*1A. Basic Descriptive Statistics sum housereg drive elecbill affidavit witness adddoc income male age literacy educ occup cityyears if control==1

Transcription

1 *1A Basic Descriptive Statistics sum housereg drive elecbill affidavit witness adddoc income male age literacy educ occup cityyears if control==1 Variable Obs Mean Std Dev Min Max --- housereg drive elecbill affidavit witness adddoc income male age literacy educ occup cityyears sum housereg drive elecbill affidavit witness adddoc income male age literacy educ occup cityyears if ngo==1 Variable Obs Mean Std Dev Min Max --- housereg drive elecbill affidavit witness adddoc income male age literacy educ occup cityyears sum housereg drive elecbill affidavit witness adddoc income male age literacy educ occup cityyears if rti==1 Variable Obs Mean Std Dev Min Max --- housereg drive elecbill affidavit witness adddoc income male age literacy educ occup cityyears

2 sum housereg drive elecbill affidavit witness adddoc income male age literacy educ occup cityyears if bribe==1 Variable Obs Mean Std Dev Min Max --- housereg drive elecbill affidavit witness adddoc income male age literacy educ occup cityyears *1B Nonparametric descriptive statistics - cardreceive summarize cardreceive if bribe ==1, detail cardreceive - Percentiles Smallest 1% % % Obs 24 25% Sum of Wgt 24 50% 82 Mean 8375 Largest Std Dev % % Variance % Skewness % Kurtosis summarize cardreceive if rti ==1, detail cardreceive - Percentiles Smallest 1% % % Obs 23 25% Sum of Wgt 23 50% 120 Mean Largest Std Dev % % Variance % Skewness % Kurtosis summarize cardreceive if ngo ==1, detail cardreceive - Percentiles Smallest 1%

3 5% % Obs 18 25% Sum of Wgt 18 50% 343 Mean Largest Std Dev % % Variance % Skewness % Kurtosis summarize cardreceive if control ==1, detail cardreceive - Percentiles Smallest 1% % % Obs 21 25% Sum of Wgt 21 50% 343 Mean Largest Std Dev % % Variance % Skewness % Kurtosis *1C Nonparametric descriptive statistics - resreceive summarize resreceive if bribe ==1, detail resreceive - Percentiles Smallest 1% 9 9 5% % 9 9 Obs 24 25% 14 9 Sum of Wgt 24 50% 17 Mean 16 Largest Std Dev % % Variance % Skewness % Kurtosis summarize resreceive if rti ==1, detail resreceive - Percentiles Smallest 1% % % Obs 23 25% Sum of Wgt 23 50% 37 Mean Largest Std Dev % % Variance % Skewness % Kurtosis

4 summarize resreceive if ngo ==1, detail resreceive - Percentiles Smallest 1% % % Obs 18 25% Sum of Wgt 18 50% 37 Mean Largest Std Dev % % Variance % Skewness % Kurtosis summarize resreceive if control ==1, detail resreceive - Percentiles Smallest 1% % % Obs 21 25% Sum of Wgt 21 50% 37 Mean Largest Std Dev % % Variance % Skewness % Kurtosis *1D Probit to check randomization probit control drive housereg elecbill income male age educ occup cityyears islam Iteration 0: log likelihood = Iteration 1: log likelihood = Iteration 2: log likelihood = Iteration 3: log likelihood = Iteration 4: log likelihood = Probit regression Number of obs = 86 LR chi2(10) = 1108 Prob > chi2 = Log likelihood = Pseudo R2 = control Coef Std Err z P> z [95% Conf Interval] - drive housereg elecbill income male age educ occup cityyears islam _cons

5 probit ngo drive housereg elecbill income male age educ occup cityyears islam Iteration 0: log likelihood = Iteration 1: log likelihood = Iteration 2: log likelihood = Iteration 3: log likelihood = Iteration 4: log likelihood = Iteration 5: log likelihood = Iteration 6: log likelihood = Iteration 7: log likelihood = Iteration 8: log likelihood = Iteration 9: log likelihood = Iteration 10: log likelihood = Iteration 11: log likelihood = Iteration 12: log likelihood = Iteration 13: log likelihood = Iteration 14: log likelihood = Iteration 15: log likelihood = Iteration 16: log likelihood = Iteration 17: log likelihood = Iteration 18: log likelihood = Iteration 19: log likelihood = Iteration 20: log likelihood = Probit regression Number of obs = 86 LR chi2(10) = 1284 Prob > chi2 = Log likelihood = Pseudo R2 = ngo Coef Std Err z P> z [95% Conf Interval] - drive housereg elecbill income male age educ occup cityyears islam _cons Note: 9 failures and 0 successes completely determined probit rti drive housereg elecbill income male age educ occup cityyears islam note: islam!= 1 predicts failure perfectly islam dropped and 2 obs not used Iteration 0: log likelihood = Iteration 1: log likelihood = Iteration 2: log likelihood = Iteration 3: log likelihood = Iteration 4: log likelihood = Probit regression Number of obs = 84 LR chi2(9) = 1431 Prob > chi2 = Log likelihood = Pseudo R2 = 01451

6 rti Coef Std Err z P> z [95% Conf Interval] - drive housereg elecbill income male age educ occup cityyears _cons probit bribe drive housereg elecbill income male age educ occup cityyears islam note: islam!= 1 predicts failure perfectly islam dropped and 2 obs not used Iteration 0: log likelihood = Iteration 1: log likelihood = Iteration 2: log likelihood = Iteration 3: log likelihood = Probit regression Number of obs = 84 LR chi2(9) = 395 Prob > chi2 = Log likelihood = Pseudo R2 = bribe Coef Std Err z P> z [95% Conf Interval] - drive housereg elecbill income male age educ occup cityyears _cons *1E LPM to check randomization reg control drive housereg elecbill income male age educ occup cityyears islam Source SS df MS Number of obs = F( 10, 75) = 097 Model Prob > F = Residual R-squared = Adj R-squared = Total Root MSE = control Coef Std Err t P> t [95% Conf Interval] - drive housereg elecbill income male

7 age educ occup cityyears islam _cons reg ngo drive housereg elecbill income male age educ occup cityyears islam Source SS df MS Number of obs = F( 10, 75) = 099 Model Prob > F = Residual R-squared = Adj R-squared = Total Root MSE = ngo Coef Std Err t P> t [95% Conf Interval] - drive housereg elecbill income male age educ occup cityyears islam _cons reg rti drive housereg elecbill income male age educ occup cityyears islam Source SS df MS Number of obs = F( 10, 75) = 134 Model Prob > F = Residual R-squared = Adj R-squared = Total Root MSE = rti Coef Std Err t P> t [95% Conf Interval] - drive housereg elecbill income male age educ occup cityyears islam _cons reg bribe drive housereg elecbill income male age educ occup cityyears islam Source SS df MS Number of obs = F( 10, 75) = 040 Model Prob > F = 09435

8 Residual R-squared = Adj R-squared = Total Root MSE = bribe Coef Std Err t P> t [95% Conf Interval] - drive housereg elecbill income male age educ occup cityyears islam _cons *2 Plots gen treatment = 1 if bribe ==1 (62 missing values generated) replace treatment = 2 if rti ==1 (23 real changes made) replace treatment = 3 if ngo ==1 (18 real changes made) replace treatment = 4 if control ==1 (21 real changes made) *2A Dotplots scatter treatment cardreceive [aweight = dotsizecr], ytitle("treatment Group") xtitle("processing Time (days)") ylabel(1 "Bribe" 2 "RTI" 3 "NG > O" 4 "Control") m(oh)

9 Treatment Group Bribe RTI NGO Control Processing Time (days) scatter treatment resreceive [aweight = dotsizeres], ytitle("treatment Group") xtitle("resver Time (days)") ylabel(1 "Bribe" 2 "RTI" 3 "NGO" 4 > "Control") m(oh) Treatment Group Bribe RTI NGO Control ResVer Time (days) *2B Boxplots graph hbox cardreceive, over(treatment, relabel (1 "Bribe" 2 "RTI" 3 "NGO" 4 "Control")) ytitle("processing Time (days)") caption("figure 1 - > Ration Card Processing Time by Treatment Group", size(medlarge))scheme(s2mono)

10 Bribe RTI NGO Control Processing Time (days) Figure 1 - Ration Card Processing Time by Treatment Group graph hbox resreceive, over(treatment, relabel (1 "Bribe" 2 "RTI" 3 "NGO" 4 "Control")) ytitle("residency Verification Time (days)") Bribe RTI NGO Control Residency Verification Time (days)

11 *2C Survival rates by group label define treatment 1 "Bribe" 2 "RTI" 3 "NGO" 4 "Control" label value treatment treatment ltable cardreceive, intervals(20) by(treatment) gr overlay xlabel(0(100) 400) xtitle("processing time") ytitle("proportion without ration card")msymbol(t)legend(off) proportion without ration card Bribe RTI NGO Control processing time *3A Mann-Whitney nonparametric difference of means tests - cardreceive *Bribe versus RTI sort brnum ranksum cardreceive in 1/47, by(bribe) bribe obs rank sum expected combined unadjusted variance adjustment for ties -153 adjusted variance Ho: cardre~e(bribe==0) = cardre~e(bribe==1) z = 5482 Prob > z = *Bribe versus NGO sort bnnum

12 ranksum cardreceive in 1/42, by(bribe) bribe obs rank sum expected combined unadjusted variance adjustment for ties -113 adjusted variance Ho: cardre~e(bribe==0) = cardre~e(bribe==1) z = 5492 Prob > z = *Bribe versus control sort bcnum ranksum cardreceive in 1/45, by(bribe) bribe obs rank sum expected combined unadjusted variance adjustment for ties -815 adjusted variance Ho: cardre~e(bribe==0) = cardre~e(bribe==1) z = 5745 Prob > z = *RTI versus NGO sort rnnum ranksum cardreceive in 1/41, by(rti) rti obs rank sum expected combined unadjusted variance adjustment for ties -341 adjusted variance

13 Ho: cardre~e(rti==0) = cardre~e(rti==1) z = 4366 Prob > z = *RTI versus Control sort rcnum ranksum cardreceive in 1/44, by(rti) rti obs rank sum expected combined unadjusted variance adjustment for ties adjusted variance Ho: cardre~e(rti==0) = cardre~e(rti==1) z = 5177 Prob > z = *NGO versus Control sort ncnum ranksum cardreceive in 1/39, by(ngo) ngo obs rank sum expected combined unadjusted variance adjustment for ties adjusted variance Ho: cardre~e(ngo==0) = cardre~e(ngo==1) z = 0965 Prob > z = *3B Mann-Whitney nonparametric difference of means tests - residency verification *Bribe versus RTI sort brnum ranksum resreceive in 1/47, by(bribe) bribe obs rank sum expected

14 combined unadjusted variance adjustment for ties adjusted variance Ho: resrec~e(bribe==0) = resrec~e(bribe==1) z = 5898 Prob > z = *Bribe versus NGO sort bnnum ranksum resreceive in 1/42, by(bribe) bribe obs rank sum expected combined unadjusted variance adjustment for ties adjusted variance Ho: resrec~e(bribe==0) = resrec~e(bribe==1) z = 5510 Prob > z = *Bribe versus control sort bcnum ranksum resreceive in 1/45, by(bribe) bribe obs rank sum expected combined unadjusted variance adjustment for ties adjusted variance Ho: resrec~e(bribe==0) = resrec~e(bribe==1) z = 5764 Prob > z = 00000

15 *RTI versus NGO sort rnnum ranksum resreceive in 1/41, by(rti) rti obs rank sum expected combined unadjusted variance adjustment for ties adjusted variance Ho: resrec~e(rti==0) = resrec~e(rti==1) z = Prob > z = *RTI versus Control sort rcnum ranksum resreceive in 1/44, by(rti) rti obs rank sum expected combined unadjusted variance adjustment for ties adjusted variance Ho: resrec~e(rti==0) = resrec~e(rti==1) z = 0226 Prob > z = *NGO versus Control sort ncnum ranksum resreceive in 1/39, by(ngo) ngo obs rank sum expected combined

16 unadjusted variance adjustment for ties adjusted variance Ho: resrec~e(ngo==0) = resrec~e(ngo==1) z = 0415 Prob > z = end of do-file *Duration Model Cox Model *Bribe, RTI and NGO vs control *Bribe, RTI and NGO vs control sort brnum stcox bribe rti control, nohr efron Cox regression -- Efron method for ties No of subjects = 86 Number of obs = 86 No of failures = 46 Time at risk = LR chi2(3) = Log likelihood = Prob > chi2 = _t Coef Std Err z P> z [95% Conf Interval] - bribe rti control stcox bribe in 1/47, nohr efron failure _d: censor analysis time _t: cardreceive Iteration 0: log likelihood = Iteration 1: log likelihood = Iteration 2: log likelihood = Iteration 3: log likelihood = Refining estimates: Iteration 0: log likelihood = Cox regression -- Efron method for ties No of subjects = 47 Number of obs = 47 No of failures = 44 Time at risk = 5469 LR chi2(1) = 3793 Log likelihood = Prob > chi2 = _t Coef Std Err z P> z [95% Conf Interval] - bribe

17