107/replication_package/Code/Stata/Tables.do

*** TABLES FOR Carrera, Royer, Stehr, Sydnor, and Taubinsky (2021) ************
*******************************************************************************

*** Initial set-up ***
clear
eststo clear
set more off

*** Set directory for output ***
cd "$main/Output"
/* Note: all tables will output to the working directory. Uses a program from

   master_do_file_for_analysis.do */


*** Load and clean experimental data ******************************************

*** Load dataset ***
use "$main/Data/cleaned_commitment_study_data", clear

*** Variable Creation & Cleaning Steps ***
*** Create an ID variable for use when reshaping long 
gen id = _n

*** Generate info treatment indicators
gen first_info = type_of_info=="1-onlygraph" 
gen new_info = type_of_info=="2-graphplus" 
gen control_info = (new_info == 0 & first_info == 0) 

*** Generate wave indicators
gen wave1 = (wave == "fall") 
gen wave2 = (wave == "winter")
gen wave3 = (wave == "spring")

*** Define anticommitment/commitment variables for each threshold
gen anticommit8 = q170 ==2 if q170<.
gen commit8 = q169 ==2 if q169<.
gen anticommit12 = chose_anticommit11
gen commit12 = chose_commit12
gen anticommit16 = q296==2 if q296<.
gen commit16 = q295 ==2 if q295<.

*** Rescale percent variable
foreach var of varlist percent*{
	replace `var' = `var'/100
	}
	
*** Store number of observations in exclusions
sum flag_low_wtp
loc lowwtp `r(sum)'
latex_write lowwtpobs `lowwtp' numbers
sum flag_exclude_exog
loc endog `r(sum)'
latex_write nexogobs `endog' numbers
loc totalexcl = `lowwtp' + `endog'
latex_write excludeobs `totalexcl' numbers

* Wave 3
sum flag_low_wtp if wave3 == 1
latex_write lowwtpobswthree `r(sum)' numbers
sum flag_exclude_exog if wave3 == 1
latex_write nexogobswthree `r(sum)' numbers

*** Store number of participants in each wave before exclusions
forval i = 1/3{

	if `i' == 1 loc wname "wone"
	else if `i' == 2 loc wname "wtwo"
	else loc wname "wthree"

	sum wave`i' if wave`i' == 1
	loc obs = r(N)
	latex_write `wname'obs `obs' numbers
	}

*** Store share of participants who prefer nothing to contingent $80
sum prefer80contingent_to_0
loc pctpreferzero : di %2.1f (1-r(mean))*100
latex_write pctpreferzero `pctpreferzero' numbers
	
*** Restrict sample to those exogenously assigned incentives
keep if flag_low_wtp == 0 & flag_exclude_exog == 0


*** Descriptive Statistics (appendix_descriptive_stats.tex) *******************

*** Generate variables and labels for the table ***
gen ft_student = student==1 if student<.
gen ft_working = working==1 if working<.
gen fpt_working = working<3 if working<.
gen advanced_degree = educ==5
replace married = married==1 if married<.
recode age  (1 = 24) (2=35.5 ) (3=45.5) (4=57.5) (5=70) , gen(imp_age)
recode income_cat  (1 = 18000) (2=37500 ) (3=75500) (4=150000) (5=.) , gen(imp_inc)

label var female "Female" 
label var imp_inc "Household income$^a$"
label var married "Married"
label var ft_student "Student, full-time"
label var fpt_working "Working, full- or part-time"
label var advanced "Advanced degree$^b$"
label var imp_age "Age$^a$"
label var past4 "Visits in the past 4 weeks, recorded"
label var visits_100 "$\hspace{0.5cm}$ Visits, recorded"
label var past100days_went "$\hspace{0.5cm}$ Visits, self-recollection"
label var past100days_should "Days that \textit{I should have gone, but didn't}"
label var days_0 "Best guess of days I will attend in next 4 weeks"
label var goal "Goal for visits in next 4 weeks"

gen wave_num = "Wave 1" if wave=="fall"
replace wave_num="Wave 2" if wave=="winter"
replace wave_num="Wave 3" if wave=="spring"
tab wave_num // N = 1248 (340 in Wave 1, 509 in Wave 2, 399 in Wave 3)

*** Produce statistics for the table ***
estpost tabstat female imp_age ft_student fpt_working married advanced_degree imp_inc past4, by(wave_num)  statistics(mean sd) columns(statistics)

*** Export table
#delimit;
estout . using `"appendix_descriptive_stats.tex"' ,
cells(mean(fmt(%3.2fc %3.2fc %3.2fc %3.2fc %3.2fc %3.2fc %8.0gc %3.2fc) star) 
sd(fmt(%3.2fc %3.2fc %3.2fc %3.2fc %3.2fc %3.2fc %8.0gc %3.2fc) par))
starlevels(\sym{*} 0.05 \sym{**} 0.01 \sym{***} 0.001, label(" \(p<@\)"))
varwidth(20) modelwidth(12) unstack delimiter(&) end(\\)
prehead(`"{"' `"\def\sym#1{\ifmmode^{#1}\else\(^{#1}\)\fi}"' `"\begin{tabular}{l*{@E}{c}}"' `"\hline"') posthead("\hline")
postfoot(`"\hline"' `"\addlinespace"' `"N & 340 & 509 & 399 & 1,248 \\"'  `"\hline\hline"' `"\end{tabular}"' `"}"' )
label varlabels(_cons Constant, end("" [1em]) nolast) mlabels(none) nonumbers
collabels(none) substitute(_ \_ "\_cons " \_cons "Total" "Overall")
interaction(" $\times$ ") notype level(95) style(tex) replace;

#delimit cr


*** (CCtable.tex) *************************************************************
*** This table was created by hand, so no tex output is produced for it *******

local commitlist "8 12 16"

*** Summarize take-up of commitment contracts ***
foreach X of local commitlist {
  * Declare local for labeling
  if `X' == 8 loc num "eight"
  else if `X' == 12 loc num "twelve"
  else loc num "sixteen"
  * Chose "more" contract
  summarize commit`X'
  * Store mean and number of observations
  loc takeup = round(r(mean)*100)
  loc takeup : di %2.0f `takeup'
  latex_write commit`num'rate `takeup' numbers
  local obsnum : di %5.0fc `r(N)'
  if r(N) > 1000 loc obsnum = subinstr("`obsnum'",",","{,}",.)
  latex_write commit`num'obs `obsnum' numbers
  * Chose "fewer" contract
  summarize anticommit`X'
  * Chose "more" given chose "fewer"
  summarize commit`X' if anticommit`X'==1
  * Chose "fewer" given chose "more"
  summarize anticommit`X' if commit`X'==1
  * Difference between "more" and "more" given chose "fewer" means
  gen commit`X'_cdtn = commit`X' if anticommit`X'==1
  ttest commit`X'_cdtn = commit`X', unpaired
  * Difference between "fewer" and "fewer" given chose "more" means
  gen anticommit`X'_cdtn = anticommit`X' if commit`X'==1
  ttest anticommit`X'_cdtn = anticommit`X', unpaired
}  


*** (CCBeliefs2.tex) **********************************************************

*** Regress expected days vs. probability of meeting threshold, all participants
reg days_0 percent_meet_commit12, robust
est store est11

reg days_0 percent_meet_anticommit, robust
est store est12

reg days_0 percent_meet_commit12 percent_meet_anticommit, robust
est store est13

* Difference of coefficients ("More"-"Fewer")
nlcom diff: _b[percent_meet_commit12]-_b[percent_meet_anticommit], post

* Store differences
local b : di %4.2f _b[diff]
local se : di %4.2f _se[diff]
estadd local diffb "`b'***" : est13 // know significance level via inspection
estadd local diffse "(`se')" : est13

*** Regress expected days vs. probability of meeting threshold, chose commit & anticommit
reg days_0 percent_meet_commit12 percent_meet_anticommit ///
	if commit12 == 1 & anticommit12 == 1, robust
est store est14

* Difference of coefficients ("More"-"Fewer")
nlcom diff: _b[percent_meet_commit12]-_b[percent_meet_anticommit], post

* Store differences
local b : di %4.2f _b[diff]
local se : di %4.2f _se[diff]
estadd local diffb "`b'***" : est14 // know significance level via inspection
estadd local diffse "(`se')" : est14

*** Export table
estout est11 est12 est13 est14 using CCBeliefs2.tex, style(tex) margin replace eqlabels(none) ///
cells(b(star fmt(%9.2f))  se(par fmt(%9.2f)) ) msign(--) starlevels(* 0.1 ** 0.05 *** 0.01) stardetach ///
keep(percent_meet_commit12 percent_meet_anticommit) wrap varwidth(25) ///
order(percent_meet_commit12 percent_meet_anticommit) ///
mgroups("Subjective expected attendance without incentives", pattern(1 0 0 0) ///
span prefix(\multicolumn{@span}{c}{) suffix(})) ///
stats(N diffb diffse, l("N" "\hline \`\`More'' $-$ \`\`Fewer''" " ") fmt(%8.0fc %8.0fc %8.0fc)) ///
varlabels(percent_meet_commit12 "Subj. prob. succeed in \lq\lq more\rq\rq\, contract" percent_meet_anticommit "Subj. prob. succeed in \lq\lq fewer\rq\rq\, contract") ///
collabels(,none) mlabels("(1)" "(2)" "(3)" "(4)") ///
prehead( "\begin{tabular}{l*{@M}{r @{} l}}" "\hline") ///
posthead("\hline") prefoot() postfoot("\hline\hline" "\end{tabular}")


*** (regs_{avg_delta_ex_1,commit}_all.tex) ************************************
	
*** Generate variable for expected attendace w/incentive given
gen days_exp = .
foreach i in 0 1 2 3 5 7{
	replace days_exp = days_`i' if incentive == `i'
	}
	
*** Generate variables for gaps between goal, past, expected, & actual attendance
gen gap_goal_exp = goal - days_0 // under no incentive
gen gap_goal_past = goal - past4
gen gap_actual_exp = visits - days_exp


*** Generate behavior change premium variables (second-order approximation)
gen delta1=(wtp1) - (days_1+days_0)/2
gen delta2 = (wtp2-wtp1) - (days_2+days_1)/2
gen delta3=(wtp3-wtp2) - (days_3+days_2)/2
gen delta5=(wtp5-wtp3)/(5-3) - (days_5+days_3)/2
gen delta7=(wtp7-wtp5)/(7-5) - (days_7+days_5)/2
gen delta12=(wtp12-wtp7)/(12-7) - (days_12+days_7)/2
gen avg_delta_ex1 = (delta2 + delta3 + delta5 + delta7 + delta12)/5
	

** Store number of observations without an incentive
gen got_contract = (incentive == .)
sum got_contract
latex_write gotcontractobs `r(sum)' numbers


*** Perform series of analyses for whole sample

* Declare restriction for following analysis
loc restrict "!missing(incentive)"
	
* Store means and standard deviations
foreach v in gap_goal_exp gap_actual_exp{
	if "`v'" == "gap_goal_exp" loc name "goalexp"
	else loc name "actexp"
	sum `v' if `restrict'
	loc mean : di %4.2f r(mean)
	loc sd : di %4.2f r(sd)
	latex_write `name'mean `mean' numbers
	latex_write `name'sd `sd' numbers
	}
	
* Generate z-scores for variables with constant restriction
foreach dvar in gap_goal_exp gap_actual_exp avg_delta_ex1{
	sum `dvar' if `restrict'
	gen z_`dvar' = (`dvar' - r(mean)) / r(sd) if `restrict'
	}
	
* Run regressions of behavior change premium with regressors & constant restriction
eststo clear
loc k = 1
foreach controls in "" "female i.income_cat married ft_student fpt_working advanced i.age_cat"{
	foreach covar in "first_info new_info" "first_info new_info z_gap_goal_exp" "first_info new_info z_gap_actual_exp"{

		* Run regression
		reg avg_delta_ex1 wave2 wave3 `covar' `controls' if `restrict', r
		eststo spec_`k'
		
		* Store coefficients from certain regressions
		if `k' == 1{
			loc coeff : di %4.2f _b[new_info]
			latex_write enhinfobcpcoeff `coeff' numbers
			}
		else if `k' == 2{
			loc coeff : di %4.2f _b[z_gap_goal_exp]
			latex_write goalexpbcpcoeff `coeff' numbers
			}
		else if `k' == 3{
			loc coeff : di %4.2f _b[z_gap_actual_exp]
			latex_write actualexpbcpcoeff `coeff' numbers
			}
		
		* Store dependent variable mean
		ci means avg_delta_ex1 if `restrict' & !missing(gap_actual_exp)
		local m : di %4.2f r(mean)
		local se : di %4.2f r(se)
		estadd local dvmean `m'
		estadd local dvse "(`se')"
		ci means avg_delta_ex1 if `restrict' & !missing(gap_actual_exp) & ///
			control_info == 1
		local mcontrol : di %4.2f r(mean)
		local secontrol : di %4.2f r(se)
		estadd local dvmeancontrol `mcontrol'
		if `k' == 1 latex_write bcpinfoctrlmean `mcontrol' numbers
		estadd local dvsecontrol "(`secontrol')"
		estadd local wavefe "Yes"
		if `k' == 1 latex_write bcpregmean `m' numbers
		if "`controls'" != "" estadd local ctrl "Yes"
		
		loc ++k
		}
	}
			
* Export regression output
estout spec_1 spec_2 spec_3 using "regs_avg_delta_ex1_all.tex", style(tex) margin replace eqlabels(none) ///
	keep(first_info new_info z_gap_goal_exp z_gap_actual_exp) wrap varwidth(25) ///
	order(first_info new_info z_gap_goal_exp z_gap_actual_exp) ///
	mgroups("\shortstack{Behavior change premium}", pattern(1 0 0) span prefix(\multicolumn{@span}{c}{) suffix(})) ///
	mlabels("(1)" "(2)" "(3)" "(4)",span prefix(\multicolumn{@span}{c}{) suffix(})) ///
	varlabels(z_gap_goal_exp "Goal $-$ exp. attend. (z-score)" z_gap_actual_exp "Actual $-$ exp. attend. (z-score)" first_info "Basic info. treatment" new_info "Enhanced info. treatment" _cons "Constant") ///
	cells(b(star fmt(%9.2f))  se(par fmt(%9.2f))) msign(--) starlevels(* 0.1 ** 0.05 *** 0.01) stardetach ///
	stats(dvmean dvse dvmeancontrol dvsecontrol wavefe N, l("Dep. var. mean:" " " "Dep. var. mean," "info. control group:" "\hline Wave FEs" "N") fmt(%8.0fc %8.0fc %8.0fc %8.0fc %8.0fc %8.0fc)) ///
	collabels(,none) prehead( "\begin{tabular}{l*{@M}{r @{} l}}" "\hline") ///
	posthead("\hline") prefoot("\hline") postfoot("\hline\hline" "\end{tabular}")
	
* Export regression output with demographic controls
estout spec_4 spec_5 spec_6 using "regs_avg_delta_ex1_all_demo.tex", style(tex) margin replace eqlabels(none) ///
	keep(first_info new_info z_gap_goal_exp z_gap_actual_exp) ///
	wrap varwidth(25) order(first_info new_info z_gap_goal_exp z_gap_actual_exp) ///
	mgroups("\shortstack{Behavior change premium}", pattern(1 0 0) span prefix(\multicolumn{@span}{c}{) suffix(})) ///
	mlabels("(1)" "(2)" "(3)" "(4)",span prefix(\multicolumn{@span}{c}{) suffix(})) ///
	varlabels(z_gap_goal_exp "Goal $-$ exp. attend. (z-score)" z_gap_actual_exp "Actual $-$ exp. attend. (z-score)" first_info "Basic info. treatment" new_info "Enhanced info. treatment" _cons "Constant") ///
	cells(b(star fmt(%9.2f))  se(par fmt(%9.2f))) msign(--) starlevels(* 0.1 ** 0.05 *** 0.01) stardetach ///
	stats(dvmean dvse dvmeancontrol dvsecontrol ctrl wavefe N, l("Dep. var. mean:" " " "Dep. var. mean," "info. control group:" "\hline Demographic controls" "Wave FEs" "N") fmt(%8.0fc %8.0fc %8.0fc %8.0fc %8.0fc %8.0fc %8.0fc)) ///
	collabels(,none) prehead( "\begin{tabular}{l*{@M}{r @{} l}}" "\hline") ///
	posthead("\hline") prefoot("\hline") postfoot("\hline\hline" "\end{tabular}")

		
* Run regressions of contract take-up on regressors with a constant restriction
preserve

* Reshape to pool contracts
reshape long commit, i(id) j(t)

* Run regressions with constant restriction
eststo clear
loc k = 1
foreach controls in "" "female i.income_cat married ft_student fpt_working  advanced i.age_cat"{
	foreach covar in "first_info new_info" "z_avg_delta_ex1 first_info new_info" "z_gap_goal_exp first_info new_info" "z_gap_actual_exp first_info new_info"{
		
		* Run regression
		reg commit i.t wave2 wave3 `covar' `controls' if `restrict', vce(cluster id)
		eststo spec_commit`k'
		
		* Store coefficients from certain regressions
		if `k' == 1{
			loc coeff : di %2.1f abs(_b[new_info]*100)
			latex_write enhinfoMCcoeff `coeff' numbers
			}
		else if `k' == 2{
			loc coeff : di %2.0f _b[z_avg_delta_ex1]*100
			latex_write bcpMCcoeff `coeff' numbers
			}
		else if `k' == 3{
			loc coeff : di %2.1f _b[z_gap_goal_exp]*100
			latex_write goalexpMCcoeff `coeff' numbers
			}
		
		* Store dependent variable mean
		ci means commit if `restrict'
		local m : di %4.2f r(mean)
		local mpct : di %4.0f r(mean)*100
		local se : di %4.2f r(se)
		estadd local dvmean `m'
		estadd local dvse "(`se')"
		ci means commit if `restrict' & control_info == 1
		local mcontrol : di %4.2f r(mean)
		local secontrol : di %4.2f r(se)
		estadd local dvmeancontrol `mcontrol'
		estadd local dvsecontrol "(`secontrol')"
		estadd local wavefe "Yes"
		estadd local ccfe "Yes"
		if `k' == 1{
			latex_write MCregmean `m' numbers
			latex_write MCregmeanPCT `mpct' numbers
			}
		if "`controls'" != "" estadd local ctrl "Yes"
		
		loc ++k
		}
	}
	
* Export regression output
estout spec_commit1 spec_commit2 spec_commit3 spec_commit4 using "regs_commit_pooled_all.tex", ///
	style(tex) margin replace eqlabels(none) wrap varwidth(25) ///
	keep(first_info new_info z_avg_delta_ex1 z_gap_goal_exp z_gap_actual_exp) ///
	order(first_info new_info z_avg_delta_ex1 z_gap_goal_exp z_gap_actual_exp) ///
	mgroups("\shortstack{Take-up of \`\`more'' visits contracts}", pattern(1 0 0 0) span prefix(\multicolumn{@span}{c}{) suffix(})) ///
	mlabels("(1)" "(2)" "(3)" "(4)",span prefix(\multicolumn{@span}{c}{) suffix(})) ///
	varlabels(z_avg_delta_ex1 "Behavior change premium (z-score)" z_gap_goal_exp "Goal $-$ exp. attend. (z-score)" z_gap_actual_exp "Actual $-$ exp. attend. (z-score)" first_info "Basic info. treatment" new_info "Enhanced info. treatment" _cons "Constant") ///
	cells(b(star fmt(%9.3f))  se(par fmt(%9.3f))) msign(--) starlevels(* 0.1 ** 0.05 *** 0.01) stardetach ///
	stats(dvmean dvse dvmeancontrol dvsecontrol wavefe ccfe N N_clust, l("Dep. var. mean:" " " "Dep. var. mean," "info. control group:" "\hline Wave FEs" "Contract FEs" "N" "Clusters") fmt(%8.0fc %8.0fc %8.0fc %8.0fc %8.0fc %8.0fc %8.0fc %8.0fc)) ///
	collabels(,none) prehead( "\begin{tabular}{l*{@M}{r @{} l}}" "\hline") ///
	posthead("\hline") prefoot("\hline") postfoot("\hline\hline" "\end{tabular}")
	
* Export regression output with demographic controls
estout spec_commit5 spec_commit6 spec_commit7 spec_commit8 using "regs_commit_pooled_all_demo.tex", ///
	style(tex) margin replace eqlabels(none) wrap varwidth(25) ///
	keep(first_info new_info z_avg_delta_ex1 z_gap_goal_exp z_gap_actual_exp) ///
	order(first_info new_info z_avg_delta_ex1 z_gap_goal_exp z_gap_actual_exp) ///
	mgroups("\shortstack{Take-up of \`\`more'' visits contracts}", pattern(1 0 0 0) span prefix(\multicolumn{@span}{c}{) suffix(})) ///
	mlabels("(1)" "(2)" "(3)" "(4)",span prefix(\multicolumn{@span}{c}{) suffix(})) ///
	varlabels(z_avg_delta_ex1 "Behavior change premium (z-score)" z_gap_goal_exp "Goal $-$ exp. attend. (z-score)" z_gap_actual_exp "Actual $-$ exp. attend. (z-score)" first_info "Basic info. treatment" new_info "Enhanced info. treatment" _cons "Constant") ///
	cells(b(star fmt(%9.3f))  se(par fmt(%9.3f))) msign(--) starlevels(* 0.1 ** 0.05 *** 0.01) stardetach ///
	stats(dvmean dvse dvmeancontrol dvsecontrol ctrl wavefe ccfe N N_clust, l("Dep. var. mean:" " " "Dep. var. mean," "info. control group:" "\hline Demographic controls" "Wave FEs" "Contract FEs" "N" "Clusters") fmt(%8.0fc %8.0fc %8.0fc %8.0fc %8.0fc %8.0fc %8.0fc %8.0fc %8.0fc)) ///
	collabels(,none) prehead( "\begin{tabular}{l*{@M}{r @{} l}}" "\hline") ///
	posthead("\hline") prefoot("\hline") postfoot("\hline\hline" "\end{tabular}")

restore

	
*** additions to numbers.tex **************************************************

*** Change in visits due to 12-visit contract assignment ***
* Generate variable for comparing treatment with control
gen gotcontract = 1 if treatment == "threshold80_12"
replace gotcontract = 0 if treatment == "control"

* Test for difference and store mean difference by take-up of contract
ttest visits if wave3 == 1 & commit12 == 1, by(gotcontract)
loc takeup : di %3.2f r(mu_2)-r(mu_1)
latex_write wantgottwelvediff `takeup' numbers

ttest visits if wave3 == 1 & commit12 == 0, by(gotcontract)
loc takeup : di %3.2f r(mu_2)-r(mu_1)
latex_write rejectgottwelvediff `takeup' numbers

* Store percentage of those who took up & got the contract who met the threshold
gen wantedgot12visits = visits if treatment == "threshold80_12" & commit12 == 1
gen under12 = wantedgot12visits < 12 if !missing(wantedgot12visits)
sum under12
loc pctunder : di %2.0f round(r(mean)*100)
latex_write pctundertwelve `pctunder' numbers


*** Statistics about WTPs at the slider maximum ***
* Reshape data for one observation per participant per incentive
preserve
keep wtp* maxwtp* id
reshape long wtp maxwtp, i(id) j(incentive)

* Generate a variable for the maximum slider value
gen maxval = incentive*30

* Store percentage of participants who maxed out slider
gen slidermaxed = !missing(maxwtp)
sum slidermaxed
loc pctmaxed : di %2.1f r(mean)*100
latex_write pctmaxed `pctmaxed' numbers

* Generate variables for fill-in-blank answers relation to maximum slider value
gen overmaxval = maxwtp > maxval if !missing(maxwtp)
gen atmaxval = maxwtp == maxval if !missing(maxwtp)
gen belowmaxval = maxwtp < maxval if !missing(maxwtp)

* Store percent of participants in each category
foreach relation in over at below{
	sum `relation'maxval
	loc pct`relation'max : di %2.0f r(mean)*100
	latex_write pct`relation'max `pct`relation'max' numbers
	}
restore


*** Share of individual behavior change value measures that are negative ***
* Reshape data for one observation per participant per incentive increase
preserve
keep delta* id
reshape long delta, i(id) j(inc)

* Estimate and store relevant statistic
gen delta_negative = delta < 0
sum delta_negative
loc pctdeltaneg : di %2.0f r(mean)*100
latex_write pctdeltaneg `pctdeltaneg' numbers
restore


*** Additional summary statistics ***
* Percent female and male
sum female
loc pctfemale : di %2.0f r(mean)*100
loc pctfemalealt : di %2.1f r(mean)*100
loc pctmalealt : di %2.1f (1 - r(mean))*100
loc fracfemale : di %4.3f r(mean)
loc fracmale : di %4.3f (1 - r(mean))
foreach stat in pctfemale pctfemalealt pctmalealt fracfemale fracmale{
	latex_write `stat' ``stat'' numbers 
	}

* Mean imputed age
sum imp_age
loc meanage : di %2.0f r(mean)
latex_write meanage `meanage' numbers

* Percent students
sum ft_student
loc pctstudent : di %2.0f r(mean)*100
latex_write pctstudent `pctstudent' numbers

* Percent working
sum fpt_working
loc pctworking : di %2.0f r(mean)*100
latex_write pctworking `pctworking' numbers

* Percent married
sum married
loc pctmarried : di %2.0f r(mean)*100
latex_write pctmarried `pctmarried' numbers

* Mean visits recorded in past 100 days
sum visits_100
loc meanpastvisits : di %2.0f r(mean)
latex_write meanpastvisits `meanpastvisits' numbers

* Mean visits recalled in past 100 days
sum past100days_went
loc meanpastvisitsrecall : di %2.0f r(mean)
latex_write meanpastvisitsrecall `meanpastvisitsrecall' numbers

* Attention check 
gen failed_attention = passed_attention_check == 0
sum failed
loc pctfailed : di %2.1f r(mean)*100
latex_write pctfailedac `pctfailed' numbers

* Chose dominated option
sum areyousure_prefer0_to_20
loc dom20 = r(N)
sum areyousure_prefer0_to_80
loc dom80 = r(N)
sum id
loc pctdominated : di %2.1f ((`dom20' + `dom80')/r(N))*100
latex_write pctchosedom `pctdominated' numbers

* Comprehension check
gen failed_cc = passed_comprehension_check == 0
sum failed_cc
loc pctfailed : di %2.1f r(mean)*100
latex_write pctfailedcc `pctfailed' numbers

* Numeracy check
gen passed_nc = 1 if !missing(q137) & !missing(q138)
replace passed_nc = 0 if (q138 == "133" | q138 == "133333" | q138 == "200000" | ///
 q138 == "2500" | q138 == "40,000" | q138 == "40,000.00" | ///
 q138 == "$400,00.00" | q138 == "400" | q138 == "400,00" | q138 == "40000" | ///
 q138 == "4000000" | q138 == "500,000" | q138 == "50000" | q138 == "800000" | ///
 q138 == "four") & !missing(passed_nc)
replace passed_nc = 0 if q137 != 100 & !missing(passed_nc)
sum passed_nc
loc pctpassed : di %2.1f r(mean)*100
latex_write passednumeracy `pctpassed' numbers

* Take-up of contract for fewer visits with & without exclusions above
preserve
reshape long commit anticommit, i(id) j(t)
sum anticommit
loc takeup : di %2.0f r(mean)*100
latex_write anticommitrate `takeup' numbers

sum anticommit if passed_comprehension_check == 1 ///
	& passed_attention_check == 1 & missing(areyousure_prefer0_to_20) ///
	& missing(areyousure_prefer0_to_80)
loc takeup : di %2.0f r(mean)*100
latex_write anticommitrateexcl `takeup' numbers
restore

* Range of percent take-up of contracts for fewer visits
loc anticommitpctmin = 1 // initialize for contract with lowest take-up
loc anticommitpctmax = 0 // initialize for contract with highest take-up
forval i = 8(4)16{
	sum anticommit`i'
	if r(mean) > `anticommitpctmax' loc anticommitpctmax = r(mean)
	if r(mean) < `anticommitpctmin' loc anticommitpctmin = r(mean)
	}
foreach stat in min max{
	loc anticommitpct`stat' : di %4.0f `anticommitpct`stat''*100
	latex_write anticommitpct`stat' `anticommitpct`stat'' numbers
	}

* Number of participants in waves 2 & 3
gen wave23 = (wave2 | wave3)
sum wave23
loc obs = r(mean)*r(N)
latex_write wtwothreeobs `obs' numbers

* Average expected visits for different incentives
sum days_0
loc avgzero : di %2.1f r(mean)
latex_write avgexpnoinc `avgzero' numbers
sum days_7
loc avgseven : di %2.1f r(mean)
latex_write avgexpseveninc `avgseven' numbers
loc diff : di %2.1f `avgseven'-`avgzero'
latex_write avgexpdiffseven `diff' numbers

* Average actual visits for different incentives
sum visits if incentive==0
loc avgzero : di %2.1f r(mean)
latex_write avgactnoinc `avgzero' numbers
sum visits if incentive==7
loc avgseven : di %2.1f r(mean)
latex_write avgactseveninc `avgseven' numbers
loc diff : di %2.1f `avgseven'-`avgzero'
latex_write avgactdiffseven `diff' numbers

* Regression of attendance on average past attendance
reg visits past4
matrix temp = r(table)'
loc b : di %4.3f temp[1,1]
loc se : di %4.3f temp[1,2]
latex_write attcoeff `b' numbers
latex_write attcoeffse `se' numbers

* Effect of $2 piece-rate incentive
gen got_2 = 1 if incentive == 2
replace got_2 = 0 if treatment == "control"
* On expectations
ttest days_2 == days_0
* On realized attendance
ttest visits, by(got_2) unpaired


*** Generate pairwise correlation estimates with bootstrapping ***
* Pairwise correlation of behavior change premium at each incentive level
capture program drop avg_pwcorr_bcp
program define avg_pwcorr_bcp, rclass

	* Initialize values
	loc m = 0
	loc agg_corr = 0
	
	* Loop through each incentive level
	foreach j in 1 2 3 5 7 {
		foreach k in 2 3 5 7 12 {
			if `j' < `k'{
				pwcorr delta`j' delta`k'
				loc agg_corr = `agg_corr' + r(rho)
				loc ++m
				}
			}
		}
		
	* Average pairwise correlation
	return scalar avg_pwcorr = `agg_corr'/`m'
	
end

* Generate estimates with bootstrapping
bootstrap avg_pwcorr_bcp = r(avg_pwcorr), reps(1000) cluster(id) seed(12345) ///
	nodrop level(95): avg_pwcorr_bcp
estat bootstrap, all

* Store estimates
matrix temp_b = e(b)
loc avg : di %3.2f temp_b[1,1]
latex_write avgbcpcorr `avg' numbers
matrix temp_se = e(se)
loc stderr : di %3.2f temp_se[1,1]
latex_write avgbcpcorrse `stderr' numbers


* Pairwise correlation of take-up of each more-visits contract
capture program drop avg_pwcorr_mc
program define avg_pwcorr_mc, rclass

	* Initialize values
	loc m = 0
	loc agg_corr = 0
	
	* Loop through each more-visits contract
	foreach j in 8 12 {
		foreach k in 12 16 {
			if `j' < `k'{
				pwcorr commit`j' commit`k'
				loc agg_corr = `agg_corr' + r(rho)
				loc ++m
				}
			}
		}
		
	* Average pairwise correlation
	return scalar avg_pwcorr = `agg_corr'/`m'
	
end

* Generate estimates with bootstrapping
bootstrap avg_pwcorr_mc = r(avg_pwcorr), reps(1000) cluster(id) seed(12345) ///
	nodrop level(95): avg_pwcorr_mc
estat bootstrap, all
	
* Store estimates
matrix temp_b = e(b)
loc avg : di %3.2f temp_b[1,1]
latex_write avgMCcorr `avg' numbers
matrix temp_se = e(se)
loc stderr : di %3.2f temp_se[1,1]
latex_write avgMCcorrse `stderr' numbers