect of years of education)", "value": 0.00295, "unit": "marginal effect on probability of being under-diagnosed per additional year of education", "effect_size": "probit marginal effect = 0.00295", "confidence_interval": {"lower": "not stated", "upper": "not stated", "level": "not stated"}, "p_value": "not stated (no significance stars are reported for this coefficient in Table 2, indicating nonsignificance at conventional levels).", "statistical_significance": 0, "direction": "0"}]}, "metadata": {"original_paper_id": "10.1016/j.socscimed.2015.11.051", "original_paper_title": "Education, individual time preferences, and asymptomatic disease detection", "original_paper_code": "not stated", "original_paper_data": "not stated"}}}, "expected_post_registration_3": null}
{"study_id": "6", "original_paper_pdf": "6/input/original_paper.pdf", "initial_details": "[CLAIM]\nOverall, most individuals favor either scientific or religious ways of understanding, but many scientifically inclined individuals prefer certain religious accounts. Members of the post-secular category were significantly less likely than members of the traditional group to respond that humans evolved from other animals (3 percent, significant at p < 0.05 on a two-tailed test).\n\n[HYPOTHESIS]\nRespondents with a post-secular perspective on science and religion will be less likely than respondents with a traditional perspective on science and religion to respond that humans evolved from other animals.", "replication_data_files": ["6/input/replication_data/GSSreplication.dta", "6/input/replication_data/OBrienReplication_OSF_Axxe_20201012.do"], "human_preregistration": "6/gt/human_preregistration.pdf", "human_report": "6/gt/human_report.pdf", "expected_post_registration": {"original_study": {"claim": {"hypothesis": "Respondents with a post-secular perspective on science and religion will be less likely than respondents with a traditional perspective on science and religion to respond that humans evolved from other animals.", "hypothesis_location": "it is discussed in Theoretical Perspectives: Conflict between Science and Religion.", "statement": "Overall, most individuals favor either scientific or religious ways of understanding, but many scientifically inclined individuals prefer certain religious accounts. Members of the post-secular category were significantly less likely than members of the traditional group to respond that humans evolved from other animals (3 percent, significant at p < 0.05 on a two-tailed test).", "statement_location": "Table 2, Conditional Means by Latent Class, Post-Secular; also discussed in Results: Perspectives on Science and Religion.", "study_type": "Observational"}, "data": {"source": "General Social Survey", "wave_or_subset": "2006, 2008, 2010", "sample_size": "622", "unit_of_analysis": "individual", "access_details": "the survey is referenced in the article but no other access details are provided: Smith, Tom W., Peter Marsden, Michael Hout, and Jibum Kim. 2011. General Social Surveys, 1972–2010 [machine-readable data file]. Principal Investigator, Tom W. Smith; Co-Principal Investigator, Peter V. Marsden; Co-Principal Investigator,  Michael Hout; Sponsored by National Science Foundation, NORC. Chicago: National Opinion Research Center [producer]; Storrs, CT: The Roper Center for Public Opinion Research, University of Connecticut [distributor].", "notes": "Accounting for the survey’s split-ballot design and missing cases, the full sample comprises 2,901 respondents (1,563 from 2006; 988 from 2008; and 350 from 2010). The Post-Secular latent class, used for the evolution item result, includes 622 respondents."}, "method": {"description": "Post-Secular respondents were identified via latent class analysis of GSS science and religion items; their responses to human evolution were compared to Traditional respondents using two-tailed tests.", "steps": "1.Combine GSS waves to get the whole dataset.\n2.Select survey items measuring science knowledge, attitudes, and religious beliefs.\n3.Conduct latent class analysis to identify groups within respondents.\n4.Assign respondents to latent classes based on based on their greatest posterior probability of class membership.\n5.Define the outcome variable as the binary response to the question: Humans evolved from earlier species of animals (1 = correct, 0 = incorrect).\n6.Compare the Post-Secular latent class to the Traditional class on the evolution item using two-tailed t-tests.", "models": "two-tailed t-test", "outcome_variable": "Binary response to Humans evolved from earlier species of animals", "independent_variables": "post-secular (latent class assigned via LCA)", "control_variables": "NA (it was a t-test)", "tools_software": "LCA: Mplus; descriptive and regression analyses: Stata"}, "results": {"summary": "Members of the post-secular category were significantly less likely than members of the traditional group to respond that humans evolved from other animals (3 percent, significant at p < 0.05 on a two-tailed test).", "numerical_results": [{"outcome_name": "Human beings developed from earlier species of animals", "value": "0.032", "unit": "proportion (in the table, 3% in the text)", "effect_size": "not stated", "confidence_interval": {"lower": "not stated", "upper": "not stated", "level": "not stated"}, "p_value": "< .05", "statistical_significance": "true", "direction": "negative"}]}, "metadata": {"original_paper_id": "10.1177/0003122414558919", "original_paper_title": "Traditional, Modern, and Post-Secular Perspectives on Science and Religion in the United States.", "original_paper_code": "not stated", "original_paper_data": "not stated"}}}, "expected_post_registration_2": {"original_study": {"claim": {"hypothesis": "Post-secular individuals are less likely than modern individuals to believe humans evolved from earlier species.", "hypothesis_location": "The descriptive comparison appears in the Results section (Table 2) where evolution beliefs are summarized for each latent perspective group.", "statement": "The study reports that only 3% of post-secular individuals agree that humans evolved from earlier species, indicating substantially lower acceptance of human evolution among post-secular respondents.", "statement_location": "Results section, Table 2.", "study_type": "Observational"}, "data": {"source": "General Social Survey (GSS).", "wave_or_subset": "2006, 2008, and 2010 waves", "sample_size": "2,901 cases (1,563 from 2006; 988 from 2008; and 350 from 2010).", "unit_of_analysis": "Individual adult survey respondent.", "access_details": "not stated", "notes": "Belief in human evolution is measured by agreement with the statement that humans evolved from earlier species. Respondents are classified into three latent perspectives—traditional, modern, and post-secular—based on latent class analysis using religion and science belief indicators."}, "method": {"description": "The authors used latent class analysis to identify three worldview groups—traditional, modern, and post-secular—and then descriptively compared the percentage in each group agreeing that humans evolved from earlier species.", "steps": ["Identify indicators of scientific and religious beliefs in the GSS dataset.", "Estimate a latent class model to classify respondents into traditional, modern, and post-secular perspectives.", "Extract predicted class membership for respondents.", "Calculate descriptive percentages of agreement with the statement that humans evolved from earlier species for each latent class.", "Compare evolution-belief percentages between post-secular and modern perspectives."], "models": "Latent class analysis used for group classification; the evolution comparison itself is descriptive and not based on regression modeling.", "outcome_variable": "Agreement that humans evolved from earlier species (percentage agreeing).", "independent_variables": "Latent perspective membership (post-secular vs. modern).", "control_variables": "not stated (no regression model is used for this descriptive comparison).", "tools_software": "not stated"}, "results": {"summary": "Post-secular individuals show substantially lower acceptance of human evolution than modern individuals, with only 3% agreeing that humans evolved from earlier species.", "numerical_results": [{"outcome_name": "Agreement that humans evolved from earlier species", "value": 3, "unit": "percentage of respondents agreeing.", "effect_size": "difference in proportions", "confidence_interval": {"lower": "not stated", "upper": "not stated", "level": "not stated"}, "p_value": "not stated.", "statistical_significance": 1, "direction": "negative"}]}, "metadata": {"original_paper_id": "10.1177/0003122414558919", "original_paper_title": "Traditional, Modern, and Post-Secular Perspectives on Science and Religion in the United States.", "original_paper_code": "not stated", "original_paper_data": "not stated"}}}, "expected_post_registration_3": null}
{"study_id": "7", "original_paper_pdf": "7/input/original_paper.pdf", "initial_details": "[CLAIM]\nState-level carbon emissions and average working hours have a strong, positive relationship, which holds across a variety of model estimation techniques and net of various political, economic, and demographic drivers of emissions. Specifically, they find that, over time, a 1 percent increase in average working hours per worker is associated with a 0.668 percent increase in emissions, holding all else constant\n\n[HYPOTHESIS]\nAverage working hours per worker in a state will be positively associated with carbon emissions.", "replication_data_files": ["7/input/replication_data/compiled.csv", "7/input/replication_data/epa.dta", "7/input/replication_data/Fitzgerald 2018 Script_clean v2.R", "7/input/replication_data/hhsize.dta", "7/input/replication_data/compiled.dta"], "human_preregistration": "7/gt/human_preregistration.pdf", "human_report": "7/gt/human_report.pdf", "expected_post_registration": {"original_study": {"claim": {"hypothesis": "Average working hours per worker in a state will be positively associated with carbon emissions.", "hypothesis_location": "it is discussed in the abtsract and the introduction", "statement": "State-level carbon emissions and average working hours have a strong, positive relationship, which holds across a variety of model estimation techniques and net of various political, economic, and demographic drivers of emissions. Specifically, they find that, over time, a 1 percent increase in average working hours per worker is associated with a 0.668 percent increase in emissions, holding all else constant.", "statement_location": "Results section, p. 1862; also Table 4 model 1.", "study_type": "Observational"}, "data": {"source": "carbon emissions: US Environmental Protection Agency (2016); \nworking hours, labor productivity, and employed population: US Bureau of Labor Statistics (BLS) Current Employment Statistics (CES) database (2016); \nGDP, and manufacturing as a percentage: US Department of Commerce Bureau of Economic Analysis (2016); \ntotal population size: US Census Bureau 2016; \nstate’s energy production: EIA’s State Energy Database System (2016);\nworking-age population, and average household size: US Census Bureau’s American Community Survey (2017);", "wave_or_subset": "2007-2013", "sample_size": "50 states (350 total state-year observations)", "unit_of_analysis": "US state", "access_details": "carbon emissions: Currently, the EPA website has been updated and no longer includes access to these data. However, it is possible to access the files through the January 19, 2017, snapshot version of the webpage (https://19january2017snapshot.epa.gov/statelocalclimate/state-energy-co2-emissions_.html). The lead author of this study will share these data upon request;\nworking hours, labor productivity, and employed population: there's a link in references (https://www.bls.gov/ces/) but no other access details are mentioned; \nGDP, and manufacturing as a percentage: there's a link in references (https://www.bea.gov/itable/index.cfm) but no other access details are mentioned;\ntotal population size: there's a link in references (https://www.census.gov/en.html) but no other access details are mentioned; \nstate’s energy production: there's a link in references (https://www.eia.gov/state/seds/) but no other access details are mentioned; \nworking-age population, and average household size: there's a link in references (https://www.census.gov/programs-surveys/acs/data.html) but no other access details are mentioned;", "notes": "All non-binary variables are transformed into logarithmic form. For such variables, the regression models estimate elasticity coefficients where the coefficient for the independent variable is the estimated net percent change in the dependent variable associated with a 1 percent increase in the independent variable. The working hours data cover all nonfarm private employees, but exclude public employees. The effect of working time on environmental outcomes is considered as a scale effect. The scale effect is measured as working time’s contribution to GDP. The authors test for the scale effect by disaggregating GDP into three parts: working hours, labor productivity, and employment to population ratio. In Model 1 they include both labor productivity and employed population percentage as the other components of GDP. Labor productivity is measured as GDP per hour of work. Employed population percentage is measured as the employed population of a state divided by its total population. The composition effect is measured net of GDP. All continuous variables are logged (ln). All models are calculated with AR(1) correction. All models contain unreported year-specific intercepts and unreported unit-specific intercept."}, "method": {"description": "The study evaluates the relationship between average working hours per worker and state-level carbon dioxide emissions across U.S. states, using panel data to test a positive association over time.", "steps": "1. Collect the data.\n2. Transform all non-binary variables into logarithmic form. \n3. Specify a model to examine the scale effect of average weekly working hours per worker on state-level carbon dioxide emissions. \n4. Include control variables: GDP per hour, employed population percentage, total population, energy production, manufacturing as a percentage of GDP, average household size, and working-age population percentage (ages 15–64).\n5. Estimate two-way fixed effects models with both state-specific and year-specific intercepts\n6. Apply Prais–Winsten estimation with panel-corrected standard errors (PCSEs) to account for heteroskedasticity and contemporaneous correlation\n7. Correct for first-order autocorrelation (AR(1)) within panels.", "models": "fixed effects regression (Prais-Winsten model with panel corrected standard errors)", "outcome_variable": "carbon dioxide emissions", "independent_variables": "working hours, GDP per hour, employed population percentage", "control_variables": "total population, manufacturing, state's energy production, working-age, average household size", "tools_software": "Stata version 13"}, "results": {"summary": "Results show a strong positive association between average working hours and state-level carbon emissions; a 1% increase in working hours corresponds to a 0.668% increase in emissions, holding other factors constant (SE = 0.179).", "numerical_results": [{"outcome_name": "CO₂ emissions", "value": "0.668", "unit": "% (increase in emissions per 1 percent increase in average working hours)", "effect_size": "not stated", "confidence_interval": {"lower": "not stated", "upper": "not stated", "level": "not stated"}, "p_value": "< 0.001", "statistical_significance": "true", "direction": "positive"}]}, "metadata": {"original_paper_id": "10.1093/sf/soy014", "original_paper_title": "Working Hours and Carbon Dioxide Emissions in the United States, 2007–2013.", "original_paper_code": "not stated", "original_paper_data": "not stated"}}}, "expected_post_registration_2": {"original_study": {"claim": {"hypothesis": "Higher average weekly working hours at the U.S. state level will be associated with higher carbon dioxide emissions from fossil fuel combustion.", "hypothesis_location": "Introduction and Literature review, which motivate examining the relationship between working hours and emissions (pages 1–3).", "statement": "The study finds that, over time, a 1 percent increase in average working hours per worker in a state is associated with a 0.668 percent increase in state-level CO2 emissions from fossil fuel combustion, net of other political, economic, and demographic drivers of emissions.", "statement_location": "Results section and Table 4 (page 16).", "study_type": "Observational."}, "data": {"source": "State-level carbon dioxide emissions from fossil fuel combustion from the U.S. Environmental Protection Agency (EPA), working hours and labor statistics from the U.S. Bureau of Labor Statistics Current Employment Statistics (CES), GDP and manufacturing share from the U.S. Bureau of Economic Analysis, energy production from the U.S. Energy Information Administration’s State Energy Data System, and demographic variables (population, household size, working-age share) from the U.S. Census Bureau and American Community Survey. State environmentalism scores come from Dietz et al. (2015).", "wave_or_subset": "Annual state-level data for all 50 U.S. states from 2007 to 2013, forming a balanced panel.", "sample_size": "350 state-year observations (50 states observed annually over 7 years, 2007–2013).", "unit_of_analysis": "State-year.", "access_details": "The authors note that the EPA state CO2 emissions files were obtained from an earlier snapshot of the EPA website and that the lead author will share these data upon request; other data sources (BLS, BEA, EIA, Census, ACS) are described as coming from publicly available databases.", "notes": "All non-binary variables are log-transformed. The dependent variable is total state-level CO2 emissions from fossil fuel combustion in million metric tons. Average weekly working hours are for nonfarm private employees only."}, "method": {"description": "The authors construct a balanced panel of the 50 U.S. states from 2007 to 2013 and estimate fixed-effects, random-effects, and hybrid panel regression models to assess whether average weekly working hours per worker are positively associated with state-level carbon dioxide emissions from fossil fuel combustion, net of economic, demographic, political, and energy-related controls.", "steps": ["Assemble annual state-level data for all 50 U.S. states from 2007 to 2013, including CO2 emissions from fossil fuel combustion, average weekly working hours per worker, GDP per capita, GDP per hour, employment-to-population ratio, population, manufacturing share of GDP, state energy production, average household size, working-age population percentage, state environmentalism scores, and census region.", "Log-transform all non-binary variables so that coefficients can be interpreted as elasticities (percent changes).", "Specify panel regression models with total CO2 emissions (logged) as the dependent variable and average weekly working hours (logged) as the main independent variable.", "Estimate two-way fixed-effects Prais–Winsten models with AR(1) corrections, including state-specific and year-specific intercepts, to assess the scale effect of working hours (models that also include GDP per hour and employed population percentage) and the composition effect (models that include GDP per capita instead of its components).", "Estimate random-effects panel models with year-specific intercepts, AR(1) correction, state environmentalism, and census-region dummy variables as additional controls to assess robustness.", "Conduct robustness checks using hybrid models that include both unit-specific means and deviations for time-varying covariates, and sensitivity analyses excluding one state at a time and excluding states experiencing recent fracking booms.", "Interpret the elasticity coefficient for logged average working hours as the percent change in CO2 emissions associated with a 1 percent change in working hours, net of controls."], "models": "Two-way fixed-effects panel regression models with Prais–Winsten estimation and AR(1) correction; random-effects panel regression models with AR(1) correction; and hybrid models that decompose time-varying covariates into unit-specific means and deviations.", "outcome_variable": "Log of total state-level carbon dioxide emissions from fossil fuel combustion (million metric tons CO2).", "independent_variables": "Log of average weekly working hours per worker in each state.", "control_variables": "Log GDP per hour; log employed population percentage; log GDP per capita (in composition-effect models); log total population; log state energy production; log manufacturing as a percentage of GDP; log average household size; log working-age population percentage (15–64); state environmentalism index; and census region dummy variables (Midwest, South, West, with Northeast as the reference category), along with state and year fixed effects in the fixed-effects models.", "tools_software": "Stata (Version 13)."}, "results": {"summary": "Across fixed-effects, random-effects, and hybrid panel models, average weekly working hours are positively and significantly associated with state-level CO2 emissions from fossil fuel combustion. In the preferred fixed-effects scale model, a 1 percent increase in average working hours per worker is associated with a 0.668 percent increase in emissions, net of productivity, employment, population, energy production, manufacturing share, household size, and working-age population. Similar positive and significant elasticities are found in other model specifications, supporting the conclusion that longer working hours contribute to higher emissions at the state level.", "numerical_results": [{"outcome_name": "Log total CO2 emissions from fossil fuel combustion (elasticity with respect to average weekly working hours, fixed-effects scale model)", "value": 0.668, "unit": "elasticity coefficient (percent change in CO2 emissions associated with a 1 percent increase in average weekly working hours)", "effect_size": "panel regression elasticity coefficient = 0.668 (model 1)", "confidence_interval": {"lower": "not stated", "upper": "not stated", "level": "not stated"}, "p_value": "< 0.001", "statistical_significance": 1, "direction": "positive"}]}, "metadata": {"original_paper_id": "10.1093/sf/soy014", "original_paper_title": "Working Hours and Carbon Dioxide Emissions in the United States, 2007–2013", "original_paper_code": "not stated", "original_paper_data": "not stated"}}}, "expected_post_registration_3": null}
{"study_id": "8", "original_paper_pdf": "8/input/original_paper.pdf", "initial_details": "[CLAIM]\na very high subsidy (such as the one under consideration by the international community) dramatically increases access, but nearly one-half of subsidized pills go to patients with-out malaria. The coefficient for “Any ACT subsidy” is 0.187 with robust standard errors clustered at the household level of 0.038, significant at the 1 percent level\n\n[HYPOTHESIS]\nACT [artemisinin combination therapies] subsidies induce take-up of ACT", "replication_data_files": ["8/input/replication_data/ReplicationData_Cohen_AmEcoRev_2015_2lb5.dta", "8/input/replication_data/Cohen et al 2015 - Replication Analysis.do"], "human_preregistration": "8/gt/human_preregistration.pdf", "human_report": "8/gt/human_report.pdf", "expected_post_registration": {"original_study": {"claim": {"hypothesis": "ACT [artemisinin combination therapies] subsidies induce take-up of ACT", "hypothesis_location": "discussed in the abstract and introduction", "statement": "a very high subsidy (such as the one under consideration by the international community) dramatically increases access, but nearly one-half of subsidized pills go to patients with-out malaria. The coefficient for “Any ACT subsidy” is 0.187 with robust standard errors clustered at the household level of 0.038, significant at the 1 percent level.", "statement_location": "Table 2, column 1, row 1.", "study_type": "Experimental"}, "data": {"source": "NA", "wave_or_subset": "NA", "sample_size": "631", "unit_of_analysis": "illness episode", "access_details": "The experimental data are presumably stored by IPAKenya field officers who collected the data or one of the authors. There is a note on the first page that there are additional materials at: http://dx.doi.org/10.1257/aer.20130267, but no mention if the repository contains the actual data or the code.", "notes": "The study was conceived and implemented in 2008–2009. The study to included three subsidy levels 80, 88, and 92 percent. The experiment was conducted in the districts of Busia, Mumias, and Samia in Western Kenya between May and December of 2009. If more than one household member got sick simultaneously, the authors included all concurrent first episodes, and therefore clustered the standard errors in all illness episode regressions at the household level."}, "method": {"description": "The study evaluates the impact of randomized subsidies for artemisinin combination therapies (ACTs) on ttake-up of ACTs, using experimental data and regression analysis to estimate effects on ACT access and targeting.", "steps": "1. The authors selected four rural market centers in Western Kenya and partnered with one drug shop in each; they sampled all households within a 4-kilometer catchment radius, excluding areas near health facilities. \n2. They administered a baseline household survey and distributed two vouchers for artemisinin-based combination therapies (ACTs). \n3. They randomly assigned households using a computerized algorithm (stratified by drug shop, distance quartile, and presence of children) to one of two groups: (a) No Subsidy: Unsubsidized ACTs at the market price. (b) ACT Subsidy: ACTs discounted at 80%, 88%, or 92% subsidy levels. \n4. Then, they recorded voucher redemptions and treatment details for illness episodes at participating drug shops over a four-month period. \n5. Finally, they conducted an endline household survey approximately four months after baseline to document illness episodes, treatment-seeking behavior, and medication use. \n6. Having all the data, they estimated the causal effect of ACT subsidies on ACT access using OLS regressions with household-clustered robust standard errors and strata fixed effects.", "models": "ordinary least squares regression", "outcome_variable": "ACT access", "independent_variables": "pooled dummies for ACT subsidy levels (80%, 88%, 92%)", "control_variables": "household head age and a full set of strata dummies (drug shop, household’s distance to the drug shop (in quartiles), presence of children in the household)", "tools_software": "not stated"}, "results": {"summary": "The coefficient for “Any ACT subsidy” is 0.187 with robust standard errors clustered at the household level of 0.038, significant at the 1 percent level.", "numerical_results": [{"outcome_name": "ACT access", "value": "0.187", "unit": "NA", "effect_size": "not stated", "confidence_interval": {"lower": "not stated", "upper": "not stated", "level": "not stated"}, "p_value": "<0.01", "statistical_significance": "true", "direction": "positive"}]}, "metadata": {"original_paper_id": "http://dx.doi.org/10.1257/aer.20130267", "original_paper_title": "Price Subsidies, Diagnostic Tests, and Targeting of Malaria Treatment: Evidence from a Randomized Controlled Trial.", "original_paper_code": "not stated", "original_paper_data": "not stated"}}}, "expected_post_registration_2": {"original_study": {"claim": {"hypothesis": "Providing ACT at a subsidized price increases the likelihood of taking ACT.", "hypothesis_location": "The expectation is developed in the discussion of the price-intervention design and the motivation for assessing whether lowering the ACT price increases medication uptake in Section II: Theoretical Framework.", "statement": "The regression results indicate that the presence of an ACT subsidy has a positive and statistically significant effect on obtaining ACTs; the estimated coefficient is 0.187.", "statement_location": "Regression results table 2: coefficient on the indicator for any ACT subsidy reported as 0.187 with a standard error of 0.038, significant at the one-percent level.", "study_type": "Experimental (Randomized Controlled Trial)."}, "data": {"source": "Data collected by the research team during the ACT price-subsidy randomized controlled trial implemented in Western Kenya.", "wave_or_subset": "May - December 2009.", "sample_size": "2789.", "unit_of_analysis": "Individual household illness episode or voucher recipient.", "access_details": "not stated", "notes": "The subsidy level was randomized, creating exogenous variation in the cost of ACT to households. Standard errors are clustered at the household level."}, "method": {"description": "The study estimates the causal influence of subsidizing ACTs on household take-up by randomly varying ACT prices and regressing ACT acquisition on the subsidy indicator.", "steps": ["Randomly assign different ACT price levels to households through voucher offers.", "Observe whether each household obtained ACT following an illness episode.", "Construct an indicator variable for whether the ACT was subsidized in that household’s voucher assignment.", "Estimate a linear probability model of ACT take-up on the subsidy indicator.", "Cluster standard errors at the household level to account for repeated observations within households."], "models": "Linear probability regression of ACT acquisition on the indicator for receiving an ACT subsidy.", "outcome_variable": "Indicator equal to 1 if the household acquired ACT for an illness episode.", "independent_variables": "Indicator for assignment to any ACT subsidy.", "control_variables": "not stated", "tools_software": "not stated"}, "results": {"summary": "Assignment to an ACT subsidy significantly increases the probability that households obtain ACTs. The estimated effect size is positive and statistically significant at the one-percent level.", "numerical_results": [{"outcome_name": "ACT take-up", "value": 0.187, "unit": "change in probability of obtaining ACT associated with subsidy assignment", "effect_size": "linear probability coefficient = 0.187", "confidence_interval": {"lower": "not stated", "upper": "not stated", "level": "not stated"}, "p_value": "< 0.01", "statistical_significance": 1, "direction": "positive"}]}, "metadata": {"original_paper_id": "10.1257/aer.20130267", "original_paper_title": "Price Subsidies, Diagnostic Tests, and Targeting of Malaria Treatment: Evidence from a Randomized Controlled Trial", "original_paper_code": "not stated", "original_paper_data": "not stated"}}}, "expected_post_registration_3": null}
{"study_id": "9", "original_paper_pdf": "9/input/original_paper.pdf", "initial_details": "[CLAIM]\nThe authors find that as the number of parties in the system increases, party dispersion increases and the effect is statistically significant for both policy dimensions (coefficient on log count of parties in system term = 0.39, robust SE clustered by country = 0.14, coefficient falls within a 95% confidence interval)\n\n[HYPOTHESIS]\nOn the economic policy dimension, the number of parties in the party system is positively associated with policy dispersion.", "replication_data_files": ["9/input/replication_data/Andrews-Money_Replication.do", "9/input/replication_data/CMP_final.dta", "9/input/replication_data/CPDS_final.dta"], "human_preregistration": "9/gt/human_preregistration.pdf", "human_report": "9/gt/human_report.pdf", "expected_post_registration": {"original_study": {"claim": {"hypothesis": "On the economic policy dimension, the number of parties in the party system is positively associated with policy dispersion.", "hypothesis_location": "section Predictions of spatial theory, it is also discussed in the abstract and introduction", "statement": "The authors find that as the number of parties in the system increases, party dispersion increases and the effect is statistically significant for both policy dimensions (coefficient on log count of parties in system term = 0.39, robust SE clustered by country = 0.14, coefficient falls within a 95% confidence interval).", "statement_location": "Table 4, Economic Policy Model 1a, Log count of parties in system.", "study_type": "Observational"}, "data": {"source": "Comparative manifesto project (CMP) - the data was published in Ian Budge, Hans-Dieter Klingemann, Andrea Volkens, Judith Bara and Eric Tanenbaum, Mapping Policy Preferences: Estimates for Parties, Electors, and Governments 1945–1998 (Oxford: Oxford University Press, 2001).", "wave_or_subset": "data from 1945 to 1999", "sample_size": "20", "unit_of_analysis": "country-election (party system per one election in established parliamentary democracies)", "access_details": "the CMP data was published so, most likely, it is open access", "notes": "Parties were included if they gained at least 1% of parliamentary seats in two consecutive elections. Parties were dropped if below 1% in three consecutive elections. Data was pooled across countries to create a shared two-dimensional policy space. The number of parties is measured both by a simple count of parties meeting the 1% seat threshold and by the effective number of parties based on seat shares. Party dispersion is measured as the distance between the two most extreme parties along each policy dimension (how far apart political parties are from each other in terms of their policy positions)."}, "method": {"description": "The study examines how the number of parties in a political system affects party dispersion in a two-dimensional policy space (economic and social dimensions).", "steps": "1. Select twenty established parliamentary democracies (Australia, Austria, Belgium, Canada, Denmark, France, Germany, Greece, Iceland, Ireland, Italy, Luxembourg, Netherlands, Norway, New Zealand, Portugal, Spain, Sweden, Switzerland, United Kingdom). \n2. Include parties with at least 1% of seats in parliament in two consecutive elections. \n3. Exclude parties below 1% in three consecutive elections. \n4.Compute number of parties. \n4. Define two policy dimensions (economic and social) using specific CMP categories (p. 814., note 37).\n5. Conduct principal components analysis (PCA) to estimate party positions along each dimension. \n6. Calculate party dispersion as the distance between the two most extreme parties on each dimension within each country–election. \n7. Regress dispersion on the log of number of parties, with robust standard errors clustered by country.", "models": "ordinary least squares regression of log party dispersion on log count of parties and electoral rules", "outcome_variable": "party dispersion", "independent_variables": "log of the count of parties in the system, electoral rules (dummy: 1 = single-member district, 0 = proportional)", "control_variables": "not stated", "tools_software": "not stated"}, "results": {"summary": "As the number of parties in the system increases, party dispersion increases and the effect is statistically significant (coefficient on log count of parties in system term = 0.39, robust SE clustered by country = 0.14, coefficient falls within a 95% confidence interval).", "numerical_results": [{"outcome_name": "Economic policy", "value": "0.39", "unit": "NA", "effect_size": "not stated", "confidence_interval": {"lower": "not stated", "upper": "not stated", "level": "95%"}, "p_value": "not stated explicitly; a 95% confidence interval indicates statistical significance.", "statistical_significance": "true", "direction": "positive"}]}, "metadata": {"original_paper_id": "10.1017/S0007123409990172", "original_paper_title": "The Spatial Structure of Party Competition: Party Dispersion within a Finite Policy Space.", "original_paper_code": "not stated", "original_paper_data": "not stated"}}}, "expected_post_registration_2": {"original_study": {"claim": {"hypothesis": "As the number of parties in a party system increases, the dispersion of parties along the economic policy dimension increases.", "hypothesis_location": "Introduction and Predictions of Spatial Theory sections (pages 1–4).", "statement": "The empirical analysis shows that systems with more parties exhibit greater dispersion: in regressions of economic-policy dispersion on the number of parties, the log count of parties has a positive and statistically significant association with the log distance between the most extreme parties along the economic dimension.", "statement_location": "Results of Empirical Analysis and Regression Analysis sections, pages 13–17, and Table 4, where the coefficient on the log count of parties predicting log economic dispersion is reported as 0.39 with a robust standard error of 0.14 and marked as statistically significant within a 95% confidence interval.", "study_type": "Observational"}, "data": {"source": "Comparative Manifesto Project (CMP) data on party manifestos for parliamentary parties, combined with information on parliamentary seat shares and electoral systems in twenty established parliamentary democracies.", "wave_or_subset": "All national parliamentary elections from the first election after the Second World War through the last election in the 1990s for twenty parliamentary democracies (Australia, Austria, Belgium, Canada, Denmark, France, Germany, Greece, Iceland, Ireland, Italy, Luxembourg, Netherlands, Norway, New Zealand, Portugal, Spain, Sweden, Switzerland, United Kingdom).", "sample_size": "295 country–election–year party systems (Table 1).", "unit_of_analysis": "Party system in a given country and election year (country–election–year).", "access_details": "not stated", "notes": "The number of parties in a party system is measured in two ways: (1) a count of parliamentary parties that win at least 1% of seats in at least two consecutive elections (with parties dropped if they fall below 1% for three consecutive elections), and (2) the effective number of parties in parliament based on the Laakso–Taagepera formula. Party positions are estimated from CMP manifesto categories via principal components analysis to construct an economic and a social policy dimension. Party dispersion is measured as the distance between the two most extreme parties on each dimension within each party system."}, "method": {"description": "The authors pool party-position data from twenty parliamentary democracies over time, estimate parties’ locations on economic and social policy dimensions using CMP manifesto data, compute dispersion as the distance between the most extreme parties in each system, and regress dispersion on the number of parties (and electoral rules) to test whether systems with more parties are more dispersed.", "steps": ["Use Comparative Manifesto Project data to code party policy positions for parliamentary parties in twenty established parliamentary democracies from 1945 to 1999.", "Construct two policy dimensions (economic and social) by performing principal components analysis on selected CMP issue categories, obtaining a score on each dimension for every party in every election year.", "Define membership in the party system using a 1% parliamentary seat-share threshold over at least two consecutive elections, dropping parties that fall below 1% for three consecutive elections; for each country–election–year, count how many parties belong to the system.", "For each country–election–year, identify the two most extreme parties along the economic policy dimension and compute the distance between them to obtain a measure of economic dispersion (and similarly for social dispersion).", "Assemble a dataset of 295 party systems (country–election–years) including dispersion measures, the count of parties, the effective number of parties, and an indicator for single-member-district versus proportional electoral rules.", "Log-transform dispersion and the number-of-parties measures, and estimate ordinary least squares regressions of logged dispersion on logged number of parties, electoral-system dummy, and the lagged dependent variable, clustering robust standard errors by country.", "Interpret the coefficient on the logged count of parties as the elasticity of party dispersion with respect to the number of parties in the system."], "models": "Ordinary least squares regression models of the natural log of party dispersion on the natural log of the number of parties in the party system and electoral rules, including a lagged dependent variable and robust standard errors clustered by country.", "outcome_variable": "Natural log of the distance between the two most extreme parties along the economic policy dimension in each country–election–year party system.", "independent_variables": "Natural log of the count of parties in the party system (based on parliamentary parties with at least 1% of seats in at least two consecutive elections).", "control_variables": "Indicator for single-member-district electoral systems versus proportional representation; lagged value of logged economic dispersion.", "tools_software": "not stated"}, "results": {"summary": "Both descriptive comparisons and regression models show that party systems with more parties are more dispersed in policy space: mean distances between extreme parties increase as the number of parties rises up to about five parties, and in regression models the log count of parties has a positive and statistically significant effect on the log distance between the most extreme parties along the economic policy dimension, while electoral rules have no independent effect once the number of parties is controlled for.", "numerical_results": [{"outcome_name": "Logged economic-policy dispersion (distance between most extreme parties) as a function of logged number of parties in the system", "value": 0.39, "unit": "OLS elasticity coefficient (percent change in economic-policy dispersion associated with a 1 percent increase in the count of parties)", "effect_size": "OLS regression coefficient = 0.39 (Model 1a)", "confidence_interval": {"lower": "not stated", "upper": "not stated", "level": 95}, "p_value": "< 0.05", "statistical_significance": 1, "direction": "positive"}]}, "metadata": {"original_paper_id": "10.1017/S0007123409990172", "original_paper_title": "The Spatial Structure of Party Competition: Party Dispersion within a Finite Policy Space", "original_paper_code": "not stated", "original_paper_data": "not stated"}}}, "expected_post_registration_3": null}
{"study_id": "10", "original_paper_pdf": "10/input/original_paper.pdf", "initial_details": "[CLAIM]\nThe focal test result concerns the location of the estimated coefficient “Imports from the South”. The dependent variable is national affluence. The result was a statistically significant estimated coefficient for “Imports from the South” (b=.910,SE=.104,p<.001)\n\n[HYPOTHESIS]\nImports from the South will be positively associated with national affluence.", "replication_data_files": ["10/input/replication_data/processed_data.csv", "10/input/replication_data/finaldata_noNA.csv", "10/input/replication_data/KMYR.do"], "human_preregistration": "10/gt/human_preregistration.pdf", "human_report": "10/gt/human_report.pdf", "expected_post_registration": {"original_study": {"claim": {"hypothesis": "Imports from the South will be positively associated with national affluence.", "hypothesis_location": "it is discussed in the introduction and literature review sections", "statement": "The focal test result concerns the location of the estimated coefficient “Imports from the South”. The dependent variable is national affluence. The result was a statistically significant estimated coefficient for “Imports from the South” (b=.910,SE=.104,p<.001).", "statement_location": "Table 2 Model 4", "study_type": "Observational"}, "data": {"source": "Organization for Economic Cooperation and Development (OECD), specifically:\nnational affluence: OECD’s Annual National Accounts, volume 1: Comparative Tables;\nimport/export data: International Trade by Commodities Database;\nunemployment: Labour Force Statistics—Summary Tables;", "wave_or_subset": "1970-2003", "sample_size": "566", "unit_of_analysis": "country-year", "access_details": "not stated; there are only mentions which OECD database was used for which variable. they are referenced in the reference section", "notes": "countries used in this study are Australia, Austria, Belgium, Canada, Denmark, Finland, France, Germany, Ireland, Italy, Japan, the Netherlands, New Zealand, Norway, Sweden, Switzerland, the United Kingdom, and the United States;\nThe study defines the South as Africa, Asia, Central and South America, and Oceania; the study defines the North as Europe and North America. Adjustments were made by moving Mexico and Turkey (from North America and Europe, respectively) to the South, and by moving Australia and New Zealand (from Oceania) and Israel, Japan, and South Korea (from Asia) to the North. To facilitate international comparison, values for imports and exports are expressed as a percentage of GDP for all countries. \nnational affluence is measured as a country’s gross domestic product (GDP) divided by its total population, with GDP expressed in U.S. dollars at prices and purchasing parities (PPP) from the year 2000."}, "method": {"description": "The study analyzes why the world’s most economically advanced countries have deindustrialized over the last few decades. Model 4 estimates the relationship between imports from the South and national affluence.", "steps": "1. Combine all datasets including 18 OECD countries (see notes) for the years 1970–2003 using databases described in the source field.\n2. Construct national affluence as GDP per capita (in 2000 PPP U.S. dollars). \n3. Define imports from the South as the total value of manufactured goods that each OECD country imports from Southern countries, and exports to the South as the total value of manufactured goods that each OECD country exports to Southern countries.\n4. Normalize all trade variables express each trade measure as a percentage of GDP.\n5. Create control variables: the unemployment rate (from OECD data) and a set of period dummies (1975–79, 1980–84, 1985–89, 1990–94, 1995–99, and 2000–2003) to capture temporal and macroeconomic effects.\n6. Estimate Model 4 with national affluence as the DV and import/export data as the IV. Include unemployment and period dummies as controls.", "models": "two-way fixed-effects regression", "outcome_variable": "national affluence", "independent_variables": "imports from the South, exports to the South", "control_variables": "unemployment, period indicators (dummies)", "tools_software": "not stated"}, "results": {"summary": "Model 4 shows a significant positive effect of imports from the South on national affluence (b = .910, SE = .104, p < .001).", "numerical_results": [{"outcome_name": "national affluence", "value": "0.910", "unit": "NA", "effect_size": "not stated", "confidence_interval": {"lower": "not stated", "upper": "not stated", "level": "not stated"}, "p_value": "<0.001", "statistical_significance": "true", "direction": "positive"}]}, "metadata": {"original_paper_id": "0002-9602/2009/11406-0002", "original_paper_title": "Explaining Deindustrialization: How Affluence, Productivity Growth, and Globalization Diminish Manufacturing Employment.", "original_paper_code": "not stated", "original_paper_data": "not stated"}}}, "expected_post_registration_2": {"original_study": {"claim": {"hypothesis": "Greater levels of imports originating from South correspond with higher levels of national affluence in advanced economies.", "hypothesis_location": "Discussion of indirect effects of global South trade in the conceptual framework.", "statement": "The analysis demonstrates that inflows of manufactured goods from developing countries have a strong positive association with national affluence. The regression coefficient for imports from the South in the affluence equation is reported as 0.910 with a standard error of 0.104, and is statistically significant at the 0.001 level.", "statement_location": "Table 2, column with national affluence as the dependent variable, row labeled “Imports from the South,” showing coefficient = .910, SE = .104, significance p < .001.", "study_type": "Observational (cross-national panel analysis)."}, "data": {"source": "OECD STAN (Structural Analysis) Database for manufacturing employment and workforce data; OECD Annual National Accounts for GDP and population used to construct national affluence; OECD International Trade by Commodities Database for imports and exports disaggregated by region and SITC codes; UN National Accounts Main Aggregates Database for real value added in manufacturing and service sectors; OECD Labour Force Statistics for unemployment data.", "wave_or_subset": "Annual observations for 18 OECD countries from 1970 to 2003.", "sample_size": "612 country–year observations.", "unit_of_analysis": "Country–year.", "access_details": "not stated", "notes": "National affluence is measured using logged GDP per capita. Imports from the South are calculated as the logged value of manufactured imports originating from non-OECD countries. All continuous variables are logged. Country fixed effects and year dummies appear in the models."}, "method": {"description": "The study estimates panel regression models (OLS) to assess how trade with developing countries relates to changes in national affluence, net of other macroeconomic conditions.", "steps": ["Assemble annual country-level data on GDP per capita, trade flows with developing countries, productivity measures, and labor market indicators for OECD nations.", "Construct measures of imports from the South using COMTRADE data and log-transform GDP per capita to represent national affluence.", "Specify a fixed-effects regression model of national affluence on imports from the South and control variables.", "Include year indicators to adjust for common temporal shocks.", "Cluster standard errors by country and estimate the regression.", "Interpret the coefficient on imports from the South as the estimated association with national affluence."], "models": "Country fixed-effects panel regression with year dummies predicting logged national affluence.", "outcome_variable": "Logged national affluence (GDP per capita).", "independent_variables": "Logged imports from the South.", "control_variables": "Exports to the South, unemployment, Period Indicators .", "tools_software": "Stata"}, "results": {"summary": "Imports from developing countries exhibit a strong and statistically significant positive association with national affluence across OECD nations. The magnitude of the coefficient indicates that increases in such imports correspond with higher levels of GDP per capita.", "numerical_results": [{"outcome_name": "Logged national affluence", "value": 0.91, "unit": "unstandardized regression coefficient (change in logged affluence per unit change in logged imports from the South)", "effect_size": "OLS fixed-effects coefficient = 0.910", "confidence_interval": {"lower": "not stated", "upper": "not stated", "level": "not stated"}, "p_value": "< .001", "statistical_significance": 1, "direction": "positive"}]}, "metadata": {"original_paper_id": "not stated", "original_paper_title": "Explaining Deindustrialization: How Affluence, Productivity Growth, and Globalization Diminish Manufacturing Employment", "original_paper_code": "not stated", "original_paper_data": "not stated"}}}, "expected_post_registration_3": null}
{"study_id": "11", "original_paper_pdf": "11/input/original_paper.pdf", "initial_details": "[CLAIM]\nControlling for cognitive abilities, age, gender, socio-economic status, parental education, and indicators of cultural capital, the analysis revealed a general positive trend between bilingualism and English foreign language achievement (estimate = 2.68; p < .01)\n\n[HYPOTHESIS]\nBilingual group membership will be positively associated with foreign language achievement when controlling for background variables", "replication_data_files": ["11/input/replication_data/Replication attempt code (FINAL).R", "11/input/replication_data/Final replication dataset.rds"], "human_preregistration": "11/gt/human_preregistration.pdf", "human_report": "11/gt/human_report.pdf", "expected_post_registration": {"original_study": {"claim": {"hypothesis": "Bilingual group membership will be positively associated with foreign language achievement when controlling for background variables.", "hypothesis_location": "the hypothesis is stated in the 1.4. Research questions section; it is also discussed in the abstract, introduction, section 1.2. Factors affecting language learning.", "statement": "Controlling for cognitive abilities, age, gender, socio-economic status, parental education, and indicators of cultural capital, the analysis revealed a general positive trend between bilingualism and English foreign language achievement (estimate = 2.68; p < .01).", "statement_location": "Table 3, Model B (Bilingual=1)", "study_type": "Observational"}, "data": {"source": "the main source: Assessment of Reading and Mathematics Development Study (ELEMENT); for socioeconomic status data: International Socio-Economic Index.", "wave_or_subset": "the sixth grade elementary school cohort", "sample_size": "2835", "unit_of_analysis": "individual - German 6th graders nested in 134 elementary school classes", "access_details": "access details are not specified; no restrictions are mentioned, perhaps the datasets are open access.", "notes": "Data were collected from a sample of students from a major European city, with about 15% of students speaking a language other than German at home. The sample is representative of public elementary school students. English achievement was assessed with a Cloze test, scaled using one-parameter item response theory in ConQuest with weighted likelihood estimates (WLEs; mean = 100, SD = 20). General cognitive abilities were measured using a composite score from two subtests of the CFT4-12R (verbal and figural analogies). All analyses used five imputed datasets to replace missing values, based on a background model including individual-level factors (grades, self-concept, interest, motivation) and classroom-level factors (achievement, socio-economic status, percentage of students with immigration background). Results were combined using MPlus 5.21 with type = imputation."}, "method": {"description": "The authors examined the effect of immigrant bilingualism on English as a foreign language achievement. They controlled for cognitive, demographic, and socio-cultural factors.", "steps": "1. Download the data and identify bilingual and monolingual groups based on language spoken at home. \n2. Exclude 111 students with no language information.\n3. Classify students into the monolingual group (n=1896) and the bilingual group (n=939).\n4. Assess English language achievement using a Cloze test, scaled with one-parameter item response theory in ConQuest using weighted likelihood estimates (WLEs; mean=100, SD=20).\n5. Measure general cognitive abilities (by the authors) using a composite score from two subtests of the CFT4-12R (verbal and figural analogies).\n6. Collect control variables from the dataset: age, gender, socio-economic status (ISEI data), parental education, and cultural capital. \n7. Handle missing data using multiple imputation.\n8. Run multiple linear regression in MPlus 5.21 with type = complex.", "models": "linear regression", "outcome_variable": "English achievement", "independent_variables": "bilingual group membership", "control_variables": "general cognitive abilities, gender, age, socio-economic status, parental education, and cultural capital", "tools_software": "For multiple regression: MPlus 5.21 taking into account the nested nature (students in classes) of the dataset (type = complex). For items scaling: ConQuest."}, "results": {"summary": "After controlling for cognitive abilities, age, gender, socio-economic status, parental education, and indicators of cultural capital, bilingual students showed a general positive trend between bilingualism and English foreign language achievement (estimate = 2.68; p < .01).", "numerical_results": [{"outcome_name": "English achievement", "value": "2.68", "unit": "points (raw Cloze test scores scaled to mean = 100, SD = 20)", "effect_size": "not stated", "confidence_interval": {"lower": "not stated", "upper": "not stated", "level": "not stated"}, "p_value": "< .01", "statistical_significance": "true", "direction": "positive"}]}, "metadata": {"original_paper_id": "http://dx.doi.org/10.1016/j.learninstruc.2014.12.001", "original_paper_title": "The effect of speaking a minority language at home on foreign language learning.", "original_paper_code": "not stated", "original_paper_data": "not stated"}}}, "expected_post_registration_2": {"original_study": {"claim": {"hypothesis": "Bilingual students, on average, achieve higher scores in English as a foreign language than their monolingual classmates, once individual and family background characteristics are taken into account.", "hypothesis_location": "Section 1.4 Research questions, Hypothesis 1a.", "statement": "After adjusting for general cognitive abilities, age, gender, socio-economic status, parental education, and cultural capital, membership in the bilingual group is positively related to English foreign language achievement: bilingual students score about 2.68 points higher than monolinguals on the English scale, and this effect is statistically significant.", "statement_location": "Results section 3.2, Table 3 ('Multiple regression models explaining English achievement of bilinguals'), Model B.", "study_type": "Observational"}, "data": {"source": "Secondary analysis of the ELEMENT study (Assessment of Reading and Mathematics Development Study), a large-scale assessment conducted in public elementary schools in a major German city.", "wave_or_subset": "Sixth-grade of the ELEMENT cohort.", "sample_size": "2,835 students nested in 134 elementary school classes.", "unit_of_analysis": "Individual student.", "access_details": "not stated", "notes": "Students attend public elementary schools; the sample is representative for sixth-grade students in the city. Language-group classification is based on parent (and, if missing, student) reports of languages spoken regularly at home. The analytic sample includes a monolingual German group (n = 1,896) and several bilingual groups (Arabic–German, Chinese–German, Polish–German, Turkish–German, and a heterogeneous ‘other bilingual’ group). Missing data were handled via multiple imputation (five imputed datasets), and analyses accounted for the clustering of students within classes using complex survey procedures in Mplus 5.21."}, "method": {"description": "The authors used multiple regression models on a large, imputed sixth-grade dataset to test whether being bilingual (speaking a minority language at home plus German) is associated with higher English foreign language achievement compared to monolingual German students, controlling for cognitive and socio-cultural background variables.", "steps": ["Use ELEMENT study data to identify students’ home language(s) from parent reports (and student reports when parent data are missing), and classify students as monolingual German or bilingual (speaking German and another language regularly at home).", "Measure English foreign language achievement in grade 6 with a Cloze test consisting of four texts and 91 word-completion items, scaled using one-parameter item response theory; obtain individual weighted likelihood estimates (WLEs).", "Collect background covariates: general cognitive abilities (from the CFT4-12R verbal and figural analogies subtests in grade 4), age, gender, socio-economic status (HISEI), parental education, and cultural capital (number of books at home).", "Handle missing data using multiple imputation (five imputed datasets) based on a background model including individual and classroom-level factors; combine results across imputations.", "Estimate an uncontrolled regression model of English achievement on a binary indicator for bilingual-group membership (Model A).", "Estimate a controlled multiple regression model of English achievement on bilingual-group membership, adding general cognitive abilities, age, gender, socio-economic status, parental education, and cultural capital as covariates (Model B).", "Account for the nested data structure (students within classes) by using complex survey options in Mplus 5.21 when fitting the regression models."], "models": "Multiple linear regression models predicting English foreign language achievement from bilingual-group membership and background covariates.", "outcome_variable": "English language achievement in grade 6.", "independent_variables": "Binary indicator for group membership (1 = bilingual student who speaks a minority language at home and German; 0 = monolingual German student).", "control_variables": "General cognitive abilities (CFT4-12R composite), age (centered), gender (girls = 1), socio-economic status (HISEI, z-score), parental education (highest parental qualification, z-score), and cultural capital (number of books at home, z-score).", "tools_software": "Mplus 5.21"}, "results": {"summary": "In the uncontrolled model, bilingual students score lower in English than monolinguals. However, once general cognitive abilities, age, gender, socio-economic status, parental education, and cultural capital are held constant, the sign reverses: bilingual-group membership is positively and significantly associated with English foreign language achievement, indicating a small advantage for bilinguals over monolinguals.", "numerical_results": [{"outcome_name": "English foreign language achievement (Cloze test WLE score)", "value": 2.68, "unit": "score points on the English achievement scale (M = 100, SD = 20)", "effect_size": "unstandardized multiple regression coefficient for bilingual-group membership in Model B of Table 3", "confidence_interval": {"lower": "not stated", "upper": "not stated", "level": "not stated"}, "p_value": "< .01", "statistical_significance": 1, "direction": "positive"}]}, "metadata": {"original_paper_id": "10.1016/j.learninstruc.2014.12.001", "original_paper_title": "The effect of speaking a minority language at home on foreign language learning", "original_paper_code": "not stated", "original_paper_data": "not stated"}}}, "expected_post_registration_3": {"original_study": {"claim": {"hypothesis": "Previous research…suggests that, once controlling for individual and familial background factors, the bilingual students will, on average, have higher scores in English as a foreign language compared to the monolingual group. Therefore we hypothesize similar advantages for the bilingual students in the present sample (Hypothesis 1a).", "hypothesis_location": "Section 1: Introduction; Subsection 1.4: Research questions; p. 78-79", "statement": "Given comparable individual and familiar background characteristics bilingual group membership is positively associated with English foreign language achievement.", "statement_location": "Section 3: Results; Subsection 3.2: Bilingualism and English achievement; p. 80.", "study_type": "Observational"}, "data": {"source": "Assessment of Reading and Mathematics Development Study", "wave_or_subset": "NA", "sample_size": "2946", "unit_of_analysis": "Student", "access_details": "not stated", "notes": "The final sample size (after cleaning) was 2835 students nested in 134 elementary\nschool classes."}, "method": {"description": "The authors use regression analysis, controlling for background characteristics, to explore the effect of bilingualism on English language achievement in school.", "steps": "(1) Clean the data by keeping only those with language information; (2) the bilingual group was divided into five proficiency groups from a standardized competency scale; (3) construct the English language achievement measure using weighted likelihood estimates (WLEs) for individual person parameters; (4) conducted all analyses using five imputed datasets, in which the missing values were replaced by plausible values.", "models": "uncontrolled regression model; regression model including controls", "outcome_variable": "English language achievement", "independent_variables": "Bilingualism", "control_variables": "general cognitive abilities, gender, age, socio-economic status, parental education, and cultural capital", "tools_software": "MPlus 5.21"}, "results": {"summary": "the bilingual group had a relatively slight advantage (b = 2.68), which can be interpreted as just over a quarter of a school year's achievement.", "numerical_results": [{"outcome_name": "English achievement", "value": "2.68", "unit": "not stated", "effect_size": "just over a quarter of a school year's achievement", "confidence_interval": {"lower": "not stated", "upper": "not stated", "level": "not stated"}, "p_value": "<0.01", "statistical_significance": "1% level", "direction": "Positive"}]}, "metadata": {"original_paper_id": "10.1016/j.learninstruc.2014.12.001.", "original_paper_title": "The effect of speaking a minority language at home on foreign language learning.", "original_paper_code": "not stated", "original_paper_data": "not sated"}}}}
{"study_id": "12", "original_paper_pdf": "12/input/original_paper.pdf", "initial_details": "[CLAIM]\nThe income is substantially higher for low-caste households residing in villages dominated by a low caste. The estimation results show the robustness of the positive relationship between agricultural income and residing in a low-caste dominated village. [The coefficient for “Low-caste villages” is 566.5 with robust standard errors of 209, significant at the 1 percent level.]\n\n[HYPOTHESIS]\nAmong low-caste households, residing in villages dominated by lower castes is associated with greater agricultural income compared to residing in villages dominated by upper castes", "replication_data_files": ["12/input/replication_data/anderson_2011_replication_data_analysis.do", "12/input/replication_data/analysis_data.dta"], "human_preregistration": "12/gt/human_preregistration.pdf", "human_report": "12/gt/human_report.docx", "expected_post_registration": {"original_study": {"claim": {"hypothesis": "Among low-caste households, residing in villages dominated by lower castes is associated with greater agricultural income compared to residing in villages dominated by upper castes.", "hypothesis_location": "the abstract and introduction", "statement": "The income is substantially higher for low-caste households residing in villages dominated by a low caste. The estimation results show the robustness of the positive relationship between agricultural income and residing in a low-caste dominated village. [The coefficient for “Low-caste villages” is 566.5 with robust standard errors of 209, significant at the 1 percent level.].", "statement_location": "Table 3 column 1", "study_type": "Observational"}, "data": {"source": "The primary data used in this paper were collected by a team of researchers based at the World Bank and in India.", "wave_or_subset": "1997–1998", "sample_size": "1295", "unit_of_analysis": "household", "access_details": "not stated, the data source is not very clear either. However, there is information that additional materials are here: http://www.aeaweb.org/articles.php?doi=10.1257/app.3.1.239; there is no information if the website includes the data or code.", "notes": "Dominant caste refers to the caste group owning the majority of land. The villages of study are located in south and southeastern Uttar Pradesh and north and central Bihar. The field survey was administered in villages drawn at random from 12 districts in Uttar Pradesh and 13 districts in Bihar. A total of 120 villages, with an overall sample size of 2,250 households, were sampled: 57 villages in Bihar and 63 in Uttar Pradesh. All of the study villages are rural and the economies in these areas are primarily dependent on agriculture. The author dropped Muslim households from the analysis (which comprises only 2 percent of the sample in Hindu dominated villages). Regression disturbance terms are clustered at the village level. There are district and state fixed effects"}, "method": {"description": "The study uses OLS regression to examine whether lower-caste households have higher agricultural income when residing in villages dominated by lower castes, relative to villages dominated by upper castes.", "steps": "1. Get the data.\n2. Restrict the sample to lower-caste households — BAC (Backward Agricultural Castes), OBC (Other Backward Castes), and SC (Scheduled Castes) — and exclude Muslim-dominated villages and Muslim households in Hindu-dominated villages.\n3. Create the binary variable for low-caste dominated villages (1 = low-caste dominated, 0 = high-caste dominated).\n4. Estimate an OLS regression of household crop income per acre on the low-caste village indicator, including household, district, and state-level controls.\n5. Cluster standard errors at the village level.", "models": "ordinary least squares regression", "outcome_variable": "crop income per acre", "independent_variables": "low-caste village", "control_variables": "literacy, land ownership, caste identity, district, state", "tools_software": "not stated"}, "results": {"summary": "Lower-caste households earn significantly more in low-caste dominated villages (“Low-caste villages” coefficient = 566.5, SE = 209, p < 0.01).", "numerical_results": [{"outcome_name": "Household Crop Income", "value": "566.5", "unit": "not stated explicitly (should be local currency though)", "effect_size": "not stated", "confidence_interval": {"lower": "not stated", "upper": "not stated", "level": "not stated"}, "p_value": "< 0.01", "statistical_significance": "true", "direction": "positive"}]}, "metadata": {"original_paper_id": "http://www.aeaweb.org/articles.php?doi=10.1257/app.3.1.239", "original_paper_title": "Caste as an Impediment to Trade.", "original_paper_code": "not stated", "original_paper_data": "not stated"}}}, "expected_post_registration_2": {"original_study": {"claim": {"hypothesis": "Among lower-caste households, the association between being a groundwater buyer and farm performance will be more favorable in villages where landownership dominance is held by a lower caste.", "hypothesis_location": "Section III: Access to Irrigation.", "statement": "The results indicate that the advantage of living in a low-caste dominated village is concentrated among groundwater buyers: the interaction between low-caste village status and water-buyer status is positive and statistically significant in the crop-income regressions.", "statement_location": "Table 4 (OLS estimations with water-market interaction terms), “LCV × water buyer”.", "study_type": "Observational"}, "data": {"source": "UP-Bihar LSMS World Bank survey data, including a household questionnaire and a village questionnaire.", "wave_or_subset": "Field survey collected in 1997-1998 in rural villages in south and southeastern Uttar Pradesh and north and central Bihar; analysis focuses on Hindu-dominated villages and excludes Muslim-dominated villages and Muslim households.", "sample_size": "120 villages and 2,250 households in the full sample; the focal regressions use a lower-caste household sample with 1,295 observations.", "unit_of_analysis": "Household (lower-caste household, BAC/OBC/SC) in a village.", "access_details": "not stated", "notes": "Village dominance is defined by which caste group owns the majority of land (upper-caste dominated vs BAC-dominated). Key irrigation measures include private tubewell ownership and groundwater buying. Regression disturbance terms are clustered at the village level in the household-level crop-income models."}, "method": {"description": "The study compares economic outcomes of lower-caste households across villages where landownership is dominated by either an upper caste or a lower backward agricultural caste group, and then tests whether differences in agricultural outcomes are explained by access to irrigation through private groundwater markets by interacting village dominance with household water-market participation.", "steps": ["Draw villages at random from districts in Uttar Pradesh and Bihar and conduct village- and household-level surveys.", "Classify villages by which caste group owns the majority of land (upper-caste dominated vs BAC-dominated).", "Restrict the main household analysis to lower-caste households (BAC, OBC, SC) and exclude Muslim-dominated villages and Muslim households.", "Construct agricultural outcome measures (e.g., crop income per acre) from reported crop sales and landholdings.", "Measure groundwater market participation using indicators for being a water buyer and tubewell owner.", "Estimate OLS regressions of crop income per acre on low-caste village status, household controls, and fixed effects, clustering standard errors at the village level.", "Estimate interaction models where low-caste village status is interacted with water-buyer status to test whether the village-dominance effect operates through groundwater-market access."], "models": "OLS regression models of crop income per acre with interaction terms between village caste dominance and groundwater market participation (water buyer), with clustered standard errors at the village level and included controls/fixed effects as specified in the table.", "outcome_variable": "Household crop income per acre (crop income per acre of total land).", "independent_variables": "Low-caste village indicator; water buyer indicator; interaction between low-caste village and water buyer (LCV × water buyer).", "control_variables": "Literacy indicator and total land; caste controls; state controls; and (depending on specification) district controls and additional sets of controls such as crop, distance, groundwater, and public-goods controls as indicated in the regression table.", "tools_software": "not stated"}, "results": {"summary": "The interaction analysis shows that the positive association between living in a low-caste dominated village and agricultural performance is concentrated among households that buy groundwater: the interaction between low-caste village status and water-buyer status is positive and statistically significant in the crop-income regression.", "numerical_results": [{"outcome_name": "Crop income per acre", "value": "850.9", "unit": "rupees per acre", "effect_size": "OLS interaction coefficient (LCV × water buyer)", "confidence_interval": {"lower": "not stated", "upper": "not stated", "level": "not stated"}, "p_value": "< 0.01", "statistical_significance": 1, "direction": "positive"}]}, "metadata": {"original_paper_id": "10.1257/app.3.1.239", "original_paper_title": "Caste as an Impediment to Trade", "original_paper_code": "not stated", "original_paper_data": "not stated"}}}, "expected_post_registration_3": {"original_study": {"claim": {"hypothesis": "Given these historical patterns, we may well expect lower castes to fair better in villages where no upper castes are present.", "hypothesis_location": "Section: Introduction; p. 240", "statement": "...lower castes fair significantly better, in terms of household income, if they reside in villages where a lower caste is dominant.", "statement_location": "Section 2: Household Outcomes by Caste Dominance; p.247", "study_type": "Observational"}, "data": {"source": "World Bank survey data on Uttar Pradesh and Bihar in India", "wave_or_subset": "NA", "sample_size": "2,250", "unit_of_analysis": "household", "access_details": "Access details (e.g., restrictions, request process)", "notes": "not stated"}, "method": {"description": "The author uses fixed effects regression analysis to explore the effect of being a low caste member in a village dominated by the lower caste on crop income.", "steps": "(1) Clean the data; (2) construct the Caste Dominance measure by landownership percentage for the castes; (3) Estimate the effects using regression analysis.", "models": "Fixed effects regression", "outcome_variable": "crop income per acre of total land", "independent_variables": "Caste Dominance", "control_variables": "Exogenous household characteristics (such as education, land ownership, and caste identity); district fixed effects; state fixed effects.", "tools_software": "not stated"}, "results": {"summary": "...lower castes fair significantly better, in terms of household income, if they reside in villages where a lower caste is dominant…The estimation results…confirm the robustness of the positive relationship between agricultural income and residing in a low-caste dominated village.", "numerical_results": [{"outcome_name": "Household Crop Income", "value": "566.5", "unit": "", "effect_size": "", "confidence_interval": {"lower": "not stated", "upper": "not stated", "level": "not stated"}, "p_value": "<0.01", "statistical_significance": "1% level", "direction": "Positive", "notes": "No confidence intervals, but robust standard errors are reported (209.0)."}]}, "metadata": {"original_paper_id": "10.1257/app.3.1.239", "original_paper_title": "Caste as an Impediment to Trade", "original_paper_code": "http://www.aeaweb.org/articles.php?doi=10.1257/app.3.1.239", "original_paper_data": "http://www.aeaweb.org/articles.php?doi=10.1257/app.3.1.239"}}}}
{"study_id": "13", "original_paper_pdf": "13/input/original_paper.pdf", "initial_details": "[CLAIM]\neven after controlling for other predictors of trust in the political system, concerns about the effect of immigration on the national community have an impact on trust in politics, that higher concern about immigration is associated with higher distrust in politics.\n\n[HYPOTHESES]\nIndividuals’ concerns about immigration is positively associated with distrust in their country’s parliament.", "replication_data_files": ["13/input/replication_data/.DS_Store", "13/input/replication_data/data_imp_5pct.rds", "13/input/replication_data/data_analysis_code.R", "13/input/replication_data/data_clean_5pct.rds", "13/input/replication_data/data_clean.rds"], "human_preregistration": "13/gt/human_preregistration.pdf", "human_report": "13/gt/human_report.pdf", "expected_post_registration": {"original_study": {"claim": {"hypothesis": "Individuals’ concerns about immigration is positively associated with distrust in their country’s parliament.", "hypothesis_location": "p. 205 Proposition 1; also discussed in the introduction and section Concern about Immigration and Political Trust.", "statement": "even after controlling for other predictors of trust in the political system, concerns about the effect of immigration on the national community have an impact on trust in politics, that higher concern about immigration is associated with higher distrust in politics.", "statement_location": "Table 3, parliament column.", "study_type": "Observational"}, "data": {"source": "European Social Survey (ESS)", "wave_or_subset": "rounds 1-4 (Fieldwork for round 1 was conducted in 2002–3, for round 2, in 2004–5 (except in Italy, where it was conducted in early 2006), for round 3, in 2006–7, and for round 4, in 2008–9.).", "sample_size": "110732", "unit_of_analysis": "individual", "access_details": "the article states: Available at http://www.europeansocialsurvey.org/. no further access details are provided.", "notes": "the analysis excludes the newer democracies of Central and Eastern Europe (CEE ); also, minorities and noncitizens have been omitted."}, "method": {"description": "The study used a three-level hierarchical linear model to test whether higher concern about immigration is associated with higher distrust in politics.", "steps": "1.Combine ESS data into one dataset.\n2. Exclude newer democracies of Central and Eastern Europe (CEE), minorities, and noncitizens.\n3.Construct the DV measuring distrust in parliament by reversing coding of the 0–10 trust item (0 = no trust, 10 = complete trust) so that higher values indicate higher distrust.\n4. Construct the key predictor “concern about immigration” as an index averaging three reversed-coded items (economic effect, cultural effect, overall effect on country).\n5.Add individual-level control variables: \n-Unhappiness: reversed 0–10 scale.\n-Dissatisfaction with life: reversed 0–10 scale.\n-Frequency of meeting friends: reversed 1–7 scale (such that high values represent rarely meeting with friends).\n-Interpersonal distrust: reversed 0–10 scale, index of three items (questions about general trust, fairness, and helpfulness of people).\n-Dissatisfied with country’s economy: reversed 0–10 scale.\n-Dissatisfied with personal income: reversed 0–10 scale.\n-Dissatisfied with health and education system: reversed 0–10 scales.\n-Winner effect: 1 if voted for governing party, 0 otherwise.\n-Vote for far-right party: 1 if respondent voted for anti-immigration party, 0 otherwise.\n-Left-right scale.\n-Household income (standardized across rounds).\n-Age.\n-Education.\n-Gender (0 = male, 1 = female).\n7.Estimate a three-level hierarchical linear model with individuals nested within country-rounds and countries using HLM software.", "models": "multilevel modeling (three-level, linear)", "outcome_variable": "political distrust", "independent_variables": "Concern about immigration", "control_variables": "Unhappiness, Dissatisfaction with life, Frequency of meeting friends, Interpersonal distrust, Dissatisfied with country’s economy, Dissatisfied with personal income, Dissatisfied with health system, Dissatisfied with education system, Winner effect, Voted for far-right party in last general election, Left-right scale, HH income, Age, Education, Female", "tools_software": "HLM"}, "results": {"summary": "Higher concern about immigration is associated with higher distrust in politics (b=0.17 SE=0.00 p=0.00).", "numerical_results": [{"outcome_name": "distrust parliament", "value": "0.17", "unit": "NA", "effect_size": "not stated", "confidence_interval": {"lower": "not stated", "upper": "not stated", "level": "not stated"}, "p_value": "0.000", "statistical_significance": "true", "direction": "positive"}]}, "metadata": {"original_paper_id": "S0043887112000032", "original_paper_title": "The Cultural Divide in Europe Migration, Multiculturalism, and Political Trust.", "original_paper_code": "not stated", "original_paper_data": "not stated"}}}, "expected_post_registration_2": {"original_study": {"claim": {"hypothesis": "Individuals who are more worried about immigration’s impact on the national community will report greater distrust of their country’s parliament.", "hypothesis_location": "Proposition 1(p.7) linking greater concern about immigration’s impact on the national community to greater distrust of politicians and political institutions.", "statement": "The multilevel models show a positive and statistically significant association between concern about immigration and distrust in parliament, even when controlling for a wide set of alternative predictors of political distrust.", "statement_location": "Table 3 (Three-Level Model of Distrust in Politics), row “Concern about immigration.”", "study_type": "Observational"}, "data": {"source": "European Social Survey (ESS).", "wave_or_subset": "ESS rounds 1–4 (fieldwork: 2002–2003; 2004–2005; 2006–2007; 2008–2009).", "sample_size": "110,732", "unit_of_analysis": "Individual survey respondent (nested within country-round and country).", "access_details": "Available via the European Social Survey website.", "notes": "The dependent variable is political distrust measured using 11-point (0–10) trust items for national institutions."}, "method": {"description": "The study uses three-level multilevel models to estimate how individual concern about immigration relates to distrust in national political institutions (including parliament), while accounting for respondents nested within survey rounds and countries and controlling for alternative explanations of political distrust.", "steps": ["Select ESS respondents from rounds 1–4 and retain countries included in the analysis.", "Construct the dependent variable for distrust in parliament from the 0–10 trust item (coded so that higher values indicate more distrust).", "Construct the key independent variable measuring concern about immigration’s impact on the national community.", "Include individual-level controls capturing pessimism/alienation, political attitudes, evaluations of government performance and the economy, and social capital.", "Estimate a three-level multilevel model (individuals nested within country-round and country).", "Compute coefficients and standard errors using HLM, as reported in the results tables."], "models": "Three-level multilevel model (HLM) with respondents nested within country-round (level 2) and country (level 3).", "outcome_variable": "Distrust in parliament (0–10 scale, higher = more distrust).", "independent_variables": "Concern about immigration.", "control_variables": "Unhappiness; dissatisfaction with life; frequency of meeting friends; interpersonal distrust; dissatisfaction with the country’s economy; dissatisfaction with personal income; dissatisfaction with the health system; dissatisfaction with the education system; winner effect; voted for far-right party in last general election; left-right scale; household income (standardized); age; education; female; plus level-2 and level-3 covariates included in the Table 3 model.", "tools_software": "HLM software"}, "results": {"summary": "Concern about immigration is positively and significantly related to distrust in parliament in the three-level model, and this association remains after controlling for a broad set of alternative predictors of political distrust.", "numerical_results": [{"outcome_name": "Distrust in parliament", "value": "1.7", "unit": "units on the 0–10 distrust scale", "effect_size": "multilevel regression coefficient", "confidence_interval": {"lower": "not stated", "upper": "not stated", "level": "not stated"}, "p_value": "0.000", "statistical_significance": 1, "direction": "positive"}]}, "metadata": {"original_paper_id": "10.1017/S0043887112000032", "original_paper_title": "The Cultural Divide in Europe: Migration, Multiculturalism, and Political Trust", "original_paper_code": "not stated", "original_paper_data": "not stated"}}}, "expected_post_registration_3": {"original_study": {"claim": {"hypothesis": "If individuals perceive newcomers as a threat to that community, then the \ngoverning institutions are likely to be called into question: those most worried about the effects of newcomers in the multicultural state may question the extent to which national political institutions will continue to represent a national citizenry.", "hypothesis_location": "Section: Concern about immigration and  \npolitical trust; p. 205", "statement": "...the effect of concern about immigration on political trust remains—even after taking into account this potential automatic correspondence via voting for the far right and via left-right self-placement, as well as pessimism, attitudes to the economy, and attitudes to government provision of health and educational services.", "statement_location": "Section:Multivariate Analyses ; p. 220", "study_type": "Observational"}, "data": {"source": "European Social Survey", "wave_or_subset": "1-4", "sample_size": "not stated", "unit_of_analysis": "respondent", "access_details": "Available at http://www.europeansocialsurvey.org/", "notes": "It is aggregated to the country level for the analysis."}, "method": {"description": "The authors use a multivariate analysis, which adjusts the standard errors for control variables at different levels of aggregation, to estimate the effect of immigration concerns on three measures of political distrust: politicians, parliament, and the legal system.", "steps": "(1) Reverse the coding of the political trust variable; (2) clean the sample by removing the excluded countries; (3) Run bivariate correlations between variables; (3) Run multilevel, multivariate analysis", "models": "Multivariate analysis", "outcome_variable": "political distrust (politicians, parliament, legal system)", "independent_variables": "concern about immigration", "control_variables": "far-right popularity;level of spending on social protection; long-term country of migration; governance quality; gdp per capita; unemployment rate", "tools_software": "not stated"}, "results": {"summary": "These results indicate that after controlling for fairly powerful predictors of distrust in politics, concern about immigration has a statistically significant effect on distrust in politics, with maximum effects of 1.7 on the 11-point measure of distrust in parliament, 1.3 on distrust in politicians, and 1.4 on distrust of the legal system.", "numerical_results": [{"outcome_name": "distrust in parliament", "value": "0.17", "unit": "not stated", "effect_size": "not stated", "confidence_interval": {"lower": "not stated", "upper": "not stated", "level": "not stated"}, "p_value": "0.00", "statistical_significance": "5% level", "direction": "Positive", "notes": "They also report standard errors = 0.000."}, {"outcome_name": "distrust in politicians", "value": "0.13", "unit": "not stated", "effect_size": "not stated", "confidence_interval": {"lower": "not stated", "upper": "not stated", "level": "not stated"}, "p_value": "0.00", "statistical_significance": "5% level", "direction": "Positive", "notes": "They also report standard errors = 0.000."}, {"outcome_name": "distrust in legal system", "value": "0.14", "unit": "not stated", "effect_size": "not stated", "confidence_interval": {"lower": "not stated", "upper": "not stated", "level": "not stated"}, "p_value": "0.00", "statistical_significance": "5% level", "direction": "Positive", "notes": "They also report standard errors = 0.000."}]}, "metadata": {"original_paper_id": "S0043887112000032.", "original_paper_title": "The Cultural Divide in Europe: Migration, Multiculturalism, and  Political Trust", "original_paper_code": "not stated", "original_paper_data": "http://www.europeansocialsurvey.org/; http://stats.oecd.org/index.aspx?DataSetCode=naG; http://titania.sourceoecd.org/vl =3262696/cl=11/nw=1/rpsv/factbook2009/06/02/01/index.htm"}}}}
{"study_id": "14", "original_paper_pdf": "14/input/original_paper.pdf", "initial_details": "[CLAIM]\noverall job satisfaction makes anemployee less likely to leave across the board: as job satisfaction increases, employees areless likely to intend to leave their agency for another within the federal government...[Leaving Agency, Job satisfaction = –0.444, SE = 0.0163, significant at p < .01, two tailtest]\n\n[HYPOTHESES]\nOverall job satisfaction makes an employee less likely to leave across the board", "replication_data_files": ["14/input/replication_data/DAR Pitts (126zz).R", "14/input/replication_data/Estimation Data - Pitts (126zz).csv"], "human_preregistration": "14/gt/human_preregistration.docx", "human_report": "14/gt/human_report.docx", "expected_post_registration": {"original_study": {"claim": {"hypothesis": "Overall job satisfaction makes an employee less likely to leave across the board.", "hypothesis_location": "The hypothesis is discussed in the Determinants of Employee Turnover: Workplace Satisfaction Factors section.", "statement": "overall job satisfaction makes anemployee less likely to leave across the board: as job satisfaction increases, employees areless likely to intend to leave their agency for another within the federal government...[Leaving Agency, Job satisfaction = –0.444, SE = 0.0163, significant at p < .01, two tailtest].", "statement_location": "Table 2", "study_type": "Observational"}, "data": {"source": "Federal Human Capital Survey (FHCS) administered by the U.S. Office of Personnel Management.", "wave_or_subset": "2006", "sample_size": "more than 200,000 U.S. federal government employees (full-time, permanent)", "unit_of_analysis": "individual - U.S. federal government employee", "access_details": "There is no mention how the data was accessed, only that it was collected and managed by the U.S. Office of Personnel Management. Also, no restriction is mentioned so maybe the data is open access.", "notes": "the sample size is not clearly stated (only: more than 200,000 U.S. federal government employees); the data was clustered by the agency; the DV - turnover intention is used as a surrogate for actual turnover. Data limitations prevented authors from predicting actual turnover, but turnover intention and actual turnover are usually highly and positively correlated in research. Gender data would have been an appropriate demographic variable to include but was missing for approximately 140,000 respondents so it was excluded from analysis."}, "method": {"description": "The authors probed associations between demographic, workplace satisfaction, and organizational/relational factors and employee turnover intention on survey data from U.S. federal employees.", "steps": "1. The authors obtained Federal Human Capital Survey data.\n2. They retained only respondents who intended to leave for another job and omitted those planning to retire. \n3. They measured turnover intention as a dichotomous variable (1  = those who plan to leave their agency to take another job within the federal government, 0 = all others) and labelled it Leaving Agency.\n4. They added independent variables.\n5. Finally, they estimated logistic regression models with robust standard errors clustered by agency.", "models": "logistic regression (with two tailed tests)", "outcome_variable": "turnover intention (dichotomous variable, where 1 represents those who plan to leave their agency to take another job within the federal government, and 0 represents all others)", "independent_variables": "Age, Agency tenure, Race/ethnicity, Job satisfaction, Satisfaction with pay, Satisfaction with benefits, Satisfaction with advancement, Performance culture, Empowerment, Relationship with supervisor, Relationship with coworkers; Interactions between age and satisfaction with benefits, and age and satisfaction with advancement were also included.", "control_variables": "not stated (all variables besides the DV are marked as independent)", "tools_software": "not stated"}, "results": {"summary": "Higher overall job satisfaction significantly reduces employees’ intention to leave their agency for another federal job (Job satisfaction = –0.444, SE = 0.0163, p < .01, two-tailed test).", "numerical_results": [{"outcome_name": "Leaving Agency", "value": "-0.444", "unit": "NA", "effect_size": "not stated", "confidence_interval": {"lower": "not stated", "upper": "not stated", "level": "not stated"}, "p_value": "< .01", "statistical_significance": "true", "direction": "negative"}]}, "metadata": {"original_paper_id": "not stated", "original_paper_title": "So Hard to Say Goodbye? Turnover Intention among U.S. Federal Employees", "original_paper_code": "not stated", "original_paper_data": "not stated"}}}, "expected_post_registration_2": {"original_study": {"claim": {"hypothesis": "Higher overall job satisfaction is associated with a lower likelihood that a federal employee intends to leave, including intentions to move to another job within the federal government.", "hypothesis_location": "Findings section under “Workplace Satisfaction Factors,” where the authors summarize that overall job satisfaction reduces turnover intention “across the board,” including intentions to leave one’s agency for another federal job.", "statement": "In the model predicting intention to leave one’s agency for another position within the federal government, overall job satisfaction is negatively and statistically significantly associated with leaving intention. The estimated coefficient for job satisfaction is -0.444 with a clustered robust standard error of 0.0163, significant at the 1% level (two-tailed test).", "statement_location": "Table 2 “Logit Results for Turnover Intention,” Leaving Agency column, row “Job satisfaction,” and the accompanying text in the Findings section describing job satisfaction as a consistent negative predictor.", "study_type": "Observational"}, "data": {"source": "2006 Federal Human Capital Survey (FHCS) administered by the U.S. Office of Personnel Management.", "wave_or_subset": "2006 FHCS.", "sample_size": "217504 observations.", "unit_of_analysis": "Individual federal employee.", "access_details": "not stated", "notes": "Turnover intention is measured using a survey question about whether the employee is considering leaving their organization within the next year and why. The focal outcome (“Leaving Agency”) is coded 1 if the employee intends to leave their agency for another job within the federal government and 0 otherwise. The job satisfaction measure is based on the item asking how satisfied the respondent is with their job overall on a five-point scale."}, "method": {"description": "The study estimates logistic regression models to examine how overall job satisfaction and other factors relate to two forms of turnover intention among U.S. federal employees, focusing here on intention to leave one’s agency for another job within the federal government.", "steps": ["Use 2006 FHCS survey data on U.S. federal employees.", "Construct the dependent variable “Leaving Agency” as a dichotomous indicator for intending to leave one’s agency for another federal job within the next year.", "Measure overall job satisfaction using the survey item asking overall satisfaction with one’s job.", "Include additional predictors spanning demographic factors, workplace satisfaction factors, and organizational/relational factors as specified in the model.", "Estimate a logistic regression model for “Leaving Agency” with robust standard errors clustered by agency."], "models": "Logistic regression with robust standard errors clustered by agency.", "outcome_variable": "Turnover intention: Leaving Agency (1 = intends to leave agency for another job within the federal government; 0 = otherwise).", "independent_variables": "Overall job satisfaction.", "control_variables": "Age category indicators; agency tenure; race/ethnicity; satisfaction with pay; satisfaction with benefits; satisfaction with advancement; performance culture; empowerment; relationship with supervisor; relationship with coworkers; and interaction terms between age categories and satisfaction with advancement and satisfaction with benefits.", "tools_software": "Stata 11"}, "results": {"summary": "Employees with higher overall job satisfaction are less likely to report an intention to leave their agency for another job within the federal government, and this association is statistically significant in the multivariate logit model.", "numerical_results": [{"outcome_name": "Leaving Agency", "value": "-0.444", "unit": "log-odds", "effect_size": "logit coefficient (unstandardized) for job satisfaction", "confidence_interval": {"lower": "not stated", "upper": "not stated", "level": "not stated"}, "p_value": "< 0.01", "statistical_significance": 1, "direction": "negative"}]}, "metadata": {"original_paper_id": "not stated", "original_paper_title": "So Hard to Say Goodbye? Turnover Intention among U.S. Federal Employees", "original_paper_code": "not stated", "original_paper_data": "not stated"}}}, "expected_post_registration_3": {"original_study": {"claim": {"hypothesis": "A large body of empirical research indicates a consistent and inverse relationship between overall job satisfaction and turnover.", "hypothesis_location": "Section; Determinants of Employee Turnover; Subsection: Workplace Satisfaction Factors; p. 752", "statement": "As expected, overall job satisfaction makes an employee less likely to leave across the board: as job satisfaction increases, employees are less likely to intend to leave their agency for another within the federal government and less likely to intend to leave the federal government for outside work.", "statement_location": "Section:Findings; Subsection: Workplace Satisfaction Factors; p. 757", "study_type": "Observational"}, "data": {"source": "Federal Human Capital Survey (FHCS) administered by the U.S. Office of Personnel Management", "wave_or_subset": "2006", "sample_size": "more than 200,000", "unit_of_analysis": "U.S. federal government employees", "access_details": "not stated", "notes": "The exact sample size is not listed in the text."}, "method": {"description": "Use logistic regressions to estimate the effect that job satisfaction has on turnover intention.", "steps": "(1) Create the two dependent, dichotomous variables of turnover intention - overall turnover intention and intention to leave for a position outside the government; (2) construct regressors by combining question responses by loading weights; (3) estimate the effects using logistic regressions with robust standard errors (4) use a series of Monte Carlo estimations to show how an employee’s probability of turnover intention is influenced by two different types of changes in our independent variables: i. moving from the minimum to the maximum possible value, and ii. increasing the value by a mean-centered standard deviation.", "models": "logistic regressions with robust standard errors", "outcome_variable": "turnover intention", "independent_variables": "overall job satisfaction", "control_variables": "demographic factors: satisfaction with benefits; advancement opportunity; empowerment; performance culture; coworker relationships; supervisor relationships; interactions of satisfaction with benefits and satisfaction with advancement", "tools_software": "Stata 11"}, "results": {"summary": "Overall job satisfaction makes an employee less likely to leave across the board", "numerical_results": [{"outcome_name": "Leaving Agency", "value": "0.444", "unit": "percentage point", "effect_size": "a standard deviation increase in job satisfaction is associated with a 5 percentage point decrease in the probability of intending to leave one’s agency", "confidence_interval": {"lower": "not stated", "upper": "not stated", "level": "not stated"}, "p_value": "<0.01", "statistical_significance": "1% level", "direction": "Negative", "notes": "Confidence intervals are not reported, but standard errors reported instead (0.0163)."}]}, "metadata": {"original_paper_id": "not stated", "original_paper_title": "So Hard to Say Goodbye? Turnover Intention among U.S. Federal Employees", "original_paper_code": "not stated", "original_paper_data": "not stated"}}}}
{"study_id": "15", "original_paper_pdf": "15/input/original_paper.pdf", "initial_details": "[CLAIM]\nFraud increases with violence up to a certain level, but then decreases again (coefficient on Violence (election, squared) term = -13.748, SE clustered at the regional command level = 4.720, p< 0.01)\n\n[HYPOTHESES]\nThe quadratic association between violence and election fraud will be negative", "replication_data_files": ["15/input/replication_data/Weidmann_Data_Analysis_Final.do", "15/input/replication_data/.DS_Store", "15/input/replication_data/Afghanistan_Election_Violence_2014.csv", "15/input/replication_data/Afghanistan_Election_Violence_2014.dta"], "human_preregistration": "15/gt/human_preregistration.pdf", "human_report": "15/gt/human_report.pdf", "expected_post_registration": {"original_study": {"claim": {"hypothesis": "The  quadratic  association  between  violence  and  election  fraud  will  be negative", "hypothesis_location": "p. 58 Violence and Fraud", "statement": "Fraud increases with violence up to a certain level, but then decreases again (coefficient  on  Violence  (election,  squared)  term  = -13.748,  SE  clustered  at  the  regional command level = 4.720, p< 0.01)", "statement_location": "Table 2, Model 1, Violence (election, squared)", "study_type": "Observational"}, "data": {"source": "fraud data: computed from polling-station level data made public by the International Election Commission (IEC) on 19 September 2009 (The IEC publicly posted the data in three waves. The authors use the earliest data release - returns from 27,163 distinct polling stations on 19 September).\nviolence data: International Security Assistance Force (ISAF). These data, which are commonly known as ‘significant activity’ or SIGACT reports; Worldwide Incident Tracking System (WITS); Armed Conflict Location and Event Dataset (ACLED).\n development data: the 2007 National Risk and Vulnerability Assessment (NRVA) household survey, which was jointly administered by the Ministry for Rural Rehabilitation and Development (MRRD) and the Central Statistics Office (CSO) of Afghanistan.\ngeography data: US Geological Survey. GTOPO30 Digital Elevation Model and GIS spatial layers used for geocoding and aggregation.\n population data: LandScan (raster GIS datasets) the complete dataset is available at http://dvn.iq.harvard.edu/dvn/dv/nilsw", "wave_or_subset": "election data: 2009, development data: 2007, geography data: 2007, population: 2008.", "sample_size": "375", "unit_of_analysis": "district", "access_details": "The IEC publicly posted the data. WITS is available at http://wits.nctc.gov/. ACLED details are to be found in: Clionadh Raleigh et al. ‘Introducing ACLED: An Armed Conflict Location and Event Dataset’, Journal of Peace Research, 47 (2010), 651–60. US Geological Survey: Available at http://edc.usgs.gov/products/elevation/gtopo30/gtopo30.html. LandScan: Oak Ridge National Laboratory, LandScan Global Population database. http://dvn.iq.harvard.edu/dvn/dv/nilsw seems to be open access", "notes": "Unlike othe work on election fraud, the authors assume (and show) that insurgent violence is not biased in favour or against a particular candidate.This assumption characterizes violence around the Afghanistan election well, but it limits the applicability of the findings to other cases where this may not hold. Also, the authors limited their view of election fraud to manipulation tactics local to the polling centre. Moreover, lacking information about the precise location of polling stations, they were unable to estimate fraud at lower levels, for example, cities. Many of the covariates used in the regression analysis are available only at the district level. All violence indicators are reported as incidents per 1,000 population. The authors created two indicators for development: first, the proportion of households supplied with electricity, and, secondly, the per capita expenditure (in 1,000 s of afghanis)."}, "method": {"description": "The authors test the prediction that the relationship between violence and election fraud follows an inverted U-shape. Using district-level data from Afghanistan’s 2009 presidential election, the authors measure fraud through forensic last-digit tests and validate it with recount-based evidence.", "steps": "1. Collect all the data. \n2. Apply the last-digit test to the total vote count and code a binary dependent variable for fraud (which takes the value of 1 if this test is significant at the 5 per cent level for a particular district).\n3. Take violence data from ISAF SIGACT reports and calculate incidents per 1,000 population within five-day (20–24 August) and sixty-day windows around election day.\n4. Georeference ISAF SIGACT, WITS, and ACLED incidents to districts.\n5. Conduct robustness checks for the violence measure using alternative geo-referenced datasets (WITS and ACLED).\n6. Compute great-circle distances from district centres to Kabul.\n7. Derive average district elevation from GTOPO30 raster data.\n8. Aggregate population data from LandScan grid cells.\n9. Construct development indicators: proportion of households with electricity, and per capita expenditure.\n10. Add the number of closed polling stations at the district level as a control.\n11. Merge datasets.\n12. Estimate OLS models clustering standard errors at the regional command level.", "models": "logit regression", "outcome_variable": "Fraud, last-digit test (total count)", "independent_variables": "Violence (election), Violence (election, squared), Violence (2 months pre-election), Violence (2 months pre-election, squared), Electrification, Per-capita expenditure, Distance from Kabul, Elevation", "control_variables": "number of closed stations at the district level", "tools_software": "not stated"}, "results": {"summary": "Fraud initially rises with violence but then falls at higher levels, following an inverted U-shaped pattern. The coefficient on the Violence (election, squared) term is -13.748 (SE clustered at the regional command level = 4.720, p < 0.01).", "numerical_results": [{"outcome_name": "Election Fraud", "value": "-13.748", "unit": "NA", "effect_size": "not stated", "confidence_interval": {"lower": "not stated", "upper": "not stated", "level": "not stated"}, "p_value": "<0.01", "statistical_significance": "true", "direction": "negative"}]}, "metadata": {"original_paper_id": "10.1017/S0007123412000191", "original_paper_title": "Violence and Election Fraud: Evidence from Afghanistan.", "original_paper_code": "not stated", "original_paper_data": "http://dvn.iq.harvard.edu/dvn/dv/nilsw"}}}, "expected_post_registration_2": {"original_study": {"claim": {"hypothesis": "Electoral fraud initially increases as violence rises but declines once violence becomes sufficiently intense, producing a curvilinear relationship between violence and fraud.", "hypothesis_location": "Theory section under “Violence and Fraud,” where Hypothesis 1 is explicitly stated predicting an inverted U-shaped relationship between violence and election fraud.", "statement": "The empirical results show that fraud is more likely in districts experiencing moderate levels of violence but becomes less likely in districts with very high levels of violence, consistent with an inverted U-shaped relationship between violence and fraud.", "statement_location": "Results section, Table 2, where the linear violence term is positive and significant and the squared violence term is negative and significant.", "study_type": "Observational"}, "data": {"source": "Polling-station level election returns from the Afghan Independent Election Commission combined with geocoded violence data from ISAF SIGACT reports and auxiliary datasets.", "wave_or_subset": "2009 Afghanistan presidential election; violence measured in a five-day window around election day and, in alternative models, a two-month pre-election window.", "sample_size": "Approximately 389 districts for fraud measures; up to 398 districts for violence and control variables.", "unit_of_analysis": "District.", "access_details": "Election results were publicly released by the Afghan Independent Election Commission; violence data originate from military and media-based event datasets.", "notes": "Fraud is measured using a digit-based forensic test (last-digit test) aggregated to the district level and validated using data from a post-election recount of ballot boxes."}, "method": {"description": "The study tests whether the relationship between violence and election fraud is nonlinear by regressing district-level fraud indicators on measures of violence and their squared terms, while controlling for development and geographic factors.", "steps": ["Aggregate polling-station election results to the district level.", "Apply the last-digit forensic test to generate a binary indicator of fraud for each district.", "Measure insurgent violence using counts of attacks per 1,000 population around election day.", "Include a squared violence term to capture potential nonlinearity.", "Add controls for closed polling stations, development, geography, and population characteristics.", "Estimate logit models for the binary fraud measure and OLS models for the recount-based fraud share, clustering standard errors at the regional command level."], "models": "Logistic regression (for binary fraud indicator) and OLS regression (for recount-based fraud share) with linear and squared violence terms.", "outcome_variable": "Election fraud at the district level.", "independent_variables": "Violence per 1,000 population; squared violence term.", "control_variables": "Number of planned polling stations closed; electrification; per-capita expenditure; distance from Kabul; average elevation.", "tools_software": "not stated"}, "results": {"summary": "The regression results support a curvilinear association between violence and fraud: fraud increases with low to moderate violence but decreases as violence becomes very high.", "numerical_results": [{"outcome_name": "Election fraud (last-digit test)", "value": "8.477", "unit": "log-odds (logit coefficient)", "effect_size": "logit regression coefficient for violence (linear term)", "confidence_interval": {"lower": "not stated", "upper": "not stated", "level": "not stated"}, "p_value": "< 0.1", "statistical_significance": 1, "direction": "positive"}, {"outcome_name": "Election fraud (last-digit test)", "value": "-13.748", "unit": "log-odds (logit coefficient)", "effect_size": "logit regression coefficient for squared violence term", "confidence_interval": {"lower": "not stated", "upper": "not stated", "level": "not stated"}, "p_value": "< 0.01", "statistical_significance": 1, "direction": "negative"}]}, "metadata": {"original_paper_id": "10.1017/S0007123412000191", "original_paper_title": "Violence and Election Fraud: Evidence from Afghanistan", "original_paper_code": "not stated", "original_paper_data": "http://dvn.iq.harvard.edu/dvn/dv/nilsw"}}}, "expected_post_registration_3": {"original_study": {"claim": {"hypothesis": "The relationship between (in)security and election fraud should be in the form of an inverted U-shape. Fraud increases with violence up to a certain level, but then decreases again.", "hypothesis_location": "Section: MANIPULATION TACTICS; Subsection: Violence and Fraud; p. 58", "statement": "We find support…for the inverted U-shaped relationship between violence and fraud", "statement_location": "Section: CONCLUSION AND POLICY IMPLICATIONS; p. 74", "study_type": "Observational"}, "data": [{"source": "International Election Commission", "wave_or_subset": "1-3", "sample_size": "27,163", "unit_of_analysis": "polling-station level", "access_details": "not stated", "notes": "After removing missing, sample size drops to 22,858."}, {"source": "SIGACT reports", "wave_or_subset": "not stated", "sample_size": "not stated", "unit_of_analysis": "violent incident", "access_details": "not stated", "notes": "The authors discuss that they use this data, but do not describe the characteristics of the data (e.g. sample size)."}, {"source": "National Risk and Vulnerability Assessment", "wave_or_subset": "2007", "sample_size": "20,576", "unit_of_analysis": "household", "access_details": "not stated", "notes": ""}, {"source": "GTOPO30", "wave_or_subset": "NA", "sample_size": "NA", "unit_of_analysis": "elevation", "access_details": "Available at http://edc.usgs.gov/products/elevation/gtopo30/gtopo30.html. 2007.", "notes": "This is geological raster data for Afghanistan so there is not real sample size."}, {"source": "LandScan population data", "wave_or_subset": "not stated", "sample_size": "not stated", "unit_of_analysis": "30 x 30 latitude/longitude cells", "access_details": "not stated", "notes": "They do not directly discuss access, but they leave the reference to the data: Oak Ridge National Laboratory, LandScan Global Population database, 2008. For this paper, the data is aggregated to the district level."}]}, "method": {"description": "Use logit (for the last-digit fraud measure) and OLS models (for the recount-based fraud measure) and regress fraud on violence and its squared term to test the inverted U-shaped prediction.", "steps": "(1) Clean, merge, and aggregate the data (remove missing, etc.) when necessary; (2) Group polling stations by district and apply the Beber–Scacco last-digit test to the total vote count; (3) Code a binary dependent variable for fraud; (4) Estimate the effects using logit and OLS models.", "models": "logit and OLS", "outcome_variable": "Fraud", "independent_variables": "Violence; Violence squared", "control_variables": "number of closed centres, economic development and geographic accessibility", "tools_software": "not stated"}, "results": {"summary": "Since the linear term of the violence measure receives a positive and significant coefficient, and the squared term a negative one, this implies the U-shaped prediction is credible.", "numerical_results": [{"outcome_name": "Election Fraud (last-digit)", "value": "13.748", "unit": "not stated", "effect_size": "not stated", "confidence_interval": {"lower": "not stated", "upper": "not stated", "level": "not stated"}, "p_value": "<0.001", "statistical_significance": "0.1% level", "direction": "Negative", "notes": "Confidence intervals are not presented, but standard errors are presented instead (4.720)."}, {"outcome_name": "Election Fraud (Recount)", "value": "1.438", "unit": "not stated", "effect_size": "not stated", "confidence_interval": {"lower": "not stated", "upper": "not stated", "level": "not stated"}, "p_value": "<0.001", "statistical_significance": "0.1% level", "direction": "Negative", "notes": "Confidence intervals are not presented, but standard errors are presented instead (0.375)."}]}, "metadata": {"original_paper_id": "10.1017/S0007123412000191", "original_paper_title": "Violence and Election Fraud: Evidence from Afghanistan", "original_paper_code": "http://dvn.iq.harvard.edu/dvn/dv/nilsw", "original_paper_data": "http://dvn.iq.harvard.edu/dvn/dv/nilsw"}}}
{"study_id": "16", "original_paper_pdf": "16/input/original_paper.pdf", "initial_details": "[CLAIM]\nSocial distancing measures decreased the mobility by anadditional 23% (95% CI: 20%, 27%)\n\n[HYPOTHESES]\nThe introduction of social distancing measures is associated with a decrease in mobility.", "replication_data_files": ["16/input/replication_data/mycode_for.replication.dataset.do", "16/input/replication_data/replicationDataset_Malik2020_with.year.csv"], "human_preregistration": "16/gt/human_preregistration.pdf", "human_report": "16/gt/human_report.pdf", "expected_post_registration": {"original_study": {"claim": {"hypothesis": "The introduction of social distancing measures is associated with a decrease in mobility.", "hypothesis_location": "The relation is mentioned in the first paragraph and explicitly stated in third (the paper has only four paragraphs) where the results are reported.", "statement": "Social distancing measures decreased the mobility by anadditional 23% (95% CI: 20%, 27%).", "statement_location": "Third paragraph with results.", "study_type": "Observational"}, "data": {"source": "Citymapper Mobility Index (CMI) for mobility data; it is mentioned that the data on implementation of governmental social distancing measures was acquired from official government and media websites but specific sources are not mentioned.", "wave_or_subset": "March 2, 2020 to March 26, 2020", "sample_size": "1,025 observations across 41 cities (25 observations per city)", "unit_of_analysis": "mobility (per city)", "access_details": "presumably that data from Citymapper is publicly available as well as the data ", "from official government and media websites": "as no restrictions or permissions are mentioned and the authors explain that they downloaded the data.", "notes": "It is explained that the data on governmental social distancing measures were obtained from official government and media websites, but the specific variables or data fields collected were not described."}, "method": {"description": "The study probed the effect of governmental social distancing measures on reducing mobility in 41 cities.", "steps": "1. The authors downloaded mobility data from the Citymapper Mobility Index and unspecified data on the implementation of governmental social distancing measures from unspecified official government and media websites. \n2. Then, they classified each city as having instituted social distancing measures if non-essential businesses were closed\n3. Further classified measures as moderate or intense based on the intensity of closure.\n4. Finally, they estimated the effect of time and social distancing measures using a multilevel mixed-effects linear regression model.", "models": "multilevel mixed-effects linear regression model", "outcome_variable": "mobility", "independent_variables": "time (in days), social distancing measures presence, intensity of social distancing measures (moderate, intense).", "control_variables": "not stated; there are probably none, judgung by the results section but the authors obtained some data on governmental social distancing measures and the specific variables or data fields collected were not described.", "tools_software": "not stated"}, "results": {"summary": "The mobility in general decreased in the discussed period. Social distancing measures decreased the mobility by an additional 23% (95% CI: 20%, 27%).", "numerical_results": [{"outcome_name": "mobility", "value": "23", "unit": "%", "effect_size": "not stated", "confidence_interval": {"lower": "20%", "upper": "27%", "level": "95%"}, "p_value": "not stated", "statistical_significance": "not stated", "direction": "negative"}]}, "metadata": {"original_paper_id": "https://doi.org/10.1101/2020.03.30.20048090", "original_paper_title": "COVID-19 related social distancing measures and reduction in city mobility.", "original_paper_code": "not stated", "original_paper_data": "not stated"}}}, "expected_post_registration_2": {"original_study": {"claim": {"hypothesis": "The implementation of government social distancing measures is associated with a reduction in urban mobility.", "hypothesis_location": "The expectation is introduced in the background and study aim where the authors state that they analyze mobility data to assess the effect of government social distancing measures on movement in cities.", "statement": "The analysis shows that government social distancing measures are associated with a substantial additional reduction in city mobility, beyond the general downward time trend in mobility.", "statement_location": "Results para describing the multilevel mixed-effects regression results, where the authors report the estimated effect of social distancing measures on the Citymapper Mobility Index.", "study_type": "Observational"}, "data": {"source": "Citymapper Mobility Index (CMI) data combined with information on government social distancing measures compiled from official government and media sources.", "wave_or_subset": "Daily city-level observations from March 2, 2020 to March 26, 2020.", "sample_size": "1025 observations across 41 cities.", "unit_of_analysis": "City-day.", "access_details": "Citymapper Mobility Index data are publicly available via Citymapper; access procedures are not further detailed.", "notes": "The CMI is based on planned trips using the Citymapper application and reflects mobility related to public transport, walking, biking, and ride-sharing, but not private automobile use. Cities are classified as having implemented social distancing measures if non-essential businesses were closed, with measures further categorized as moderate or intense."}, "method": {"description": "The study estimates the association between social distancing policies and changes in urban mobility using longitudinal city-level mobility data and regression modeling.", "steps": ["Download daily Citymapper Mobility Index data for 41 cities from March 2 to March 26, 2020.", "Compile dates and intensity of government social distancing measures for each city using official and media sources.", "Classify social distancing measures as present or absent, and further as moderate or intense.", "Construct a longitudinal dataset with repeated daily observations for each city.", "Estimate a multilevel mixed-effects linear regression model including time trends and social distancing indicators."], "models": "Multilevel mixed-effects linear regression model with city-level repeated observations.", "outcome_variable": "City mobility measured by the Citymapper Mobility Index (percentage relative to baseline).", "independent_variables": "Indicator for implementation of social distancing measures; time (day).", "control_variables": "Time trend (days since March 2, 2020).", "tools_software": "not stated"}, "results": {"summary": "City mobility declined steadily over time and decreased substantially more in cities after the introduction of social distancing measures.", "numerical_results": [{"outcome_name": "City mobility (Citymapper Mobility Index)", "value": "-23", "unit": "percentage-point change in mobility", "effect_size": "regression coefficient for social distancing measures", "confidence_interval": {"lower": "-27", "upper": "-20", "level": "95"}, "p_value": "not stated", "statistical_significance": 1, "direction": "negative"}]}, "metadata": {"original_paper_id": "10.1101/2020.03.30.20048090", "original_paper_title": "COVID-19 related social distancing measures and reduction in city mobility", "original_paper_code": "not stated", "original_paper_data": "not stated"}}}, "expected_post_registration_3": {"original_study": {"claim": {"hypothesis": "One approach to slowing the pandemic is reducing the contact rate in the population through social distancing. Governments the world over have instituted different measures to \nincrease social distancing but information on their effectiveness in reducing mobility is lacking.", "hypothesis_location": "p. 1", "statement": "Social distancing measures decreased the mobility by an additional 23%.", "statement_location": "p. 1", "study_type": "Observational"}, "data": {"source": "Citymapper Mobility Index", "wave_or_subset": "March 2, 2020 through March 26, 2020", "sample_size": "1,025", "unit_of_analysis": "city-by-day level", "access_details": "https://citymapper.com/cmi", "notes": "Data is for 41 cities. Social distancing measures are also used (as treatment) and collected from various government and media websites."}, "method": {"description": "Use linear regression models to estimate the effect of social distancing measures on mobility.", "steps": "(1) Download the data; (2) Tabulate data on implementation of governmental social distancing measures; (3) Estimate the effect of time and social distancing measures using a multilevel mixed-effects linear regression model.", "models": "multilevel mixed-effects linear regression model", "outcome_variable": "mobility", "independent_variables": "social distancing measures", "control_variables": "time", "tools_software": "not stated"}, "results": {"summary": "Social distancing measures decreased the mobility by an additional 23%.", "numerical_results": [{"outcome_name": "mobility", "value": "23", "unit": "percent", "effect_size": "not stated", "confidence_interval": {"lower": "20", "upper": "27", "level": "95%"}, "p_value": "not stated", "statistical_significance": "not stated", "direction": "Positive"}]}, "metadata": {"original_paper_id": "10.1101/2020.03.30.20048090", "original_paper_title": "COVID-19 related social distancing measures and reduction in city mobility", "original_paper_code": "not stated", "original_paper_data": "not stated"}}}}
{"study_id": "17", "original_paper_pdf": "17/input/original_paper.pdf", "initial_details": "[CLAIM]\nPolitical polarization is strongly associatedwith smaller government in democratic countries, but there is no relationship between polarization and the size ofgovernment in undemocratic countries. When the sample is restricted to strong democracies, the estimated effect of polarization on government consumption is statistically significant (coefficient in the ‘long’ specification for the “Private” measure of polarization=-18.73, heteroscedasticity robust SE=4.79, p=.01)\n\n[HYPOTHESIS]\nAmong strong democracies (countries with a Polity IV score of 9 or greater), polarization (as measured by the SD of the “private ownership” responses) will be negatively associated with the size of government", "replication_data_files": ["17/input/replication_data/.DS_Store", "17/input/replication_data/analyze.R", "17/input/replication_data/data.csv"], "human_preregistration": "17/gt/human_preregistration.pdf", "human_report": "17/gt/human_report.pdf", "expected_post_registration": {"original_study": {"claim": {"hypothesis": "Among strong democracies (countries with a Polity IV score of 9 or greater), polarization (as measured by the SD of the “private ownership” responses) will be negatively associated with the size of government.", "hypothesis_location": "discussed in the introduction", "statement": "Political polarization is strongly associatedwith smaller government in democratic countries, but there is no relationship between polarization and the size ofgovernment in undemocratic countries. When the sample is restricted to strong democracies, the estimated effect of polarization on government consumption is statistically significant (coefficient in the ‘long’ specification for the “Private” measure of polarization=-18.73, heteroscedasticity robust SE=4.79, p=.01).", "statement_location": "Table 5 column PRIVATE.", "study_type": "Observational"}, "data": {"source": "polarization: World Values Surveys (WVS); government size: World Bank and International Monetary Fund; GDP, export/import, population: World Bank", "wave_or_subset": "polarization: primarily the 2000 wave of the WVS, but for some countries the 1995 wave is used; the appendix A shows survey years vary across countries.\ngovernment size: 2003-2005.\nGDP, export/import, population: 2009.\nstrong democracies: Polity IV project.", "sample_size": "24", "unit_of_analysis": "country-year", "access_details": "polarization: European Values Study Group andWorld Values Association. 2006. European and World Values Surveys Four-wave Integrated Data File 1981−2004. Version 20060423. http://www.wvsevsdb.com/wvs/WVSData.jsp (accessed July 17, 2010).\ngovernment size: Gwartney, J., and R. Lawson. 2008. Economic Freedom of the World: 2008 Annual Report. Vancouver, Canada: Fraser Institute. http://www.freetheworld.com/release_2008.html (accessed July 17, 2010).\nGDP, export/import, population: World Bank. 2009.World Development Indicators 2009.Washington, DC: World Bank. (no further access details)\nstrong democracies: Marshall, M.G., and K. Jaggers. 2007. Political Regime Characteristics and Transitions 1800−2007. Polity IV Project. University of Maryland. http://www.systemicpeace.org/inscr/inscr.htm (accessed May 4, 2010).Polity IV project", "notes": "The WVS is based on face-to-face interviews with about 1,000 respondents in 83 different countries, but 9 of these are not included in the analysis (Northern Ireland, Puerto Rico, Taiwan, Serbia and Montenegro, Iraq, Belarus, Saudi Arabia, Israel and India). Broad measure of government size: general government consumption as a fraction of total consumption (GOVCONS), the average of the years 2003 to 2005. The control variable ASIAE is included in Table 5 but eventually omitted in strong democracy sample. Regional dummies, FEDERAL and OECD variables should be coded as per: Persson, T., and G. Tabellini. 2003. The Economic Effects of Constitutions. Cambridge, MA: MIT Press."}, "method": {"description": "The article studies the relationship between political polarization and public spending using the dispersion of self-reported political preferences as a measure of polarization.", "steps": "1.Get the government size data (government consumption as a fraction of total consumption, averaged over 2003–2005).\n2. Get the political attitudes data for the 2000 wave. For countries missing in 2000, supplement with the 1995 wave.\n3.Construct the polarization measure by extracting responses to the “private ownership of business” question from the WVS. Compute the SD of responses within each country.\n4.Identify strong democracies by selecting only countries with Polity score ≥ 9.\n5.Add control variables (long specification):\n-Mean response to the PRIVATE question.\n-AFRICA = 1 if in Africa.\n-ASIAE = 1 if in East Asia.\n-LAAM = 1 if in Latin America, Central America, or the Caribbean.\n-COL_ESPA = Spanish colonial origin, weighted by years since independence.\n-COL_UKA = British colonial origin, weighted by years since independence.\n-COL_OTHA = Other colonial origin, weighted similarly.\n-LYP = log of real GDP per capita (constant 2000 USD, year 2000).\n-TRADE = (exports + imports) / GDP (year 2000).\n-PROP1564 = share of population aged 15–64 (year 2000).\n-PROP65 = share of population aged 65+ (year 2000).\n-FEDERAL = 1 if the country has a federal structure.\n-OECD = 1 if OECD member before 1993 (excluding Turkey).\n6.Run an OLS of GOVCONS on PRIVATE polarization, including all control variables. Use heteroscedasticity-robust standard errors.", "models": "ordinary least squares regression with heteroscedasticity-robust standard errors", "outcome_variable": "Government consumption", "independent_variables": "Polarization (SD of PRIVATE responses)", "control_variables": "Mean of PRIVATE responses, AFRICA, ASIAE, LAAM, COL_ESPA, COL_UKA, COL_OTHA, LYP, TRADE, PROP1564, PROP65, FEDERAL, OECD", "tools_software": "not stated"}, "results": {"summary": "Higher polarization is associated with smaller government consumption (b = −18.73, SE = 4.79, p = .01).", "numerical_results": [{"outcome_name": "GOVCONS", "value": "-18.73", "unit": "NA", "effect_size": "not stated", "confidence_interval": {"lower": "not stated", "upper": "not stated", "level": "not stated"}, "p_value": "0.01", "statistical_significance": "true", "direction": "negative"}]}, "metadata": {"original_paper_id": "10.1017/S0003055410000262", "original_paper_title": "Political Polarization and the Size of Government.", "original_paper_code": "not stated", "original_paper_data": "not stated"}}}, "expected_post_registration_2": {"original_study": {"claim": {"hypothesis": "Within countries classified as strong democracies (Polity IV score at least 9), greater economic-policy polarization—measured as the standard deviation of responses to the survey item on private versus government ownership—will be associated with a smaller government sector.", "hypothesis_location": "The conditional expectation that polarization matters in democracies is developed in the theory discussion and then operationalized in the empirical section where the authors estimate models for the full sample, undemocratic countries, and strong democracies, using the “Private ownership” polarization measure.", "statement": "Restricting the analysis to strong democracies, the association between the private-ownership polarization measure and government consumption is negative and statistically significant.", "statement_location": "Results table 4 for the strong-democracy subsample (Polity IV ≥ 9), long specification, using the “Private” polarization measure as the key explanatory variable for government consumption (coefficient = -18.73, robust SE = 4.79, p = .01).", "study_type": "Observational"}, "data": {"source": "World Values Survey (WVS) for individual responses used to construct polarization; Polity IV for regime type (strong democracy defined as Polity IV ≥ 9); country-level government consumption data (government size) and additional country covariates from standard international macro datasets.", "wave_or_subset": "Focus on the 2000 wave for most countries, but data from the 1995 wave are used for some countries so as to increase the sample size.", "sample_size": "Face-to-face interviews with about 1,000 respondents in 83 different countries.", "unit_of_analysis": "Country (country-level analysis).", "access_details": "not stated", "notes": "Polarization is operationalized as dispersion (standard deviation) in responses to the WVS ‘private versus government ownership’ item. Strong democracies are defined using Polity IV scores of 9 or greater. The focal outcome is government consumption (government size)."}, "method": {"description": "The study estimates cross-country regression models of government size on polarization, with regime-based sample restrictions. The focal test estimates the association between the ‘private ownership’ polarization measure and government consumption within strong democracies (Polity IV ≥ 9) using a long specification and heteroscedasticity-robust standard errors.", "steps": ["Construct a country-level polarization measure as the standard deviation of survey responses to the private-versus-government ownership item.", "Classify countries by regime type using Polity IV and restrict the analytic sample to strong democracies (Polity IV ≥ 9).", "Measure government size using government consumption.", "Estimate the long-specification regression of government consumption on the polarization measure for the strong-democracy subsample.", "Compute heteroscedasticity-robust standard errors and assess statistical significance."], "models": "Cross-country linear regression model (long specification) with heteroscedasticity-robust standard errors.", "outcome_variable": "Government consumption (government size).", "independent_variables": "Polarization measured as the standard deviation of the ‘private ownership’ responses.", "control_variables": "Geographic regional dummy variables: Africa (AFRICA), South and East Asia (ASIAE), and Latin and South America and the Caribbean (LAAM); colonial origin variables weighted by years of independence: British colonial origin (COL_UKA), Spanish colonial origin (COL_ESPA), and other colonial origin (COL_OTHA). Additional controls that may be endogenous: log GDP per capita in 2000 (LYP), trade openness in 2000 (TRADE), proportion of population ages 15–64 in 2000 (PROP1564), proportion of population above 65 in 2000 (PROP65), federal political structure indicator (FEDERAL), and OECD membership before 1993 (OECD; Turkey excluded).", "tools_software": "not stated"}, "results": {"summary": "In strong democracies, higher polarization on the private-ownership dimension is associated with significantly lower government consumption, whereas the relationship is not present in undemocratic countries.", "numerical_results": [{"outcome_name": "Government consumption", "value": "-18.73", "unit": "units of government consumption", "effect_size": "linear regression coefficient", "confidence_interval": {"lower": "not stated", "upper": "not stated", "level": "not stated"}, "p_value": "0.01", "statistical_significance": 1, "direction": "negative"}]}, "metadata": {"original_paper_id": "10.1017/S0003055410000262", "original_paper_title": "Political Polarization and the Size of Government", "original_paper_code": "not stated", "original_paper_data": "not stated"}}}, "expected_post_registration_3": {"original_study": {"claim": {"hypothesis": "...more recent research in economics and political science has suggested that the size of government is also determined by the dispersion or polarization of political preferences.", "hypothesis_location": "Section: Introduction; p. 543", "statement": "We find that the relationship between polarization and size of government is substantially stronger in democratic countries, supporting the view that polarization affects public spending rather than the other way around.", "statement_location": "Section: Introduction; p. 544", "study_type": "Observational"}, "data": {"source": "World Values Surveys", "wave_or_subset": "2000 and 1995", "sample_size": "1,000", "unit_of_analysis": "individual", "access_details": "not stated", "notes": "Primarily the 2000 wave, but use the 1995 wave for some countries and analysis. The access is not stated, but there is a reference to the dataset: http://www.wvsevsdb.com/wvs/WVSData.jsp."}, "method": {"description": "Authors use standard regression analysis to test the effect of polarization on government size", "steps": "(1) Clean the data (e.g. remove unused countries); (2) construct polarization measure using standard errors and also using the Esteban and Ray method; (3) construct the measure of government size as the general government consumption as a fraction of total consumption; (4) ", "models": "regression analysis", "outcome_variable": "government size", "independent_variables": "polarization", "control_variables": "geographic dummy variables; colonial origin; logarithm of GDP per capita in 2000; openness to trade in 2000; proportion of population between 15 and 64 in 2000; proportion of population above 65 in 2000; dummy variable indicating whether the country has a federal political structure; indicator variable for Organisation for Economic Cooperation and Development membership before 1993", "tools_software": "not stated"}, "results": {"summary": "Political polarization has a negative and statistically significant relationship with government consumption in the specifications with controls for the mean response and exogenous set of control variables…When the sample is restricted to strong democracies, the estimated effect of polarization on government consumption is statistically significant and robust to the different sets of control variables.", "numerical_results": [{"outcome_name": "government consumption", "value": "18.73", "unit": "standard deviation", "effect_size": "between 2.0 and 6.1 percentage points", "confidence_interval": {"lower": "not stated", "upper": "not stated", "level": "not stated"}, "p_value": "0.01", "statistical_significance": "1% level", "direction": "Negative", "notes": "The confidence intervals are not presented, but the standard error is (4.79)."}]}, "metadata": {"original_paper_id": "10.1017/S0003055410000262", "original_paper_title": "Political Polarization and the Size of Government", "original_paper_code": "not stated", "original_paper_data": "not stated"}}}}
{"study_id": "18", "original_paper_pdf": "18/input/original_paper.pdf", "initial_details": "[CLAIM]\nA  one standard deviation decrease in median age (equal to 3.5 years in 2010) results in a 2.5 percentage point increase in the entrepreneurship rate, which is over 40 percent of the mean entrepreneurship rate across countries (equal to 0.061 in 2010).\n\n[HYPOTHESIS]\nThe entrepreneurship rate in a country is negatively associated with the country’s median age.", "replication_data_files": ["18/input/replication_data/REPEntireDataset2_Country_Year_Entre_Regression.do", "18/input/replication_data/replication_data_mkk9.csv"], "human_preregistration": "18/gt/human_preregistration.pdf", "human_report": "18/gt/human_report.pdf", "expected_post_registration": {"original_study": {"claim": {"hypothesis": "The entrepreneurship rate in a country is negatively associated with the country’s median age.", "hypothesis_location": "p. 154 Corollary 3; p.157 empirical implications: implication 5; also discussed in the abstract.", "statement": "A  one standard deviation decrease in median age (equal to 3.5 years in 2010) results in a 2.5 percentage point increase in the entrepreneurship rate, which is over 40 percent of the mean entrepreneurship rate across countries (equal to 0.061 in 2010).", "statement_location": "p. S167, section V. Empirical Implementation (B. Country-Level Analysis); also discussed in the abstract.", "study_type": "Observational"}, "data": {"source": "entrepreneurship: Global Entrepreneurship Monitor; the GEM data is checked against: Flash Eurobarometer Survey on Entrepreneurship.\ndemographics: US Census Bureau’s International Data Base", "wave_or_subset": "2001-2010", "sample_size": "393", "unit_of_analysis": "country-year", "access_details": "GEM: Bosma, Niels, Alicia Coduras, Yana Litovsky, and Jeff Seaman. 2012. “Global Entrepreneurship Monitor Manual.” Version 2012-9 (May). Global Entrepreneurship Res. Assoc., London. no further access details are provided; \nFESE: https://dbk.gesis.org\nIDB: http://www.census.gov/population/international/data/idb/index.php", "notes": "The total number of observations is 1.3 million, the number of country-year cells is 393, and the number of country-age-year cells is 17,554. The entrepreneurship rate can be constructed at each cell level. China was excluded from the study."}, "method": {"description": "The study examines how demographic structure influences entrepreneurship, regressing country-level entrepreneurship rates on median age", "steps": "1.Get entrepreneurship data defined as managing and owning a business up to 42 months old that pays wages. Aggregate the data to construct entrepreneurship rates at the country-year level.\n2.Get population data and calculate median age for each country-year.\n3.Estimate a regression and weight observations are weighted by the number of individuals who make up each country-year cell. Include year dummies as controls. Standard errors clustered at the country level are in brackets.", "models": "ordinary least squares regression", "outcome_variable": "Entrepreneurship Rate (Early Stage, Pays Wages)", "independent_variables": "Median age (ages 20–64)", "control_variables": "year dummies", "tools_software": "not stated"}, "results": {"summary": "A one–standard deviation (3.5-year) reduction in median age raises the entrepreneurship rate by 2.5 percentage points, equivalent to roughly 40% of the sample mean (0.061) across countries.", "numerical_results": [{"outcome_name": "Entrepreneurship Rate (Early Stage, Pays Wages)", "value": "2.5", "unit": "percentage point", "effect_size": "not stated", "confidence_interval": {"lower": "not stated", "upper": "not stated", "level": "not stated"}, "p_value": "not stated; p was not stated explicitly, but the result is calculated using the estimates from column 3 of table 2, and regression coefficients on median age there are significant at 1 percent level", "statistical_significance": "not stated; p was not stated explicitly, but the result is calculated using the estimates from column 3 of table 2, and regression coefficients on median age there are significant at 1 percent level", "direction": "negative"}]}, "metadata": {"original_paper_id": "0022-3808/2018/126S1-0007", "original_paper_title": "Demographics and Entrepreneurship.", "original_paper_code": "not stated", "original_paper_data": "not stated"}}}, "expected_post_registration_2": {"original_study": {"claim": {"hypothesis": "Countries with younger populations tend to exhibit higher rates of entrepreneurial activity, implying an inverse relationship between median age and entrepreneurship.", "hypothesis_location": "The Effect of Age on Entrepreneurship & Country-Level Aggregation and Summary of Empirical Predictions sections, where the authors argue that population age structure shapes aggregate entrepreneurship.", "statement": "The empirical estimates indicate that countries with lower median ages have higher entrepreneurship rates. A decline in median age is associated with a sizable increase in entrepreneurship.", "statement_location": "Results (Table 2) section discussing the baseline cross-country regressions and the standardized effect of median age on entrepreneurship rates.", "study_type": "Observational"}, "": {"data": {"source": "Global Entrepreneurship Monitor (GEM) individual-level survey data collected by the Global Entrepreneurship Research Association (GERA); Flash Eurobarometer Survey on Entrepreneurship (FESE) for robustness checks; population statistics from the US Census Bureau’s International Data Base (IDB); macroeconomic and institutional controls from the Penn World Table (version 7.1), World Bank databases, Barro and Lee (2010) education attainment data, and the Property Rights Alliance.", "wave_or_subset": "GEM surveys conducted in multiple waves between 2001 and 2010 across 82 countries; FESE surveys conducted in overlapping years starting in 2002; demographic data matched annually to country-year observations.", "sample_size": "More than 1.3 million individuals aged 15–60 from the GEM; 82 countries; 393 country-year cells; and 17,554 country-age-year cells. FESE sample sizes vary by country-year and are reported separately in the appendix.", "unit_of_analysis": "Individual respondents in the GEM and FESE surveys, aggregated to country-year or country-age-year cells for analysis.", "access_details": "not stated", "notes": "Entrepreneurship is primarily defined as owning and managing a business up to 42 months old that pays wages. Alternative definitions include all new businesses (including recent failures), start-ups without paid wages, and total early-stage entrepreneurship; results are robust across definitions."}}, "method": {"description": "The study estimates cross-country regression models to assess how demographic structure, particularly median age, is associated with national entrepreneurship rates.", "steps": ["Assemble country-level measures of entrepreneurship rates.", "Collect demographic indicators including median age.", "Standardize median age in selected specifications.", "Estimate regression models relating entrepreneurship rates to median age.", "Interpret coefficients in both raw and standardized terms to assess economic magnitude."], "models": "Linear regression models estimated at the country level.", "outcome_variable": "Entrepreneurship rate.", "independent_variables": "Median age of the population.", "control_variables": "GDP growth rate, college enrolment rate, start-up cost, property rights index, military service, percentage of agriculture in GDP.", "tools_software": "not stated"}, "results": {"summary": "Countries with younger median ages have higher entrepreneurship rates. The magnitude of the association implies that demographic differences can account for a substantial share of cross-country variation in entrepreneurship.", "numerical_results": [{"outcome_name": "Entrepreneurship rate", "value": "2.5", "unit": "percentage-point change in entrepreneurship rate associated with a one-standard-deviation decrease in median age", "effect_size": "standardized regression effect", "confidence_interval": {"lower": "not stated", "upper": "not stated", "level": "not stated"}, "p_value": "<0.01", "statistical_significance": 1, "direction": "negative"}]}, "metadata": {"original_paper_id": "not stated", "original_paper_title": "Demographics and Entrepreneurship", "original_paper_code": "not stated", "original_paper_data": "not stated"}}}, "expected_post_registration_3": {"original_study": {"claim": {"hypothesis": "... this factor [the advantages of youth] has a positive effect on entrepreneurship and the importance of that factor declines with age.", "hypothesis_location": "Section: Introduction: ; p. 142", "statement": "The estimates imply that a median age that is one standard deviation lower is associated with a 2.5 percentage point higher country rate of entrepreneurship, which is about 40 percent of the mean rate.", "statement_location": "Section: Introduction; p.144 ", "study_type": "Observational"}, "data": [{"source": "Global Entrepreneurship Monitor", "wave_or_subset": "2001–2010", "sample_size": "1.3 million", "unit_of_analysis": "individual level who are 15 to 60 years of age", "access_details": "http://www.gemconsortium.org", "notes": "The authors aggregate the data up to the country level."}, {"source": "Penn World Table", "wave_or_subset": "version 7.1", "sample_size": "over 200", "unit_of_analysis": "country level", "access_details": "Available at http://www.census.gov/population/international/data/idb/index.php.", "notes": "Used for population data to construct sampling weights."}, {"source": "Education attainment data set", "wave_or_subset": "NA", "sample_size": "not stated", "unit_of_analysis": "country-year level", "access_details": "constructed by Barro and Lee (2010)", "notes": "Limited details as it is only used for control variable."}, {"source": "World Bank database", "wave_or_subset": "not stated", "sample_size": "not stated", "unit_of_analysis": "country level", "access_details": "Available at https://data.worldbank.org/indicator/NV.AGR.TOTL.ZS.", "notes": "Limited details as it is only used for control variable."}, {"source": "international property rights index from the Property Rights Alliance", "wave_or_subset": "NA", "sample_size": "not stated", "unit_of_analysis": "country level", "access_details": "not stated", "notes": "Limited details as it is only used for control variable. Access is not discussed, but a reference is included (Strokova and Mitra 2010). Data is for 2010 only."}, {"source": "Wikipedia", "wave_or_subset": "NA", "sample_size": "not stated", "unit_of_analysis": "country level", "access_details": "Available at http://en.wikipedia.org/wiki/Military_service.", "notes": "Limited details as it is only used for control variable."}], "method": {"description": "The authors use reduced-form regressions at the country level of aggregation to estimate the effect that the age distribution in a country has on the rate of entrepreneurship.", "steps": "(1) Construct sampling weights to aggregate data; (2) Clean data and aggregate to the country level using sampling weights; (3) run reduced-form, country-level regressions.", "models": "reduced-form regressions at the country level of aggregation", "outcome_variable": "entrepreneurship rate", "independent_variables": "age of the particular group; share of population younger", "control_variables": "year dummies, log of per capita GDP, the 5-year per capita average growth rate, the share of agriculture sector in GDP, the tertiary education completion rate, an estimate of start-up costs as a percentage of gross national income, an index of intellectual property, and a dummy for whether the country has more than 1 year of compulsory military service.", "tools_software": "not stated"}, "results": {"summary": "The estimates imply that a median age that is one standard deviation lower is associated with a 2.5 percentage point higher country rate of entrepreneurship, which is about 40 percent of the mean rate.", "numerical_results": [{"outcome_name": "Entrepreneurship Rate", "value": "0.007", "unit": "standard deviation", "effect_size": "2.5 percentage point increase", "confidence_interval": {"lower": "not stated", "upper": "not stated", "level": "not stated"}, "p_value": "<0.01", "statistical_significance": "at 1% level", "direction": "negative", "notes": "Confidence intervals are not presented, but standard errors are (0.001). Discussed positively but written negatively because these results are discussed in terms of a decrease in median age."}]}, "metadata": {"original_paper_id": "0022-3808/2018/126S1-0007", "original_paper_title": "Demographics and Entrepreneurship", "original_paper_code": "not stated", "original_paper_data": "not stated"}}}}
{"study_id": "19", "original_paper_pdf": "19/input/original_paper.pdf", "initial_details": "[CLAIM]\nNations with efficient governments and tight cultures have been most effectiveat limiting COVID-19’s infection rate and mortality likelihood (a significant interaction betweentightness and efficiency, b = -.17, SE = .07, t(41) = -2.23, p = .031).\n\n[HYPOTHESIS]\nThe interaction between cultural tightness and government efficiency will be negative in its association with the COVID-19 infection rate.", "replication_data_files": ["19/input/replication_data/gelfand_replication_data.csv", "19/input/replication_data/Analysis_script_v2.do"], "human_preregistration": "19/gt/human_preregistration.pdf", "human_report": "19/gt/human_report.docx", "expected_post_registration": {"original_study": {"claim": {"hypothesis": "The interaction between cultural tightness and government efficiency will be negative in its association with the COVID-19 infection rate.", "hypothesis_location": "p. 3, second paragraph", "statement": "Nations with efficient governments and tight cultures have been most effective at limiting COVID-19’s infection rate and mortality likelihood (a significant interaction between tightness and efficiency, b = -.17, SE = .07, t(41) = -2.23, p = .031).", "statement_location": "p. 7, first paragraph", "study_type": "Observational"}, "data": {"source": "COVID-19: European Center for Disease Control; government efficiency: the World Bank’s Government Efficiency Index; cultural tightness: index from Gelfand, M. J., Raver, J. L., Nishii, L., Leslie, L. M., Lun, J., Lim, B. C., ... & Aycan, Z. (2011). Differences between tight and loose cultures: A 33-nation study. science, 332(6033), 1100-1104; economic development: GDP per capita retrieved  from the International Monetary Fund; inequality: nations’ Gini coefficients retrieved from the World Bank; median ages: CIA World Factbook.", "wave_or_subset": "World Bank: 2017(for government efficiency), and unspecified for inequality (the authors mention most recent release for each nation); Covid: 2020; International Monetary Fund: 2019; CIA World Factbook: 2018", "sample_size": "141", "unit_of_analysis": "nation", "access_details": "all the data is available on the OSF (https://osf.io/pc4ef/); it is not certain if the original data collected by the authors is open access or not but no restrictions are mentioned.", "notes": "The authors downloaded data on cases per million citizens, and indexed death rate through the number of mortalities divided by the number of total cases. According to the government efficiency metric, efficient governments score highly on 5 dimensions: they are efficient in spending public revenue, they do not place strong compliance burdens on the private sector, they are able to efficiently settle legal and judicial disputes in the private sector, they are receptive to challenges from the private sector, and they offer transparent information about changes in government policies and regulations affecting private sector 5 activities. The mortality likelihood was measured through the number of deaths from COVID-19 divided by the number of COVID-19 cases in a nation. Economic development was indexed through GDP per capita, Inequality was indexed through the nations’ Gini coefficients."}, "method": {"description": "The authors tested the hypothesis that nations with efficient governments and tight cultures have been most effective at limiting COVID-19’s infection rate and mortality likelihood.", "steps": "1. Retrieve and filter the virus data to focus on the period after each nation surpassed 1 case per million people. \n2. Calculate the infection rate per million citizens for each day.\n3. Capture infection rate by fitting regression equations for each nation, log-transforming the outcome variable (cases per million people) and the predictor variable (days) to account for the exponential growth rate of the virus.\n4. Get government efficiency and cultural tightness data.\n5. Combine nation-level growth rate estimates with government efficiency and cultural tightness scores.\n6. Add and standardize control variables.\n7. Examine the distribution of the rate of cases and mortality likelihood before fitting the models like the authors did. They discovered that the growth rate of cases was normally distributed, but the mortality likelihood was highly skewed. If the distribution is similar, proceed to steps 8 and 9. If not, choose a model that better fits the data’s distributional characteristics.\n8. Conduct weighted ordinary least squares regression predicting log-transformed COVID-19 infection rate per million, including both main effects of cultural tightness and government efficiency and their interaction.\n9. Weight each nation’s observation by the number of days of data available to account for reliability.", "models": "ordinary least squares regression", "outcome_variable": "COVID-19 infection rate", "independent_variables": "government efficiency, cultural tightness", "control_variables": "economic development, inequality, median age", "tools_software": "not stated"}, "results": {"summary": "The authors tested the interaction of cultural tightness and government efficiency on growth rates of COVID-19 and found a significant interaction betweentightness and efficiency, b = -.17, SE = .07, t(41) = -2.23, p = .031.", "numerical_results": [{"outcome_name": "NA", "value": "-0.17", "unit": "NA", "effect_size": "not stated", "confidence_interval": {"lower": "not stated", "upper": "not stated", "level": "not stated"}, "p_value": "0.31", "statistical_significance": "true", "direction": "negative."}, {"outcome_name": "t-statistic", "value": "-2.23", "unit": "NA", "effect_size": "not stated", "confidence_interval": {"lower": "not stated", "upper": "not stated", "level": "not stated"}, "p_value": "NA", "statistical_significance": "NA", "direction": "NA"}]}, "metadata": {"original_paper_id": "not stated", "original_paper_title": "Cultural and Institutional Factors Predicting the Infection Rate and Mortality Likelihood of the COVID-19 Pandemic.", "original_paper_code": "https://osf.io/pc4ef/", "original_paper_data": "https://osf.io/pc4ef/"}}}, "expected_post_registration_2": {"original_study": {"claim": {"hypothesis": "Countries with stronger social norms (greater cultural tightness) and more capable public institutions (greater government efficiency) will experience slower COVID-19 infection growth, with the slowest growth occurring where both tightness and efficiency are high.", "hypothesis_location": "Early theory and prediction statement (p. 1-2) where the authors propose that cultural tightness and government efficiency should combine to predict slower COVID-19 growth rates.", "statement": "The regression results (p. 7) support the interactive prediction for infection growth: the tightness-by-efficiency interaction is negative and statistically significant, indicating that tightness is linked to slower infection growth particularly in countries with more efficient governments (and efficiency is likewise linked to slower growth particularly in tighter cultures).", "statement_location": "Results paragraph reporting the interaction model for infection rate and the coefficient for the tightness × government efficiency interaction (reported with b, SE, t, and p).", "study_type": "Observational"}, "data": {"source": "European Centre for Disease Prevention and Control (ECDC) daily country-level COVID-19 cases and deaths; World Bank Government Efficiency Index (2017); cultural tightness index (Gelfand et al. 2011) expanded to additional nations (Eriksson et al. 2020); covariates from IMF (GDP per capita, 2019 release), World Bank (Gini coefficient; most recent estimate per nation), and CIA World Factbook (median age, 2018 release).", "wave_or_subset": "COVID-19 case/death data downloaded for 161 nations and updated between March 21 and March 30, 2020. Infection-growth modeling begins after each country exceeds 1 case per million people. Government efficiency data available for 126 nations (2017). Cultural tightness data available for 57 nations.", "sample_size": "The paper reports 528,019 confirmed cases (including 23,672 deaths) across 141 nations as the global case/death totals used in the study framing.", "unit_of_analysis": "Country (nation).", "access_details": "The paper states that all data and code associated with the analyses are available on OSF.", "notes": "Infection rate is operationalized as a country-specific growth-rate estimate obtained by fitting a log–log model of cases-per-million over days (after exceeding 1 case per million). Mortality likelihood is measured as deaths divided by total cases. In the second-stage country-level regressions predicting infection rate, cases are weighted by the number of observations (days) used to estimate each country’s growth curve."}, "method": {"description": "The study first estimates country-specific COVID-19 infection growth rates from daily case-per-million time series and then uses country-level regressions to test whether cultural tightness and government efficiency—individually and interactively—predict cross-national variation in infection growth (and separately mortality likelihood), controlling for economic development, inequality, and median age.", "steps": ["Download daily COVID-19 cases and deaths by country and convert cases to cases per million people.", "For each country, start the growth-rate series once the country exceeds 1 case per million people.", "Estimate each country’s infection growth rate by fitting a regression with log(cases per million) as the outcome and log(days) as the predictor.", "Compute mortality likelihood as deaths divided by total cases for each country.", "Assemble country-level predictors: government efficiency and cultural tightness, plus GDP per capita, Gini, and median age.", "Standardize the covariates (GDP per capita, Gini, median age) prior to estimation.", "Estimate an OLS regression (Gaussian) predicting infection rate and include the tightness × government efficiency interaction; weight the country-level regression by the number of observations contributing to each country’s growth estimate."], "models": "Country-level OLS regression predicting infection rate (Gaussian) with an interaction term between cultural tightness and government efficiency; weighted by the number of daily observations used to estimate each nation’s growth curve.", "outcome_variable": "Infection rate (country-specific log-transformed growth-rate estimate of COVID-19 cases per million).", "independent_variables": "Cultural tightness; government efficiency; cultural tightness × government efficiency interaction.", "control_variables": "GDP per capita; Gini coefficient; median age (all included in the regressions; covariates standardized prior to estimation).", "tools_software": "not stated"}, "results": {"summary": "The interaction between cultural tightness and government efficiency is negative and statistically significant for infection growth, consistent with the idea that countries that are both tighter and more institutionally efficient show especially slow growth in COVID-19 cases per million during the early phase of the pandemic.", "numerical_results": [{"outcome_name": "Infection rate (log-transformed growth rate of cases per million)", "value": "-0.17", "unit": "regression coefficient units (on the modeled infection-rate scale)", "effect_size": "interaction-term coefficient (cultural tightness × government efficiency)", "confidence_interval": {"lower": "not stated", "upper": "not stated", "level": "not stated"}, "p_value": "0.031", "statistical_significance": 1, "direction": "negative"}]}, "metadata": {"original_paper_id": "not stated", "original_paper_title": "Cultural and Institutional Factors Predicting the Infection Rate and Mortality Likelihood of the COVID-19 Pandemic", "original_paper_code": "https://osf.io/pc4ef/", "original_paper_data": "https://osf.io/pc4ef/"}}}, "expected_post_registration_3": {"original_study": {"claim": {"hypothesis": "We predicted that cultural tightness and government efficiency would predict slower growth rates of COVID-19 and lower mortality likelihoods, and that nations with high cultural tightness and high government efficiency would show especially slow growth rate and mortality likelihood.", "hypothesis_location": "Section: Introduction; p. 3.", "statement": "Culturally tight nations and 20 nations with higher levels of government efficiency each had significantly slower COVID-19 infection rates.", "statement_location": "Section: Introduction; p. 5.", "study_type": "Observational"}, "data": [{"source": "European Center for Disease Control", "wave_or_subset": "not stated", "sample_size": "528,019", "unit_of_analysis": "cases", "access_details": "not stated", "notes": "For COVID data."}, {"source": "World Bank’s Government Efficiency Index", "wave_or_subset": "2017", "sample_size": "126", "unit_of_analysis": "country level", "access_details": "not stated", "notes": "For government efficiency data. World Bank data was also used for Gini coefficient data."}, {"source": "index from Gelfand et al.", "wave_or_subset": "2017", "sample_size": "126", "unit_of_analysis": "country level", "access_details": "not stated", "notes": "For cultural tightness data. No access discussed but a reference is included (Gelfand et al., 2011)."}, {"source": "International Monetary Fund", "wave_or_subset": "2019", "sample_size": "not stated", "unit_of_analysis": "country level", "access_details": "not stated", "notes": "For economic development, proxied by GDP per capita, data."}, {"source": "CIA World Factbook", "wave_or_subset": "2018", "sample_size": "not stated", "unit_of_analysis": "country level", "access_details": "not stated", "notes": "For median age data."}], "method": {"description": "The authors run a OLS regression to test the effect of the interaction between government efficiency and cultural tightness on the infection rate of COVID-19 and a logistic regression to test the effect of this interaction on mortality likelihood.", "steps": "(1) Download the data and construct the variables, including constructing death rate by dividing the number of mortalities by the number of cases; (2) Standardize relevant covariates; (3) Take the log of the outcome (cases per million people) and explanatory variable (days) and run a country-level, logged regression of COVID cases on days; (4)  conducted a second set of regressions using the estimates from the initial 10 general linear models to predict cross-cultural variation in the infection rate of COVID-19, weighting cases by the number of observations across nations; (5) Run regressions using interacted terms as well.", "models": "ordinary least squares regression with gaussian distribution and logistic regression with exponential distribution", "outcome_variable": "infection rate of COVID-19 and mortality likelihood", "independent_variables": "government efficiency interacted with cultural tightness", "control_variables": "economic development; inequality; median age", "tools_software": "not stated"}, "results": {"summary": "...nations with high cultural tightness and high government efficiency would have a much lower log-transformed rate of 1.06 new cases per million. Culturally tight nations and nations with higher levels of government efficiency each had significantly lower death rates of COVID-19.", "numerical_results": [{"outcome_name": "growth rates of COVID-19", "value": "0.17", "unit": "not stated", "effect_size": "log-transformed rate of 1.41 new cases per million people per day or 103.21 fewer cases per million people", "confidence_interval": {"lower": "not stated", "upper": "not stated", "level": "not stated"}, "p_value": "0.031", "statistical_significance": "5% level", "direction": "Negative", "notes": "No confidence intervals presented, but standard errors are presented instead (0.07)."}, {"outcome_name": "mortality likelihood", "value": "0.30", "unit": "not stated", "effect_size": "1.43 more people per million", "confidence_interval": {"lower": "not stated", "upper": "not stated", "level": "not stated"}, "p_value": "< .001", "statistical_significance": "0.1% level", "direction": "Negative", "notes": "No confidence intervals presented, but standard errors are presented instead (0.03)."}]}, "metadata": {"original_paper_id": "not stated", "original_paper_title": "Cultural and Institutional Factors Predicting the Infection Rate and Mortality Likelihood of the COVID-19 Pandemic ", "original_paper_code": "https://osf.io/pc4ef/", "original_paper_data": "https://osf.io/pc4ef/"}}}}