{
  "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"
  }
}