From 232ce91caa2647e49157318ab1c348e842a12b2a Mon Sep 17 00:00:00 2001 From: Koen Leuveld Date: Tue, 13 Feb 2024 12:56:32 +0100 Subject: [PATCH] clean up folder --- congogbv_analysis.do | 217 --- congogbv_customfunctions.tex | 32 - congogbv_listanalysis.r | 41 - congogbv_orig.tex | 1624 ----------------- congogbv_paths.tex | 3 - congogbv_preamble.tex | 73 - .../congogbv_dataprep.do | 0 .../congogbv_dhs.do | 0 .../congogbv_helpers.do | 0 9 files changed, 1990 deletions(-) delete mode 100644 congogbv_analysis.do delete mode 100644 congogbv_customfunctions.tex delete mode 100644 congogbv_listanalysis.r delete mode 100644 congogbv_orig.tex delete mode 100644 congogbv_paths.tex delete mode 100644 congogbv_preamble.tex rename congogbv_dataprep.do => stata_do_files/congogbv_dataprep.do (100%) rename congogbv_dhs.do => stata_do_files/congogbv_dhs.do (100%) rename congogbv_helpers.do => stata_do_files/congogbv_helpers.do (100%) diff --git a/congogbv_analysis.do b/congogbv_analysis.do deleted file mode 100644 index e7fb7d7..0000000 --- a/congogbv_analysis.do +++ /dev/null @@ -1,217 +0,0 @@ -/* -Analysis of GBV in Congo, based on MFS II baseline data. - -Dependencies: -congogbv_dataprep.do: prepares MFS II data, and ACLED data. -congogbv_helpers.do: defines programs to create figures and tables. - -Author: Koen Leuveld -Git repo: https://github.com/freetambo/congogbv.git - -Date: 10/02/2020 - -*/ - -set scheme lean1 - -*set controls -global dataloc X:\Dropbox (Personal)\PhD\Papers\CongoGBV\Data_encrypted //holds raw and clean data -global tableloc C:\Users\Koen\Dropbox (Personal)\PhD\Papers\CongoGBV\Tables //where tables are put -global figloc C:\Users\Koen\Dropbox (Personal)\PhD\Papers\CongoGBV\Figures //where figures are put -global gitloc C:\Users\Koen\Documents\GitHub //holds do files - -global allcontrols agewife agehusband genderhead eduwife_prim eduwife_sec eduhusband_prim eduhusband_sec tinroof livestockany terrfe_* treatment - - -*run helpers -run "$gitloc\congogbv\congogbv_helpers.do" //contains functions used to generate tables and figures. -run "$gitloc\congogbv\congogbv_dataprep.do" //cleans data. - - -********************************************* -**TABLE 1: Comparison between DHS and my sample -********************************************* -local using using "$tableloc/dhs_compare.tex" - -use "$dataloc\clean\dhs.dta", clear -eststo dhs_nat: estpost su agewife tinroof eduwife_prim eduwife_sec [iweight=wgt] -eststo dhs_sk: estpost su agewife tinroof eduwife_prim eduwife_sec [iweight=wgt] if province == 11 - -use "$dataloc\clean\analysis.dta", clear -eststo sample_all: estpost su agewife tinroof eduwife_prim eduwife_sec -eststo sample_selected: estpost su agewife tinroof eduwife_prim eduwife_sec if !missing(ball5) - - -esttab dhs_nat dhs_sk sample_all sample_selected `using', cells("mean(fmt(2))") label noobs mtitles("DHS National" "DHS South Kivu" "Full Sample" "Gender Module") replace nonumbers - -eststo clear - -********************************************* -**Table 2: Gender module sample make up -********************************************* -use "$dataloc\clean\analysis.dta", clear - -*sample make up -tab2csv riskwifestatus riskhusbandstatus using "$tableloc/sample_tabs.csv" -tabout riskwifestatus riskhusbandstatus using "$tableloc/sample_tabs.tex", replace style(tex) format(0c) // h3(nil) - -************************** -**Table 4: Balance Table** -************************** -use "$dataloc\clean\analysis.dta", clear -drop if ball5 == . - -balance_table /// - numballs ///list experiment - victimproplost victimfamlost acledviolence10 /// conflict - husbmoreland wifemoreland /* contribcash contribcashyn*/ riskwife riskhusband barghusbandcloser bargwifecloser /// bargainin and empowerment - $allcontrols /// - if !missing(ball5) using "$tableloc\balance.tex", /// - rawcsv treatment(ball5) cluster(vill_id) - -/* -reg ball5 /// - victimproplost victimfamlost acledviolence10 /// conflict - husbmoreland wifemoreland /* contribcashyn */ riskwife riskhusband barghusbandcloser bargwifecloser /// bargainin and empowerment - atthusbtotal attwifetotal /// gender attitidues - , vce(cluster vill_id) - */ - -********************************************** -**Figure 1: Mean Comparisons Overall** -********************************************** -tempfile diffs -meandiffs numballs using "$figloc/meancompare_overall.png", treatment(ball5) coeffs(`diffs') //!!!meandiffs fuction is defined in congogbv_helpers.do - - - -********************************************** -**Figure 2: Mean Comparisons across Conflict** -********************************************** -graph drop _all -meandiffs numballs, treatment(ball5) by(victimproplost) coeffs(`diffs') append name(meancompare_conf1, replace) -meandiffs numballs, treatment(ball5) by(victimfamlost) coeffs(`diffs') append name(meancompare_conf2, replace) -meandiffs numballs, treatment(ball5) by(acledviolence10d) coeffs(`diffs') append name(meancompare_conf3, replace) - -grc1leg meancompare_conf1 meancompare_conf2 meancompare_conf3, position(4) ring(0) -graph export "$figloc/meancompare_conf.png", as(png) replace - -*********************************** -**Table 5: conflict by region -*********************************** -local using using "$tableloc/conflict_by_terr.tex" -eststo conflict_comp: estpost tabstat victimproplost victimfamlost acledviolence10, by(territory) statistics(mean sd) columns(statistics) -esttab conflict_comp `using', main(mean) aux(sd) nostar unstack nonote label noobs nonumbers replace - - -*********************************** -**Table 6:Mean Comparisons across Conflict -*********************************** -meandifftab numballs using "$tableloc\meandifftab_conf.csv",by(victimproplost victimfamlost acledviolence10d) treat(ball5) robust - - - - -********************************************** -**Figure 3: Mean Comparisons Marriage** -********************************************** -meandiffs numballs, treatment(ball5) by(statpar) coeffs(`diffs') append name(meancompare_mar1,replace) -meandiffs numballs, treatment(ball5) by(bargresult) coeffs(`diffs') append name(meancompare_mar2,replace) -//meandiffs numballs, treatment(ball5) by(contribcashyn) coeffs(`diffs') append name(meancompare_mar3,replace) - -grc1leg meancompare_mar1 meancompare_mar2 // , position(4) ring(0) -graph export "$figloc/meancompare_mar.png", as(png) replace - - -********************************************** -**Table 7: Mean Comparisons Marriage** -********************************************** -meandifftab numballs using "$tableloc\meandifftab_mar.csv",by(husbmoreland wifemoreland barghusbandcloser bargwifecloser) treat(ball5) robust - - -*export to CSV -preserve -use `diffs', clear -export delimited using "$tableloc\incidence.csv", datafmt replace -restore - -********************************************** -**Table 8: Multivariate Regression -********************************************** - -local using using "$tableloc\results_regression.tex" - -global controls agewife agehusband genderhead eduwife_sec eduhusband_prim tinroof livestockany terrfe_* treatment - -tempfile regs //"$tableloc\regs.csv" -eststo l1: kict ls numballs husbmoreland $controls, condition(ball5) nnonkey(4) estimator(linear) vce(cluster vill_id) -regsave using "`regs'", replace addlabel(reg,l1) pval -eststo l2: kict ls numballs victimfamlost $controls , condition(ball5) nnonkey(4) estimator(linear) vce(cluster vill_id) -regsave using "`regs'" , append addlabel(reg,l2) pval -eststo l3: kict ls numballs acledviolence10 $controls, condition(ball5) nnonkey(4) estimator(linear) vce(cluster vill_id) -regsave using "`regs'", append addlabel(reg,l3) pval -eststo l5: kict ls numballs husbmoreland victimfamlost acledviolence10 attwifetotal $controls, condition(ball5) nnonkey(4) estimator(linear) vce(cluster vill_id) -regsave using "`regs'", append addlabel(reg,l5) pval - - -esttab l? `using', replace /// - nomtitles keep(Delta:*) order(Delta:husbmoreland Delta:victimfamlost Delta:acledviolence10 Delta:attwifetotal) se label /// - starlevels(* 0.10 ** 0.05 *** 0.01) nonotes - -preserve -use `regs', clear -gen coef_pct = abs(coef) * 100 -format coef stderr pval %9.2f -format coef_pct %9.0f -export delimited using "$tableloc/regs.csv", datafmt replace - -restore - -********************************************* -**TABLE A1: Sample Selection -********************************************* -use "$dataloc\clean\analysis.dta", clear - -gen wifeconsent = riskwifestatus == 1 if !missing(riskwifestatus) -gen husbandconsent = riskhusbandstatus == 1 if !missing(riskhusbandstatus) -gen coupleconsent = wifeconsent * husbandconsent - - -*sample selected -local using using "$tableloc/attrition.tex" - - -eststo attrwife, t("Wife"): logit wifeconsent $allcontrols, vce(cluster vill_id) -eststo attrhusband, t("Husband"): logit husbandconsent $allcontrols, vce(cluster vill_id) -eststo attrcouple, t("Couple"): logit coupleconsent $allcontrols, vce(cluster vill_id) - -esttab attr* `using', replace /// - nodepvar se label /// - starlevels(* 0.10 ** 0.05 *** 0.01) nonotes eqlabels("" "") - -********************************************** -**Table A2: Determinants** -********************************************** -local using using "$tableloc\determinants_regression.tex" - -use if !missing(ball5) using "$dataloc\clean\analysis.dta" , clear -eststo clear -local depvars husbmoreland victimfamlost acledviolence10 -local rh_vars $allcontrols -foreach var of varlist `depvars' { - local rh_depvars : list depvars - var - di "var: `var'" - di "rh_depvars: `rh_depvars'" - eststo det_`var': reg `var' `rh_depvars' `rh_vars', vce(cluster vill_id) -} - - -esttab det_* `using', replace /// - mtitles("\specialcell{Family MR\\had more land}" /// - "\specialcell{Bargaining:\\closer to FR}" /// - "\specialcell{Conflict pre-2012:\\HH member killed}" /// - "\specialcell{Conflict 2013-2014:\\Viol. against civilians}") /// - depvars se label order(`depvars' `rh_vars') drop(genderhead) /// - starlevels(* 0.10 ** 0.05 *** 0.01) nonotes - - diff --git a/congogbv_customfunctions.tex b/congogbv_customfunctions.tex deleted file mode 100644 index d84c9df..0000000 --- a/congogbv_customfunctions.tex +++ /dev/null @@ -1,32 +0,0 @@ -%get summstats from balance table -\newcommand{\summstat}[2]{\csvreader[filter strcmp={\var}{#1}]{\congogbvTables/balance.csv}{var=\var,#2=\stat}{\stat}} - -%create table for mean differences -\csvstyle{meandifftable}{tabular=lccccc,table head= \toprule Variable & N &Control & Treatment & Diff & St. Err. \\\toprule,late after line=\\,table foot=\bottomrule} - -\newcommand{\meandifftab}[1]{ - \begin{threeparttable} - \csvreader[meandifftable]{#1}% - {n0 = \na, n1= \nb, label0 = \laba, label1 = \labb, varlabel=\var,meancontrol0 = \mca, meantreat0 = \mta ,stardiff0=\diffa, sediff0=\sea, meancontrol1 = \mcb, meantreat1 = \mtb , stardiff1=\diffb, sediff1=\seb, stardd = \dd, sedd=\sedd}% - {% - %\multicolumn{6}{l}{\var} \\ - \var & & & & & \\ % - \quad \laba & \na & \mca & \mta & \diffa & \sea \\% - \quad \labb & \nb & \mcb & \mtb & \diffb & \seb \\% - \quad Diff in Diff & & & & \dd & \sedd - }% - \begin{tablenotes} - \small - \item Robust Standard errors reported. - \item * p $<$ 0.1, **, p $<$ 0.05, *** p $<$ 0.01 - \end{tablenotes} - \end{threeparttable} -} - - -\newcommand{\coeffget}[3]{\csvreader[filter=\equal{\reg}{#1} \and \equal{\var}{#2}]{\congogbvTables/regs.csv}{var=\var,reg=\reg,#3=\coeff}{\coeff}} - -\newcommand{\incid}[2]{\csvreader[filter strcmp={\key}{#1}]{\congogbvTables/incidence.csv}{key=\key,#2=\inc}{\inc}} - -\newcommand{\specialcell}[2][c]{% - \begin{tabular}[#1]{@{}c@{}}#2\end{tabular}} \ No newline at end of file diff --git a/congogbv_listanalysis.r b/congogbv_listanalysis.r deleted file mode 100644 index 6148873..0000000 --- a/congogbv_listanalysis.r +++ /dev/null @@ -1,41 +0,0 @@ -library(list) -library(readstata13) -library(pastecs) -library(fastDummies) - -dta <- read.dta13("C:/Users/Koen/Dropbox (Personal)/PhD/Papers/CongoGBV/Data/clean/analysis.dta") - -reg <- ictreg(numballs ~ terrfe_2 + terrfe_3 + terrfe_4 + husbmoreland,data=dta,treat="ball5") -summary(reg) - - -round(stat.desc(dta[,c("numballs","terrfe_2","terrfe_3","terrfe_4","husbmoreland")]),2) - -dta_complete <- na.omit(dta[,c("numballs","ball5","terrfe_2","terrfe_3","terrfe_4","husbmoreland")]) -round(stat.desc(dta_complete[,c("numballs","terrfe_2","terrfe_3","terrfe_4","husbmoreland")]),2) - - -reg <- ictreg(numballs ~ terrfe_2 + terrfe_3 + terrfe_4 + husbmoreland ,data=dta_complete,treat="ball5") - - -dta$villid2 <- as.factor(dta[,"vill_id"]) - -reg <- ictreg(numballs ~ husbmoreland ,data=dta,treat="ball5") -round(stat.desc(dta[,c("vill_id","numballs","terrfe_2","terrfe_3","terrfe_4","husbmoreland")]),2) -tabulate(dta$villid2) - - - -reg <- ictreg(numballs ~ factor(vill_id) + husbmoreland ,data=dta,treat="ball5") - -dta[1:10,grepl("^terrfe_", names(dta))] - - -analysis <- dta[c("numballs","ball5","vill_id","husbmoreland")] - -analysis <- fastDummies::dummy_cols(analysis, select_columns = "vill_id") - -analysis$vill_id <-NULL - -reg <- ictreg(numballs ~ . -ball5 ,data=analysis,treat="ball5") - diff --git a/congogbv_orig.tex b/congogbv_orig.tex deleted file mode 100644 index cf694c2..0000000 --- a/congogbv_orig.tex +++ /dev/null @@ -1,1624 +0,0 @@ -% Options for packages loaded elsewhere -\PassOptionsToPackage{unicode}{hyperref} -\PassOptionsToPackage{hyphens}{url} -% -\documentclass[ -]{article} -\usepackage{amsmath,amssymb} -\usepackage{iftex} -\ifPDFTeX - \usepackage[T1]{fontenc} - \usepackage[utf8]{inputenc} - \usepackage{textcomp} % provide euro and other symbols -\else % if luatex or xetex - \usepackage{unicode-math} % this also loads fontspec - \defaultfontfeatures{Scale=MatchLowercase} - \defaultfontfeatures[\rmfamily]{Ligatures=TeX,Scale=1} -\fi -\usepackage{lmodern} -\ifPDFTeX\else - % xetex/luatex font selection -\fi -% Use upquote if available, for straight quotes in verbatim environments -\IfFileExists{upquote.sty}{\usepackage{upquote}}{} -\IfFileExists{microtype.sty}{% use microtype if available - \usepackage[]{microtype} - \UseMicrotypeSet[protrusion]{basicmath} % disable protrusion for tt fonts -}{} -\makeatletter -\@ifundefined{KOMAClassName}{% if non-KOMA class - \IfFileExists{parskip.sty}{% - \usepackage{parskip} - }{% else - \setlength{\parindent}{0pt} - \setlength{\parskip}{6pt plus 2pt minus 1pt}} -}{% if KOMA class - \KOMAoptions{parskip=half}} -\makeatother -\usepackage{xcolor} -\usepackage[margin=1in]{geometry} -\usepackage{color} -\usepackage{fancyvrb} -\newcommand{\VerbBar}{|} -\newcommand{\VERB}{\Verb[commandchars=\\\{\}]} -\DefineVerbatimEnvironment{Highlighting}{Verbatim}{commandchars=\\\{\}} -% Add ',fontsize=\small' for more characters per line -\usepackage{framed} -\definecolor{shadecolor}{RGB}{248,248,248} -\newenvironment{Shaded}{\begin{snugshade}}{\end{snugshade}} -\newcommand{\AlertTok}[1]{\textcolor[rgb]{0.94,0.16,0.16}{#1}} -\newcommand{\AnnotationTok}[1]{\textcolor[rgb]{0.56,0.35,0.01}{\textbf{\textit{#1}}}} -\newcommand{\AttributeTok}[1]{\textcolor[rgb]{0.13,0.29,0.53}{#1}} -\newcommand{\BaseNTok}[1]{\textcolor[rgb]{0.00,0.00,0.81}{#1}} -\newcommand{\BuiltInTok}[1]{#1} -\newcommand{\CharTok}[1]{\textcolor[rgb]{0.31,0.60,0.02}{#1}} -\newcommand{\CommentTok}[1]{\textcolor[rgb]{0.56,0.35,0.01}{\textit{#1}}} -\newcommand{\CommentVarTok}[1]{\textcolor[rgb]{0.56,0.35,0.01}{\textbf{\textit{#1}}}} -\newcommand{\ConstantTok}[1]{\textcolor[rgb]{0.56,0.35,0.01}{#1}} -\newcommand{\ControlFlowTok}[1]{\textcolor[rgb]{0.13,0.29,0.53}{\textbf{#1}}} -\newcommand{\DataTypeTok}[1]{\textcolor[rgb]{0.13,0.29,0.53}{#1}} -\newcommand{\DecValTok}[1]{\textcolor[rgb]{0.00,0.00,0.81}{#1}} -\newcommand{\DocumentationTok}[1]{\textcolor[rgb]{0.56,0.35,0.01}{\textbf{\textit{#1}}}} -\newcommand{\ErrorTok}[1]{\textcolor[rgb]{0.64,0.00,0.00}{\textbf{#1}}} -\newcommand{\ExtensionTok}[1]{#1} -\newcommand{\FloatTok}[1]{\textcolor[rgb]{0.00,0.00,0.81}{#1}} -\newcommand{\FunctionTok}[1]{\textcolor[rgb]{0.13,0.29,0.53}{\textbf{#1}}} -\newcommand{\ImportTok}[1]{#1} -\newcommand{\InformationTok}[1]{\textcolor[rgb]{0.56,0.35,0.01}{\textbf{\textit{#1}}}} -\newcommand{\KeywordTok}[1]{\textcolor[rgb]{0.13,0.29,0.53}{\textbf{#1}}} -\newcommand{\NormalTok}[1]{#1} -\newcommand{\OperatorTok}[1]{\textcolor[rgb]{0.81,0.36,0.00}{\textbf{#1}}} -\newcommand{\OtherTok}[1]{\textcolor[rgb]{0.56,0.35,0.01}{#1}} -\newcommand{\PreprocessorTok}[1]{\textcolor[rgb]{0.56,0.35,0.01}{\textit{#1}}} -\newcommand{\RegionMarkerTok}[1]{#1} -\newcommand{\SpecialCharTok}[1]{\textcolor[rgb]{0.81,0.36,0.00}{\textbf{#1}}} -\newcommand{\SpecialStringTok}[1]{\textcolor[rgb]{0.31,0.60,0.02}{#1}} -\newcommand{\StringTok}[1]{\textcolor[rgb]{0.31,0.60,0.02}{#1}} -\newcommand{\VariableTok}[1]{\textcolor[rgb]{0.00,0.00,0.00}{#1}} -\newcommand{\VerbatimStringTok}[1]{\textcolor[rgb]{0.31,0.60,0.02}{#1}} -\newcommand{\WarningTok}[1]{\textcolor[rgb]{0.56,0.35,0.01}{\textbf{\textit{#1}}}} -\usepackage{longtable,booktabs,array} -\usepackage{calc} % for calculating minipage widths -% Correct order of tables after \paragraph or \subparagraph -\usepackage{etoolbox} -\makeatletter -\patchcmd\longtable{\par}{\if@noskipsec\mbox{}\fi\par}{}{} -\makeatother -% Allow footnotes in longtable head/foot -\IfFileExists{footnotehyper.sty}{\usepackage{footnotehyper}}{\usepackage{footnote}} -\makesavenoteenv{longtable} -\usepackage{graphicx} -\makeatletter -\def\maxwidth{\ifdim\Gin@nat@width>\linewidth\linewidth\else\Gin@nat@width\fi} -\def\maxheight{\ifdim\Gin@nat@height>\textheight\textheight\else\Gin@nat@height\fi} -\makeatother -% Scale images if necessary, so that they will not overflow the page -% margins by default, and it is still possible to overwrite the defaults -% using explicit options in \includegraphics[width, height, ...]{} -\setkeys{Gin}{width=\maxwidth,height=\maxheight,keepaspectratio} -% Set default figure placement to htbp -\makeatletter -\def\fps@figure{htbp} -\makeatother -\setlength{\emergencystretch}{3em} % prevent overfull lines -\providecommand{\tightlist}{% - \setlength{\itemsep}{0pt}\setlength{\parskip}{0pt}} -\setcounter{secnumdepth}{-\maxdimen} % remove section numbering -% definitions for citeproc citations -\NewDocumentCommand\citeproctext{}{} -\NewDocumentCommand\citeproc{mm}{% - \begingroup\def\citeproctext{#2}\cite{#1}\endgroup} -\makeatletter - % allow citations to break across lines - \let\@cite@ofmt\@firstofone - % avoid brackets around text for \cite: - \def\@biblabel#1{} - \def\@cite#1#2{{#1\if@tempswa , #2\fi}} -\makeatother -\newlength{\cslhangindent} -\setlength{\cslhangindent}{1.5em} -\newlength{\csllabelwidth} -\setlength{\csllabelwidth}{3em} -\newenvironment{CSLReferences}[2] % #1 hanging-indent, #2 entry-spacing - {\begin{list}{}{% - \setlength{\itemindent}{0pt} - \setlength{\leftmargin}{0pt} - \setlength{\parsep}{0pt} - % turn on hanging indent if param 1 is 1 - \ifodd #1 - \setlength{\leftmargin}{\cslhangindent} - \setlength{\itemindent}{-1\cslhangindent} - \fi - % set entry spacing - \setlength{\itemsep}{#2\baselineskip}}} - {\end{list}} -\usepackage{calc} -\newcommand{\CSLBlock}[1]{\hfill\break\parbox[t]{\linewidth}{\strut\ignorespaces#1\strut}} -\newcommand{\CSLLeftMargin}[1]{\parbox[t]{\csllabelwidth}{\strut#1\strut}} -\newcommand{\CSLRightInline}[1]{\parbox[t]{\linewidth - \csllabelwidth}{\strut#1\strut}} -\newcommand{\CSLIndent}[1]{\hspace{\cslhangindent}#1} -\usepackage{multirow} -\usepackage{multicol} -\usepackage{colortbl} -\usepackage{hhline} -\newlength\Oldarrayrulewidth -\newlength\Oldtabcolsep -\usepackage{longtable} -\usepackage{array} -\usepackage{hyperref} -\usepackage{float} -\usepackage{wrapfig} -\ifLuaTeX - \usepackage{selnolig} % disable illegal ligatures -\fi -\usepackage{bookmark} -\IfFileExists{xurl.sty}{\usepackage{xurl}}{} % add URL line breaks if available -\urlstyle{same} -\hypersetup{ - pdftitle={Sexual violence, conflict, and female empowerment: Exploratory evidence from a list experiment in Eastern DR Congo}, - pdfauthor={Koen Leuveld}, - hidelinks, - pdfcreator={LaTeX via pandoc}} - -\title{Sexual violence, conflict, and female empowerment: Exploratory -evidence from a list experiment in Eastern DR Congo} -\author{Koen Leuveld} -\date{} - -\begin{document} -\maketitle - -\section*{Introduction}\label{introduction} -\addcontentsline{toc}{section}{Introduction} - -Over the past decades, tremendous progress has been made worldwide to -improve the lives of the world's poor. The proportion of people living -under the poverty line of \$1.25 per day dropped from over half to just -14\%; gender disparity in primary education has been drastically reduced -or even eliminated; under-five mortality rates have been halved (United -Nations 2015). However, such progress has largely bypassed fragile -states, like the Democratic Republic of the Congo (DRC) (Asadullah and -Savoia 2018; Samy and Carment 2011). Congolese women in particular face -economic hardships and human rights violations, including a high rate of -Sexual and Gender Based Violence (SGBV): estimates of the proportion of -women who have suffered from this range from 15\% to 40\% (Johnson et -al. 2010; Peterman, Palermo, and Bredenkamp 2011). Aid workers have -called the country ``the world's worst place to be a woman or a child'', -and the UN's Special Representative on Sexual Violence in Conflict, -Margot Wallström, even called the country the ``rape capital of the -world'' (Human Rights Watch 2009). The issue of SGBV is of specific -concern, given its high psychological, social and economic costs (Post -et al. 2002; Peterson et al. 2018). Consequently, tremendous -international efforts have been made to implement or support projects to -assist the victims of SGBV. The 2018 Nobel Peace prize was awarded to -Dr.~Denis Mukwege, for his work on victims of SGBV at Bukavu's Panzi -Hospital. - -Despite this attention, very little reliable data exists on the topic -(Palermo and Peterman 2011). Data collection efforts have been hampered -by the conflicts the country has faced, which have made large-scale data -collection from representative samples difficult. As a result, most data -available on the topic is from surveys within clinics and NGOs aimed at -assisting victims of SGBV, making comparison between victims of SGBV and -non-victims difficult. These samples are obviously skewed, as they only -survey women who have already come forward in search for help. Even when -survey data is available, the sensitive nature of the topic may cause -respondents to withhold information due their unease in discussing -sensitive topics with survey field staff. - -In this paper, I explore the characteristics of victims of SGBV as well -as non-victims to study the dynamics and potential drivers of SGBV. I -classify female survey respondents according to potential risk factors -for SGBV, and analyse whether these factors are in fact associated with -the incidence of SGBV. In this way, this paper aims to address the -question what the drivers of SGBV in Eastern Congo are. Specifically, I -consider conflict and the position of women in Congolese society as -potential drivers, and compare their relative contribution. - -The conflict that has persisted in the country for the past decades is -the most often-cited driver of the high rate of SGBV. This is -particularly true for policy circles, where the framing of SGBV in Congo -as ``weapon of war'' is popular (Baaz Eriksson and Stern 2013; Kirby -2015). There is empirical evidence to support this notion. (Johnson et -al. 2010) carried out a large-scale survey in Eastern Congo to -investigate incidence and perpetrators of SGBV, and found that the -majority of sexual violence reported by their respondents was -conflict-related; of female victims of sexual violence, 74.3\% reported -the perpetrators to be conflict-related. Likewise, Bartels et al. (2013) -find that the majority of the victims of SGBV treated at Panzi Hospital --- in Bukavu, South Kivu -- indicate that the perpetrators were armed -groups. It is therefore not surprising that the topic of SGBV in Congo -has often been analysed within the context of violent conflict (Baaz -Eriksson and Stern 2013). Conversely, the aspect of the conflict that -has received the most world-wide media attention has been SGBV -(Autesserre 2012). - -However, this view of the central role of conflict in sexual violence in -the DRC has come under increasing scrutiny. It has been argued that this -focus on the relationship between sexual violence and conflict has been -counter-productive, as it has distracted attention from other pressing -problems the DRC faces (Autesserre 2012; Hilhorst and Douma 2018; Porter -2019). Moreover, it risks missing the civilian perpetrators of SGBV. -There is empirical evidence for this position too. Based on DHS data, -(Peterman, Palermo, and Bredenkamp 2011) find that rates of Sexual -Intimate Partner Violence (IPV) are higher than rates of other forms of -Sexual Violence in Congo. - -This increased focus on IPV, rather than conflict, shifts attention from -conflict to the bargaining position of women in Congolese society and -households as a driver of SGBV. The bargaining position of a woman is -the level of autonomy she has, and is determined by things such as her -outside options; its effect on SGBV is ambiguous (Eswaran and Malhotra -2011). One the one hand, a woman's welfare may depend on her bargaining -position: women with more income, and better prospects in case of a -divorce would thus face less risk of IPV. On the other hand a woman's -partner may use IPV as an instrument to assert power as a response to -her increased empowerment. The empirical record reflects this ambiguity. -(Bhattacharya, Bedi, and Chhachhi 2009) find that increase in -employment, and the increasing of status of a woman within the household -reduces violence. Similarly, (Hidrobo, Peterman, and Heise 2016) find -that cash transfers to women, decrease the risk of violence. However, -when specifically looking at sexual IPV in the Dominican Republic, -(Bueno and Henderson 2017) find that an increase in women's (economic) -empowerment led to an increase in IPV. In Vietnam, (Bulte and Lensink -2019) find that a project aiming to increase women's income, may have -led to increased IPV. In Afghanistan, (Gibbs et al. 2020) find no link -(positive or negative) between economic empowerment and IPV. The link -between IPV and women's intra-household bargaining position may be -moderated by local customs, and depend on exactly the type of IPV and -the type of empowerment under consideration. - -These two main drivers of SGBV -- conflict and empowerment -- are not -necessarily separate, as conflict may affect empowerment on the long -run. In the short run, conflict may have a direct effect on SGBV through -perpetration by armed groups, potentially in a strategic manner (Baaz -Eriksson and Stern 2013; Kirby 2015). This caused the topic to be on the -international agenda, as a ``weapon of war''. However, in the long run -there is a more indirect effect as well: conflict causes the breakdown -of norms, which may have long-lasting effects. For example, (Kelly et -al. 2018) find that that IPV increased in districts that experienced -conflict in Liberia, while (Müller and Tranchant 2019) draw similar -conclusions from data from the Gaza strip. (Saile et al. 2013) -investigate the correlates of IPV for a sample of conflict-exposed women -in Northern Uganda. They find that while the level of conflict exposure -predicts physical violence, sexual violence is more associated with the -level of childhood familial violence. This link between current and past -experiences of violence suggests that the effect of conflict on violence -is deeper than just the direct effect. People traumatized during the -conflict (either because they were victims or perpetrators) are more -likely to be victimized later on. - -In answering the question what the main drivers of SGBV are, I thus -consider two main drivers: the position of women in Congolese society -and conflict. Within conflict, I distinguish between historic conflict -(here I use pre-2012 data) with long-term, indirect, effects and recent -conflict (up to one year prior to the interview) with short-term, direct -effects. I argue that the indirect long-term effects of conflict are -likely to be related to the position of women, through changing norms, -while more recent conflict events may not have had an additional impact -on norms yet. For empowerment, I use a bargaining game, and survey -questions that determine women's pre-marriage relative status. - -This paper contributes to the empirical evidence base on the incidence -of SGBV in Eastern Congo by drawing on a sample of beneficiaries of -development assistance projects in South Kivu province, in the Eastern -DRC. While the selection of respondents was not done to produce a -representative sample for the province, it does not suffer from the same -problems that clinic-based surveys have, allowing me to compare victims -of SGBV with non-victims. Data on SGBV comes from a list experiment, a -technique which has been gaining popularity as a way to obtain -information on sensitive topics (see e.g. Sniderman, Tetlock, and Piazza -1991; Holbrook and Krosnick 2010; Bulte and Lensink 2019; Peterman et -al. 2018; LaBrie and Earleywine 2000; Corstange 2009). Put briefly, list -experiments allow group-level analysis of SGBV victimization, without -individuals revealing their own victimization status. This eliminates -the need for respondents to withhold information and thus reduces the -social desirability bias that results (Blair and Imai 2012). Such bias -may explain the fact that studies on the drivers of SGBV often -contradict each other. While one study finds conflict-related -perpetrators are responsible for the majority of cases of SGBV (Johnson -et al. 2010), another finds intimate partners as the most common -culprits (Peterman, Palermo, and Bredenkamp 2011). Stark et al. (2017) -provides an example of how different methodologies can provide different -answers: when using Audio Assisted Self-Interviews (ACASI) they find -that intimate partners are the main perpetrators of SBV. However, in -complementary group discussions, where social desirability bias is -likely to be present, respondents did not bring up intimate partners at -all. - -I combine list experiments with detailed survey data on the household -and outcomes from behavioural experiments, which allows for a rich -characterization of victims of SGBV. Because such a characterization is -lacking thus far, this data is useful in addressing and preventing SGBV. -Moreover, while the potential drivers of SGBV mentioned above -- -conflict and empowerment -- have been studied in isolation, this paper -contributes by analysing these in one framework. - -I find high victimization rates in my sample: 30\% of the women report -SGBV in the past twelve months. These victims are likely to be married -to higher-status men, have low intra-household bargaining power, and -have been exposed to violent conflict to the extent where they have lost -family or household members before 2012 (two years before the list -experiment). I find no evidence of a link between SGBV and recent -conflict exposure. These findings are consistent with recent findings in -the literature that conflict has long-lasting impact on SGBV through -IPV. This paper is structured as follows: first I describe the research -setting, the sample, and then the various sources of data. The -subsequent section describes my empirical framework, which revolves -around the use of a list experiment. I then present the results of the -analyses. In the concluding section I contextualize the findings and -present policy implications. - -\section*{Background}\label{background} -\addcontentsline{toc}{section}{Background} - -Congo's 2006 constitution grants equal rights to men and women. In -practice, however, women hold an inferior position in Congolese society. -This is reflected in social and economic outcomes. The literacy rates -among women and girls aged 15-24 is 73.6\% (compared to 91.2\% among men -and boys of the same age); only 8.5\% of women have completed secondary -education (compared to 16.2\% of the men); while 67\% of women work, -only 7.8\% work outside of agriculture or trading and services (MPSMRM, -MSP, and ICF International 2014). Within the household, women occupy an -inferior position as well: the husband is the head by law, and marital -rape is not considered a crime (Kilonzo et al. 2009). - -In addition to the difficulties inherent to their inferior position, -women have faced widespread human rights abuses during the conflicts -that have swept the country since the mid 1990s. South Kivu (the setting -for the present study) has been greatly affected by these conflicts. The -first Congo war started with an invasion by Rwanda and armed groups -supported by Rwanda to clear perpetrators of the Rwanda genocide from -the refugee camps in the east of the country, putting the province on -the front line. Throughout this First Congo War (1996-1997), the Second -Congo War (1998-2003) and the subsequent fragile peace, ethnic tensions -have remained high throughout the province, resulting in frequent -localized bursts of violence (see e.g. Verwijen 2016). While some of the -human rights abuses during these phases of the conflict occurred during -large-scale attacks on civilians, often they occurred during ambushes -while women were conducting their day-to-day tasks (Freedman 2011; Human -Rights Watch 2002). Women were often assaulted by multiple perpetrators. -These were not only members of rebel groups, but also the government -army (Human Rights Watch 2009). - -The consequences of (conflict-related) SGBV for the victims have been -well-researched. It has severe mental and physical health consequences -(Johnson et al. 2010). However, due to the remote nature and lack of -resources, victims have difficulty finding professional help, often -having to travel more than a day to clinics (Harvard Humanitarian -Initiative 2009; Kohli et al. 2012). The negative consequences persist -until long after the event, as victims face stigmatization within their -communities and households (Albutt et al. 2017; Harvard Humanitarian -Initiative 2009). - -The adverse consequences of conflict-related SGBV do not remain limited -to the direct victims. The violence against women during the conflict -resulted in a change in norms, where armed groups were no longer the -main perpetrators of SGBV, but civilians (including intimate partners) -(Freedman 2011). Risk factors for sexual IPV include partner problematic -use of alcohol and partner controlling behaviours (Babalola, -Gill-Bailey, and Dodo 2014). While the Congolese government has made -attempts to address the situation, such as through the Law on the -Suppression of Sexual Violence, implementation of these measures has -been marred by the general lack of resources state authority in the -country (Steiner et al. 2009). - -\section*{Sample}\label{sample} -\addcontentsline{toc}{section}{Sample} - -The main source of data for this study is the gender module from a -household survey that was undertaken in 2014 as the endline survey for -the evaluation of Dutch development aid. This evaluation concerned -projects ran by four NGOs in the territories of Kabare, Fizi and Uvira, -and the commune of Bagira.\footnote{In the remainder of the paper, I - will consider Kabare and Bagira to be one ``territory'', since the - selected communities in Kabare and Bagira are located close together, - in the peri-urban zone of Bukavu.} The baseline for this evaluation -was done in 2012. Half of the respondents were selected from communities -that benefited from the projects, the other half were selected from -comparable households in non-intervention communities. These projects -were about agriculture, women's rights and education. Overall, the -beneficiaries of the projects were vulnerable, mostly rural, households. -An indicator for being beneficiary to any of these projects is included -in the full analysis below. In total data was collected in 73 -communities. In each community, baseline data was collected on 15 -households in 2012; however, due to attrition, 2014 data is available -for an average of 12 households per community, for a total of 889 -households. - -\begin{Shaded} -\begin{Highlighting}[] -\FunctionTok{read\_csv}\NormalTok{(}\FunctionTok{here}\NormalTok{(}\StringTok{"tables/table1.csv"}\NormalTok{)) }\SpecialCharTok{\%\textgreater{}\%} \FunctionTok{flextable}\NormalTok{()} -\end{Highlighting} -\end{Shaded} - -\begin{verbatim} -## Warning: fonts used in `flextable` are ignored because the `pdflatex` engine is -## used and not `xelatex` or `lualatex`. You can avoid this warning by using the -## `set_flextable_defaults(fonts_ignore=TRUE)` command or use a compatible engine -## by defining `latex_engine: xelatex` in the YAML header of the R Markdown -## document. -\end{verbatim} - -\global\setlength{\Oldarrayrulewidth}{\arrayrulewidth} - -\global\setlength{\Oldtabcolsep}{\tabcolsep} - -\setlength{\tabcolsep}{0pt} - -\renewcommand*{\arraystretch}{1.5} - - - -\providecommand{\ascline}[3]{\noalign{\global\arrayrulewidth #1}\arrayrulecolor[HTML]{#2}\cline{#3}} - -\begin{longtable}[c]{|p{0.75in}|p{0.75in}|p{0.75in}|p{0.75in}|p{0.75in}} - - - -\ascline{1.5pt}{666666}{1-5} - -\multicolumn{1}{>{\raggedleft}m{\dimexpr 0.75in+0\tabcolsep}}{\textcolor[HTML]{000000}{\fontsize{11}{11}\selectfont{agewife}}} & \multicolumn{1}{>{\raggedleft}m{\dimexpr 0.75in+0\tabcolsep}}{\textcolor[HTML]{000000}{\fontsize{11}{11}\selectfont{tinroof}}} & \multicolumn{1}{>{\raggedleft}m{\dimexpr 0.75in+0\tabcolsep}}{\textcolor[HTML]{000000}{\fontsize{11}{11}\selectfont{eduwife\_prim}}} & \multicolumn{1}{>{\raggedleft}m{\dimexpr 0.75in+0\tabcolsep}}{\textcolor[HTML]{000000}{\fontsize{11}{11}\selectfont{eduwife\_sec}}} & \multicolumn{1}{>{\raggedright}m{\dimexpr 0.75in+0\tabcolsep}}{\textcolor[HTML]{000000}{\fontsize{11}{11}\selectfont{label}}} \\ - -\ascline{1.5pt}{666666}{1-5}\endfirsthead - -\ascline{1.5pt}{666666}{1-5} - -\multicolumn{1}{>{\raggedleft}m{\dimexpr 0.75in+0\tabcolsep}}{\textcolor[HTML]{000000}{\fontsize{11}{11}\selectfont{agewife}}} & \multicolumn{1}{>{\raggedleft}m{\dimexpr 0.75in+0\tabcolsep}}{\textcolor[HTML]{000000}{\fontsize{11}{11}\selectfont{tinroof}}} & \multicolumn{1}{>{\raggedleft}m{\dimexpr 0.75in+0\tabcolsep}}{\textcolor[HTML]{000000}{\fontsize{11}{11}\selectfont{eduwife\_prim}}} & \multicolumn{1}{>{\raggedleft}m{\dimexpr 0.75in+0\tabcolsep}}{\textcolor[HTML]{000000}{\fontsize{11}{11}\selectfont{eduwife\_sec}}} & \multicolumn{1}{>{\raggedright}m{\dimexpr 0.75in+0\tabcolsep}}{\textcolor[HTML]{000000}{\fontsize{11}{11}\selectfont{label}}} \\ - -\ascline{1.5pt}{666666}{1-5}\endhead - - - -\multicolumn{1}{>{\raggedleft}m{\dimexpr 0.75in+0\tabcolsep}}{\textcolor[HTML]{000000}{\fontsize{11}{11}\selectfont{31.82608}}} & \multicolumn{1}{>{\raggedleft}m{\dimexpr 0.75in+0\tabcolsep}}{\textcolor[HTML]{000000}{\fontsize{11}{11}\selectfont{0.3345647}}} & \multicolumn{1}{>{\raggedleft}m{\dimexpr 0.75in+0\tabcolsep}}{\textcolor[HTML]{000000}{\fontsize{11}{11}\selectfont{0.4705190}}} & \multicolumn{1}{>{\raggedleft}m{\dimexpr 0.75in+0\tabcolsep}}{\textcolor[HTML]{000000}{\fontsize{11}{11}\selectfont{0.09845465}}} & \multicolumn{1}{>{\raggedright}m{\dimexpr 0.75in+0\tabcolsep}}{\textcolor[HTML]{000000}{\fontsize{11}{11}\selectfont{dhs\_national}}} \\ - - - - - -\multicolumn{1}{>{\raggedleft}m{\dimexpr 0.75in+0\tabcolsep}}{\textcolor[HTML]{000000}{\fontsize{11}{11}\selectfont{31.08862}}} & \multicolumn{1}{>{\raggedleft}m{\dimexpr 0.75in+0\tabcolsep}}{\textcolor[HTML]{000000}{\fontsize{11}{11}\selectfont{0.5968956}}} & \multicolumn{1}{>{\raggedleft}m{\dimexpr 0.75in+0\tabcolsep}}{\textcolor[HTML]{000000}{\fontsize{11}{11}\selectfont{0.2929335}}} & \multicolumn{1}{>{\raggedleft}m{\dimexpr 0.75in+0\tabcolsep}}{\textcolor[HTML]{000000}{\fontsize{11}{11}\selectfont{0.05500434}}} & \multicolumn{1}{>{\raggedright}m{\dimexpr 0.75in+0\tabcolsep}}{\textcolor[HTML]{000000}{\fontsize{11}{11}\selectfont{dhs\_skivu}}} \\ - - - - - -\multicolumn{1}{>{\raggedleft}m{\dimexpr 0.75in+0\tabcolsep}}{\textcolor[HTML]{000000}{\fontsize{11}{11}\selectfont{40.57590}}} & \multicolumn{1}{>{\raggedleft}m{\dimexpr 0.75in+0\tabcolsep}}{\textcolor[HTML]{000000}{\fontsize{11}{11}\selectfont{0.5846501}}} & \multicolumn{1}{>{\raggedleft}m{\dimexpr 0.75in+0\tabcolsep}}{\textcolor[HTML]{000000}{\fontsize{11}{11}\selectfont{0.2491309}}} & \multicolumn{1}{>{\raggedleft}m{\dimexpr 0.75in+0\tabcolsep}}{\textcolor[HTML]{000000}{\fontsize{11}{11}\selectfont{0.02433372}}} & \multicolumn{1}{>{\raggedright}m{\dimexpr 0.75in+0\tabcolsep}}{\textcolor[HTML]{000000}{\fontsize{11}{11}\selectfont{sample\_full}}} \\ - - - - - -\multicolumn{1}{>{\raggedleft}m{\dimexpr 0.75in+0\tabcolsep}}{\textcolor[HTML]{000000}{\fontsize{11}{11}\selectfont{41.09106}}} & \multicolumn{1}{>{\raggedleft}m{\dimexpr 0.75in+0\tabcolsep}}{\textcolor[HTML]{000000}{\fontsize{11}{11}\selectfont{0.6053963}}} & \multicolumn{1}{>{\raggedleft}m{\dimexpr 0.75in+0\tabcolsep}}{\textcolor[HTML]{000000}{\fontsize{11}{11}\selectfont{0.2529511}}} & \multicolumn{1}{>{\raggedleft}m{\dimexpr 0.75in+0\tabcolsep}}{\textcolor[HTML]{000000}{\fontsize{11}{11}\selectfont{0.02866779}}} & \multicolumn{1}{>{\raggedright}m{\dimexpr 0.75in+0\tabcolsep}}{\textcolor[HTML]{000000}{\fontsize{11}{11}\selectfont{sample\_gendermodule}}} \\ - -\ascline{1.5pt}{666666}{1-5} - - - -\end{longtable} - - - -\arrayrulecolor[HTML]{000000} - -\global\setlength{\arrayrulewidth}{\Oldarrayrulewidth} - -\global\setlength{\tabcolsep}{\Oldtabcolsep} - -\renewcommand*{\arraystretch}{1} - -The sampling procedure outlined here is thus unlikely to have produced a -nationally (or even provincially) representative sample. In Table -\hyperref[tab:dhs_compare]{1}, I present a comparison across selected -demographics between the full study sample (column 3), and the -representative sample from the DHS Program (columns 1-2). Women in the -study sample are older, and less likely to have finished school, than -the provincial average in South Kivu. - -\phantomsection\label{tab:bargsample} -\begin{longtable}[]{@{}lcllll@{}} -\caption{Gender module sample make up}\tabularnewline -\toprule\noalign{} -\endfirsthead -\endhead -\bottomrule\noalign{} -\endlastfoot -& Male Respondent & & & & \\ -Female Respondent & Consented & Refused & Absent & No Husband & Total \\ -& No. & No. & No. & No. & No. \\ -Consented & 184 & 3 & 253 & 153 & 593 \\ -Refused & 0 & 0 & 0 & 1 & 1 \\ -Absent & 4 & 0 & 0 & 1 & 5 \\ -No Wife & 282 & 3 & 2 & 3 & 290 \\ -Total & 470 & 6 & 255 & 158 & 889 \\ -\end{longtable} - -Not all households participated fully in the gender module. Where -possible, it was administered to both the head of the household and -their spouse, so that there were a Female and a Male Respondent to the -interview \footnote{In tables, I refer to Male and Female Respondents as - MR and FR respectively.}. In the vast majority of the cases, the -husband is considered the head, but it was left open to the respondents -to indicate the head. Table \hyperref[tab:bargsample]{2} displays how -the sample is built up. In total, there were 889 respondents to the -survey. In 593 households, the Female Respondent (the wife of the -household head or the female head) consented to responding to the gender -module. In 1 household, the female respondent refused; in 5, the Female -Respondent was absent during the interview, and in 290 households the -head of the household had no wife, and there was thus no Female -Respondent. In 470 households, the Male Respondent (usually the -household head) consented to the module, 6 refused, 255 Male Respondents -were absent, and in 158 households the head of the household was an -unmarried woman, meaning that there was no Male Respondent. In 184 -households, both husband and wife responded to the module. Efforts to -increase this number, by tracking down absent household heads, were -constricted by the limited time field teams had in each community due to -the security situation at the time of field work. As a consequence of -this, sample sizes between various analyses are different: analyses -relying on both partners being present - e.g.~for the bargaining game - -will have a lower sample size than others. - -The selection of respondents to the gender module is unlikely to have -been random. In column 4 of Table \hyperref[tab:dhs_compare]{1} selected -demographics for the Female Respondents to the gender module are -presented. The respondents are slightly older than the full sample, and -considerably older than the provincial average. They are slightly more -likely to have completed secondary school than the full sample, but less -likely than the provincial average. In Table -\hyperref[tab:sample_selection]{6}, I present results from logit models -to find correlations between household characteristics and participation -in the gender module. The dependent columns of the columns are whether -the wife, the husband and the couple participated in the gender module, -respectively. There are some selection effects. Households that own tin -roofs, are more likely to have a Female Respondent. In households that -own livestock, it was less likely that there was a female respondent to -the gender module, and more likely to have male respondent. The final -analysis below will include these as controls. - -\section*{Methods}\label{methods} -\addcontentsline{toc}{section}{Methods} - -This paper combines data from the 2014 and 2012 rounds of the survey, -with ACLED data. The gender module from the 2014 survey is the main -source of data for this paper. The module was administered separately to -Male and Female Respondent (with a small part being administered -jointly). It contained (i) a list experiment designed to elicit the -incidence of SGBV among Female Respondents; (ii) a risk bargaining game -to elicit the relative intra-household bargaining position of the Male -and Female Respondent; and (iii) a set of propositions to collect -detailed information on gender attitudes. I present the List Experiment, -and the analysis thereof, in more detail in the Empirical Framework -below. - -\phantomsection\label{tab:bargaining} -\begin{longtable}[]{@{}lcccc@{}} -\caption{Bargaining game lotteries}\tabularnewline -\toprule\noalign{} -\# & Low & High & Expected & Risk aversion \\ -\midrule\noalign{} -\endfirsthead -\toprule\noalign{} -\# & Low & High & Expected & Risk aversion \\ -\midrule\noalign{} -\endhead -\bottomrule\noalign{} -\endlastfoot -1 & 4,000 CDF & 4,000 CDF & 4,000 CDF & Extremely risk-averse \\ -2 & 3,600 CDF & 4,800 CDF & 4,200 CDF & Extremely risk-averse \\ -3 & 3,200 CDF & 5,600 CDF & 4,400 CDF & Moderately risk-averse \\ -4 & 2,800 CDF & 6,400 CDF & 4,600 CDF & Moderately risk-averse \\ -5 & 2,400 CDF & 7,200 CDF & 4,800 CDF & Risk-neutral \\ -6 & 1,400 CDF & 8,200 CDF & 4,800 CDF & Risk-loving \\ -\end{longtable} - -The risk bargaining game in the gender module was modified from -(Martinsson, Sutter, and Carlsson 2009). In the game, the respondents -chose between a set of six risky lotteries, based on (Eckel and Grossman -2002). The lotteries presented ranged from fairly low-risk ones -- where -low and high pay-out were nearly equal -- to high-risk one -- where -there was a large difference between high and low pay-outs (see Table -\hyperref[tab:bargaining]{3} for details of the lotteries). The Male and -Female Respondents first chose privately (without knowing their -partner's choice), and then jointly. By comparing the couple decision -with the individual decision, I obtain an indicator for bargaining -power: the closer the couple decision is to the Female Respondent's -decision -- relative to the Male Respondent's decision -- the higher her -bargaining power. The difference between the procedure used by -(Martinsson, Sutter, and Carlsson 2009) and the one here, is that they -use a risk experiment based on (Holt and Laury 2002); a more complicated -experiment compared to Eckel and Grossman (2002). This added -complication may cause some participants to not fully understand the -procedure, leading to poor results (Dave et al. 2010). Given the low -numeracy of the subjects, I implemented the simpler of the two -experiments. - -I draw on two sources for conflict data: data from the 2012 round of the -survey, and ACLED data from 2013-2014. The 2012 data contains detailed -information of the conflict history of the respondents dating back to -the start of the First Congo War in 1996. Among other things, -respondents were asked whether they lost family members, whether they -lost property, and when these events took place. I use this to construct -indicators for historic victimization, which may have indirect effects -on SGBV victimization. Due to time constraints, the 2014 round of the -survey did not contain a detailed conflict exposure module. In order to -get more detailed information on recent victimization, complement the -household-level data with more recent data from the Armed Conflict -Location \& Event Data Project {[}ACLED; Raleigh et al. (2010){]}. The -2014 data contains GPS coordinates for all interviewed households. Using -these coordinates, I can link households to nearby conflict events from -the ACLED database that took place within the 12 months preceding the -interview. Because this data then coincides with the window for SGBV -used here, any direct effects from conflict -- like perpetration of SGBV -by armed groups -- will be captured by this indicator. However, while -this data is more recent, it does not capture individual experience; -only exposure based on the distance from the household to conflict -events. - -\section*{Empirical Strategy}\label{empirical-strategy} -\addcontentsline{toc}{section}{Empirical Strategy} - -A major concern in collecting data on SGBV is reporting bias. -Respondents are unlikely to be comfortable to truthfully answer -questions about SGBV. Respondents may want to hide undesirable answers, -leading to what's called social desirability bias. Not only may this -lead to an underestimate of the incidence of SGBV, the unwillingness to -divulge information may be correlated to the identity of the -perpetrators: people may be more willing to divulge victimization from -armed groups, than from intimate partners (Stark et al. 2017). This -non-random nature of non-response would thus lead to an underestimate of -the incidence of SGBV, and biased estimates for the correlates of SGBV -when using direct questions. This is why such direct questions are not -used in list experiments. Instead, interviewers present respondents with -a list of issues and ask them to indicate the number of issues from the -list they have faced. By adding the sensitive item to the list of issues -for half of the respondents (randomly selected), estimates for incidence -of the sensitive item can be obtained by comparing the mean number of -issues faced in both groups (hence ``item count technique'' as an -alternative name for list experiments). The advantage is thus that -answers are guaranteed to be anonymous: the interviewer (or the data -analyst) does not know the number of non-sensitive issues the respondent -faces and so has no way of knowing the answer to sensitive item. This -anonymity removes the need to hide the answer, and thus the social -desirability bias. - -Over the past decades, list experiments have grown in popularity as a -way to obtain accurate data on sensitive topics. (Holbrook and Krosnick -2010) review 48 studies using list experiments, and found that they are -effective at decreasing social desirability bias. Comparing studies that -use list experiments with studies that do not, they find that reporting -rates of sensitive items are higher in studies using list experiments. -It is therefore not surprising that this approach has been applied to a -wide range of topics, such as sensitive political opinions (Frye et al. -2017; Blair, Imai, and Lyall 2014; Meng, Pan, and Yang 2017; Corstange -2009), over-reporting of voting (Holbrook and Krosnick 2010), risky -behaviours (LaBrie and Earleywine 2000) and SGBV (Bulte and Lensink -2019; Peterman et al. 2018). - -For the list experiment in this study, the female respondents were -randomly divided into two groups. This was done by the electronic survey -software (ODK), based on the randomly assigned ID codes. I follow (Imai -2011) in calling these groups Treatment and Control. An interviewer told -each respondent: ``I will read 4 \emph{(or 5)} problems that women can -experience. These can be sensitive problems. When you've experienced a -problem in the last year, please drop one of the balls to the ground. I -will not look at when you drop these balls, and only want to know the -total number of balls at the end. In the past 12 months, did you -experience... - -\begin{itemize} -\item - Lack of food; -\item - Lack of money; -\item - Theft; -\item - Sterility; and, -\item - Sexual Violence (Treatment group only)'' -\end{itemize} - -The interviewer only presented women randomly selected to be in the -treatment group with the fifth item (Sexual Violence). I selected the -four control items in such a way that it is unlikely women in the sample -face none, or all, of the issues. In such cases the interviewer knows -the respondent's answer to the sensitive issue ("no" if the total number -of issues is 0, "yes" if the total number is 5). Not all the control -items are non-sensitive, as the item ``sterility'' is a sensitive item. -This was done to reduce respondent suspicion when one sensitive items is -juxtaposed with a number of completely non-sensitive items (see Chuang -et al. (2019) for a more detailed explanation). After all items were -read, the interviewer asked the respondent to count the number of balls, -and report the number. The questionnaire was field tested prior to field -work to ensure that respondents understood these concepts. All -interviewers were thoroughly trained in the protocols, and the -electronic questionnaire was programmed in such a way to ensure -compliance to the protocol. - -A crucial assumption for the list experiment is that the randomization -ensures that Treatment and Control groups are identical. Table -\hyperref[tab:balance]{\[tab:balance\]} (Column 7) provides a comparison -of the two groups within the sample. The treatment and control group are -not perfectly balanced across some of the variables. However, an F-test -on the differences between treatment and control being jointly equal to -zero fails to reject the null-hypothesis that they are equal (p=0.20). -This suggests that the differences found are due to chance, rather than -any bias in the randomization procedure. - -While the indirect nature of list experiments prevents reporting bias, -this does come at a cost of efficiency in statistical analysis. The -incidence is easily computed by subtracting the mean of issues faced in -the control group from the mean number of issues in the treatment group. -This means that sample sizes have to be far larger for list experiments -than for direct questions. - -In a regression framework, the incidence would be estimated as follows -(Holbrook and Krosnick 2010): - -\[\label{eq:basic} -NumIssues_i = \beta_0 + \beta_1 Treatment_i + \epsilon_i\] - -Where \(NumIssues_i\) is the number of issues experienced by respondent -\(i\), and \(Treatment_i\) is her treatment assignment. Coefficient -\(\beta_1\) yields the estimate for the incidence. To find correlates of -SGBV, equation \hyperref[eq:basic]{\[eq:basic\]} can be augmented using -interaction terms as follows: \[\label{eq:interaction} -NumIssues_i = \beta_0 + \beta_1 Treatment_i + \beta_2 X_i + \beta_3 Treatment_i X_i + \epsilon_i\] - -Where \(X_i\) is an explanatory variable and coefficient \(\beta_3\) -gives the estimate for the additional incidence of SGBV associated with -a unit increase of \(X\). This can be easily modified to allow for more -variables. Again, this is much less efficient than when using direct -questioning. By using more sophisticated methods proposed by Imai (2011) -(and implemented by (Tsai 2019) in Stata), more efficient estimates can -be obtained. - -lCCCCCCC - -\&(1)\&(2)\&(3)\&(4)\&(5)\&(6)\&(7) \&\&\&\&(4)-(6) -\&N\&Mean\&N\&Mean\&N\&Mean\& Number of reported -issues\&593\&2.49\&291\&2.65\&302\&2.34\&0.30*** -\&\&(0.94)\&\&(1.03)\&\&(0.83)\& Conflict pre-2012: property -lost\&530\&0.77\&264\&0.79\&266\&0.75\&0.04 -\&\&(0.42)\&\&(0.41)\&\&(0.43)\& Conflict pre-2012: HH member -killed\&530\&0.49\&264\&0.51\&266\&0.48\&0.03 -\&\&(0.50)\&\&(0.50)\&\&(0.50)\& Conflict 2013--2014: Viol. against -civilians\&496\&6.73\&239\&6.68\&257\&6.77\&--0.08 -\&\&(4.69)\&\&(4.70)\&\&(4.69)\& Family MR had more -land\&450\&0.33\&224\&0.33\&226\&0.33\&--0.00 -\&\&(0.47)\&\&(0.47)\&\&(0.47)\& Family FR had more -land\&450\&0.21\&224\&0.22\&226\&0.19\&0.02 -\&\&(0.41)\&\&(0.41)\&\&(0.40)\& Bargaining: choice Female -Respondent\&593\&3.58\&291\&3.59\&302\&3.56\&0.04 -\&\&(2.06)\&\&(2.08)\&\&(2.05)\& Barganing: choice Male -Respondent\&184\&3.45\&97\&3.49\&87\&3.40\&0.09 -\&\&(2.14)\&\&(2.12)\&\&(2.18)\& Bargaining: closer to -MR\&184\&0.40\&97\&0.37\&87\&0.44\&--0.07 -\&\&(0.49)\&\&(0.49)\&\&(0.50)\& Bargaining: closer to -FR\&184\&0.27\&97\&0.32\&87\&0.21\&0.11* -\&\&(0.44)\&\&(0.47)\&\&(0.41)\& Age of -FR\&593\&41.09\&291\&40.49\&302\&41.67\&--1.17 -\&\&(14.01)\&\&(14.06)\&\&(13.96)\& Age of -MR\&449\&45.67\&224\&44.48\&225\&46.85\&--2.37** -\&\&(13.80)\&\&(13.09)\&\&(14.40)\& HH Head -Female\&593\&0.26\&291\&0.24\&302\&0.27\&--0.03 -\&\&(0.44)\&\&(0.43)\&\&(0.45)\& FR completed primary -education\&593\&0.25\&291\&0.26\&302\&0.25\&0.01 -\&\&(0.44)\&\&(0.44)\&\&(0.43)\& FR completed secondary -education\&593\&0.03\&291\&0.02\&302\&0.04\&--0.02 -\&\&(0.17)\&\&(0.13)\&\&(0.20)\& MR completed primary -education\&449\&0.63\&224\&0.63\&225\&0.63\&0.01 -\&\&(0.48)\&\&(0.48)\&\&(0.48)\& MR completed secondary -education\&449\&0.20\&224\&0.19\&225\&0.20\&--0.01 -\&\&(0.40)\&\&(0.39)\&\&(0.40)\& Household has a tin -roof\&593\&0.61\&291\&0.62\&302\&0.59\&0.03 -\&\&(0.49)\&\&(0.49)\&\&(0.49)\& Household owns -livestock\&593\&0.49\&291\&0.51\&302\&0.47\&0.03 -\&\&(0.50)\&\&(0.50)\&\&(0.50)\& -territory==Uvira\&593\&0.24\&291\&0.24\&302\&0.23\&0.01 -\&\&(0.43)\&\&(0.43)\&\&(0.42)\& -territory==Fizi\&593\&0.63\&291\&0.65\&302\&0.62\&0.03 -\&\&(0.48)\&\&(0.48)\&\&(0.49)\& Project -Beneficary\&593\&0.50\&291\&0.49\&302\&0.50\&--0.01 -\&\&(0.50)\&\&(0.50)\&\&(0.50)\& - -FR = Female Respondent; MR = Male Respondent; Standard Deviations in -parentheses; *p \(<\) 0.1,**p \(<\) 0.05,***p \(<\) 0.01 - -\section*{Results}\label{results} -\addcontentsline{toc}{section}{Results} - -Comparison of means of issues faced: treatment vs. control. - -In this section, I will first compare the results of the list experiment -in the whole sample, then in different sub-groups. I then present -results from a full multivariate regression that aims to minimize -potential bias caused by confounding variables. - -In the full sample, the difference between the group who were presented -with only four issues (the Control group) and the group who were -presented four issues plus SGBV (the Treatment group) is the estimate of -the incidence of SGBV. The average number of issues reported by the -control group is , while the number if issues reported by the treatment -group is (see Figure \hyperref[fig:meancompare_overall]{1}). The -difference of implies that the incidence of SGBV is \% in this sample. -The p-value for a t-test on this difference is . This estimate appears -substantially higher than previous estimates. These previous estimates -(e.g. Peterson et al. 2018; Stark et al. 2017; Johnson et al. 2010) -arrive at a similar rate of victimization, but over the life of the -respondent, whereas here we only consider victimization the past twelve -months. A higher incidence is expected, since the sample is non random, -drawing mostly from vulnerable rural households. - -\subsection*{Conflict}\label{conflict} -\addcontentsline{toc}{subsection}{Conflict} - -Comparison of means of issues faced across conflict exposure. - -With respect to conflict, I distinguish between recent conflict (as -indicated by ACLED events that happened within the 12 months before the -list experiment) and historic conflict (1996-2012). Historic conflict -can only have had an indirect effect on SGBV, e.g.~through changed -norms, as the list experiment only covers SGBV events within the past 12 -months. Recent conflict can have a direct effect through perpetration -during the conflict event. - -I first analyse victimization patterns by comparing sub-groups of the -respondents, based on one variable at a time. A full, multivariate -analysis will follow. With respect to conflict, I consider three ways of -splitting the sample in sub-groups: (i) respondents who live in -households that indicated (or not) in 2012 to have suffered loss of (or -damage to) property, including agricultural fields, due to conflict; -(ii) whether the respondent's household indicated in 2012 to have lost -any household members or family as a consequence of the conflict (or -not); and (iii) whether number of instances of violence against -civilians in ACLED data within a 10km radius during the past twelve -months was higher than the number of instances for the median household -(nor not) \footnote{The results presented here are robust to using - number of battles or number of fatalities rather than the instances of - violence against civilians; using 5,15,20, 25 or 30km as a radius; and - using a continuous variable, rather than a binary variable.}. - -Conflict exposure was high in the sample (see Table -\hyperref[tab:balance]{\[tab:balance\]}): \% of the respondents reported -having lost property due to conflict between 1996 and 2012. \% of the -respondents reported the loss of a family or household member. Again, -conflict exposure was high, even when limiting the time-span to one year -prior to the data collection. The mean number of violent conflicts -within a 10km radius was . This exposure differs across the territories -(Table \hyperref[tab:conflict_by_terr]{4}). While respondents in all -territories were greatly affected by conflict prior to 2012, those in -Fizi were hit harder. In the 12 months before the survey however, Uvira -was in the midst in an outbreak of violence, related to conflicts -surrounding the succession of traditional rulers in the chefferies of -Bafuliiro and Plaine de la Ruzizi. In fact, weeks before data collection -in 2014 took place, 30 civilians were killed in Mutarule, a village in -the Plaine, but not in my sample. This difference in recent and historic -conflict patterns means that households with conflict exposure pre-2012 -are not more likely to be victimized in 2013-2014 (see also Table -\hyperref[tab:determinants_regression]{\[tab:determinants_regression\]}). -Associations between pre-2012 violence and SGBV will thus not be the -result of re-targeting of the same households. - -\phantomsection\label{tab:conflict_by_terr} -\begin{center}\rule{0.5\linewidth}{0.5pt}\end{center} - -\begin{verbatim} - Kabare/Bagira Uvira Fizi Total -\end{verbatim} - -Conflict pre-2012: property lost 0.440 0.700 0.847 0.770 (0.501) (0.460) -(0.360) (0.421) Conflict pre-2012: HH member killed 0.120 0.393 0.588 -0.492 (0.328) (0.490) (0.493) (0.500) Conflict 2013-2014: Viol. against -civilians 7.289 10.05 4.892 6.726 (1.797) (2.752) (5.052) (4.689) ---------------------------------------------- --------------- --------- ---------- --------- - -: Conflict exposure by territory - -The results of the sub-group analysis is displayed graphically in Figure -\hyperref[fig:meancompare_conf]{2}. From the top two panels, it can be -seen that the difference between treatment and control is greater among -conflict-victimized respondents than among non-conflict victimized -respondents. The size of these differences is listed in Table -\hyperref[tab:meandiff_conf]{\[tab:meandiff_conf\]}. Among those that -indicated not having lost property, the difference in number of issues -faced between treatment and control is 0.19, implying a SGBV -victimization rate of 19\%. The difference between Treatment and Control -among respondents who did lose property was 0.38. The difference in the -differences between these groups of 0.19 issues (this corresponds to -coefficient \(\beta_3\) in equation -\hyperref[eq:interaction]{\[eq:interaction\]} above) is not -statistically significant. When splitting the sample by households -indicating having lost a family or household member to conflict before -2012, the difference-in-difference estimate is 0.37, indicating that -incidence of SGBV among respondents who lost family due to conflict is -37 percentage points higher than among those who have not. This effect -is significant at the 5\% level. Note that the SGBV could not have -happened during the same time as the conflict event(s): the SGBV -happened twelve months before the interview in 2014, while the conflict -events happened before 2012. - -When looking at more recent exposure to conflict, no clear patterns -emerge (see bottom panel of Figure \hyperref[fig:meancompare_conf]{2}). -SGBV incidence among women who have more instances of violence near them -than the median is 7 percentage points lower than women who do not -(bottom row of Table -\hyperref[tab:meandiff_conf]{\[tab:meandiff_conf\]}). However, this is -not statistically significant. I thus find no evidence of large-scale -direct perpetration of SGBV by armed groups in the one year before data -collection, but also no evidence of indirect effects of recent conflict. - -The fact that conflict before 2012 correlates with SGBV, but recent -conflict does not, points at a more complex relationship between -conflict and SGBV than a simple direct effect due to perpetration by -armed groups. It is more likely that violence has an indirect effect -through changed norms. The fact that recent conflict seems not to have -an indirect effect either, may mean that this change of norms takes -time, or that the nature of recent conflict is different from historic -conflict. - -n0 = , n1= , label0 = , label1 = , varlabel=,meancontrol0 = , meantreat0 -= ,stardiff0=, sediff0=, meancontrol1 = , meantreat1 = , stardiff1=, -sediff1=, stardd = , sedd= \& \& \& \& \&\\ -\& \& \& \& \&\\ -\& \& \& \& \&\\ -Diff in Diff \& \& \& \& \& - -Robust Standard errors reported. - -p \(<\) 0.1, **, p \(<\) 0.05, *** p \(<\) 0.01 - -\subsection*{Intra-household bargaining -position}\label{intra-household-bargaining-position} -\addcontentsline{toc}{subsection}{Intra-household bargaining position} - -\begin{figure} -\centering -\includegraphics{figures/meancompare_mar.eps} -\caption{Comparison of means of issues faced by pre-marriage -status.}\label{fig:meancompare_mar} -\end{figure} - -I then create sub-groups based on the intra-household bargaining -position of the respondents. I compare women across two variables. -First, I compare women across the relative status of the partners at the -time of marriage, by using family land-holdings as a proxy for status. -The 2014 survey contained a section on the marriage of the (spouse of -the) household head. In this section, respondents were asked whose -family owned more land, prior to the marriage: the wife's, the -husband's, or whether they had equal land. This choice of proxy was made -in consultation with local partners (including NGOs and universities), -and based on the importance of agriculture in the area. In \% of the -cases, the husband's family had more land, in \% of the cases the wife's -family did. Note that only households responded to this question, as -some refused to give a definite answer (Table -\hyperref[tab:balance]{\[tab:balance\]}). - -The second intra-household aspect I explore is derived from the results -of the bargaining game played with couples during the 2014 survey. I -create three groups, based on whether the joint decision is closer to -the husband's decision, to the wife's, or if the distance is equal. The -mean choice of the Female Respondents in the sample was ; the Male -Respondents were slightly more risk-averse: their mean choice was . In -\% of the cases, the couple decision was closest to the Male -Respondent's choice. In \% it was closer to the Female Respondent's. -Note that the size of the sample here is smaller than for the other -variables presented, as it was not always possible to have both the Male -and Female Respondent present at the same time for the interview. - -Figure \hyperref[fig:meancompare_mar]{3} displays the results from the -sub-group analysis. Overall, the difference between treatment and -control is larger for the sub-groups of respondents with a worse -intra-household bargaining position, indicating that the incidence of -SGBV is higher among these respondents. As suggested by the large size -of the 95\% confidence intervals, some of these sub-groups are small. In -Table \hyperref[tab:meadifftab_mar]{\[tab:meadifftab_mar\]}, these -differences are tabulated, including the sizes of the sub-groups. -However, the variable definitions are slightly different, due to the -difficulties in interpreting difference-in-differences between three -sub-groups. For each variable, two comparisons are tabulated: one, -comparing households where the female respondents had the better -bargaining position with the two other sub-groups, and one comparing -households where the male respondent had the better bargaining position -with the two other sub-groups. Female respondent in households where the -family of the husband had the most land prior to marriage were victims -of SGBV in 50\% of the cases, while 16\% of the other respondents were. -The difference of 33 percentage points is statistically significant at -the 10\% level. In the other comparison for the same variable, the -difference is even larger, but not statistically significant; perhaps -due to the low number of women with more pre-marital status than their -husbands. The differences when split by results from the bargaining game -are larger still: 57 or 61 percentage points, depending on the groups -used. - -While these results may suggest that IPV is an important driver of SGBV, -the fact that I have no information on perpetrators means that this is -not certain. - -n0 = , n1= , label0 = , label1 = , varlabel=,meancontrol0 = , meantreat0 -= ,stardiff0=, sediff0=, meancontrol1 = , meantreat1 = , stardiff1=, -sediff1=, stardd = , sedd= \& \& \& \& \&\\ -\& \& \& \& \&\\ -\& \& \& \& \&\\ -Diff in Diff \& \& \& \& \& - -Robust Standard errors reported. - -p \(<\) 0.1, **, p \(<\) 0.05, *** p \(<\) 0.01 - -\subsection*{Multivariate Regression -analysis}\label{multivariate-regression-analysis} -\addcontentsline{toc}{subsection}{Multivariate Regression analysis} - -In the preceding sections, I examined univariate relations between -variables of interest and the incidence of SGBV. However, such analysis -may suffer from omitted variables and spurious correlations. Here I move -to a richer specification, in order to prevent such biases, and assess -the relative importance of each driver. I expand equation -\hyperref[eq:interaction]{\[eq:interaction\]} to simultaneously include -indicators for conflict and intra-household bargaining position. To -reduce the risk of multi-collinearity, I do not include the full set of -variables discussed above, but select one indicator for each, guided by -the results obtained above. A key criterion for selection is the number -of respondents for each indicator. The analysis of list experiments -suffers from rapid loss of power due to the indirect nature of the -analysis. To mitigate this, indicators that are available for large -groups of respondents were selected. For conflict, I include both the -indicator for household member killed before 2012 (as an indicator for -historic conflict) and violence against civilians from the ACLED data -(as an indicator for recent conflict); and for intra-household -bargaining position a dummy for the husband's family having the most -land. I use the KICT Stata package developed by Tsai (2019) to estimate -these models. Interpretation of the coefficients is the same as equation -\hyperref[eq:interaction]{\[eq:interaction\]}, but estimation is more -efficient. - -In order to reduce missing variable bias, I include a set of controls -that likely (co-) determine SGBV and the right-hand side variables -listed above. A full analysis of these determinants is provided in the -Appendix, Table -\hyperref[tab:determinants_regression]{\[tab:determinants_regression\]}. -In addition, I include variables that determine sample selection, as -displayed in Table \hyperref[tab:sample_selection]{6}. In particular, I -include the age of the Female Respondent; indicators for the education -of the Male and Female respondents; asset holdings of the household, -including livestock and a tin roof; territory dummies; and an indicator -for being in the treatment group of any of the projects under evaluation -for the survey. - -In Table \hyperref[tab:results_regression]{5} I display the results of -these regressions. In columns 1-3 I rerun the univariate models from -above. Results are the same as before: both conflict history and -intra-household bargaining are associated with increased incidence of -SGBV. I In column 4 I present the full model. I find that women in a -marriage where their husband's family had more land before the marriage, -are percentage points more likely than other women to be victim of SGBV. -Note that the pre-marriage status of women within the household is -uncorrelated to conflict (see Table -\hyperref[tab:determinants_regression]{\[tab:determinants_regression\]}). -Women who live in households that lost a family or household member due -to conflict prior to 2012 are percentage points more likely to be -victimized by SGBV than other women. Of note is also the negative -associated of the Female Respondent having a secondary education: in -this linear model, women with secondary education are percentage points -less likely to be victimized. The fact that the absolute value of this -coefficient is higher than 1 is due to the fact that linear models do -not constrain predictions of probabilities between 0 and 1. - -The finding that conflict history is associated with an increase in -SGBV, while recent conflict is not, points to the indirect relationship -between conflict and SGBV, where conflict may affect SGBV rates not -through perpetration by armed groups, but by an increase in IPV. The -notion that IPV is a major driver of SGBV is reinforced by the fact that -both intra-household bargaining position and secondary education are -negatively associated with SGBV. This suggests that the position of -women is important in protecting them from human rights violations. - -Caution should be taken with this interpretation, as results presented -here are not necessarily causal: women with higher education may differ -from other women in non-observable ways, and face lower victimization -because of that, rather than education. Furthermore, no data exists on -the perpetrators of the violence. The method of a list experiment does -not allow for follow-up questions to victimized women, as the -interviewer cannot know who to ask these follow up questions to. - -\phantomsection\label{tab:results_regression} -\begin{longtable}[]{@{}lcccc@{}} -\caption{Multivariate regression Results}\tabularnewline -\toprule\noalign{} -& \(1\) & \(2\) & \(3\) & \(4\) \\ -\midrule\noalign{} -\endfirsthead -\toprule\noalign{} -& \(1\) & \(2\) & \(3\) & \(4\) \\ -\midrule\noalign{} -\endhead -\bottomrule\noalign{} -\endlastfoot -Family MR had more land & 0.419** & & & 0.451* \\ -& (0.204) & & & (0.240) \\ -Conflict pre-2012: HH member killed & & 0.409** & & 0.374** \\ -& & (0.182) & & (0.179) \\ -Conflict 2013-2014: Viol. against civilians & & & 0.0120 & 0.0147 \\ -& & & (0.0224) & (0.0230) \\ -FR empowerment attitudes & & & & 0.000101 \\ -& & & & (0.0199) \\ -Age of FR & 0.00843 & 0.00684 & 0.00563 & 0.0110 \\ -& (0.0161) & (0.0182) & (0.0186) & (0.0207) \\ -Age of MR & -0.0122 & -0.00929 & -0.00778 & -0.0111 \\ -& (0.0149) & (0.0162) & (0.0174) & (0.0189) \\ -HH Head Female & 0.00951 & -0.0766 & 0.257 & 0.421 \\ -& (0.445) & (0.525) & (0.318) & (0.410) \\ -FR completed secondary education & -1.111*** & -1.347*** & -1.034*** & --1.249*** \\ -& (0.320) & (0.351) & (0.336) & (0.332) \\ -MR completed primary education & -0.0390 & -0.0655 & -0.193 & -0.263 \\ -& (0.166) & (0.178) & (0.181) & (0.177) \\ -Household has a tin roof & 0.292 & 0.279 & 0.184 & 0.214 \\ -& (0.203) & (0.221) & (0.234) & (0.233) \\ -Household owns livestock & -0.0455 & -0.00342 & -0.142 & -0.193 \\ -& (0.160) & (0.181) & (0.177) & (0.186) \\ -territory==Uvira & 0.418 & 0.202 & 0.438 & 0.205 \\ -& (0.264) & (0.342) & (0.288) & (0.360) \\ -territory==Fizi & 0.511* & 0.232 & 0.504* & 0.191 \\ -& (0.292) & (0.365) & (0.302) & (0.379) \\ -Project Beneficary & 0.0542 & 0.0121 & 0.0382 & 0.0732 \\ -& (0.163) & (0.177) & (0.157) & (0.167) \\ -Constant & -0.162 & -0.00267 & -0.00290 & -0.111 \\ -& (0.483) & (0.506) & (0.491) & (0.614) \\ -Observations & 449 & 402 & 379 & 350 \\ -\end{longtable} - -FR = Male Respondent; MR = Female Respondent - -Standard errors clustered at the village level; * p \(<\) 0.1, **, p -\(<\) 0.05, *** p \(<\) 0.01 - -\section*{Conclusion}\label{conclusion} -\addcontentsline{toc}{section}{Conclusion} - -In this paper, I analysed the results from a list experiment, in order -to identify potential drivers of SGBV in Eastern Congo. Prevalence of -SGBV is high in Congo, however little is known about the victims, and -the drivers of victimization. In order to address this, I combined the -results from the list experiment with rich data, including a household -survey, a bargaining game, and conflict data. - -The incidence rates I find are very high: 30\% of the women in the -sample report having been the victim of SGBV in the past twelve months. -Most data collected on lifetime victimization arrives at similar rates, -suggesting that this estimate for a one-year window is high. The rate -found here may thus not be nationally, or regionally, representative. -This is likely due to the fact that women in the sample were recruited -among beneficiaries and potential beneficiaries of programs aimed at -assisting the most vulnerable women and households. It is to be expected -that incidence rates in this group are higher than for other groups. In -fact, I find that secondary schooling rates among women in my sample is -lower than the national or provincial average, and that incidence of -SGBV among women who have attended secondary school are significantly -lower than among other women. - -When examining the backgrounds of the victims, I find that they are -likely to be married to higher-status men, have low intra-household -bargaining power, and have been exposed to violent conflict to the -extent where they have lost family or household members before 2012 (two -years before the list experiment). When comparing these effects in one -analysis, I find that the effect of intra-household dynamics is larger -than the effect of conflict. This contrasts with popular frames where -the conflict is seen as the primary driver of SGBV, but is in line with -previous literature suggesting that intimate partners are more likely -perpetrators of SGBV than members of armed groups (see e.g. Peterman, -Palermo, and Bredenkamp 2011). - -Taken together, these finding imply that human rights violations do not -end when the conflict ends. The disruption of social norms may cause -women (and perhaps men, but the present data set does not cover them) to -suffer from violence long after the last shot has been fired. A focus of -rape as a ``weapon of war'' may thus be too narrow to address these -violations. This is not to say there direct perpetration of SGBV by -armed forces is not a problem in Congo. There is ample proof that -large-scale violations have been committed by armed forces, especially -historically. The conflict has undergone changes throughout the years, -and with it the kinds of human rights violations perpetrated. The -massacre in Mutarule in the weeks before data collections did see 30 -innocent civilians murdered, but there are no reports of rape. -Furthermore, focusing efforts to assist women on the victims from such -attacks risks missing women victimized in their homes, far away from any -fighting. Structural changes encouraging women's education and tangibly -raising their status are needed to protect these women as well. - -There are three large caveats with these findings: (i) causal -interpretation is difficult due to the cross-sectional nature of the -data; (ii) little analysis could be done on the perpetrators of the -violence, as indirect questioning precludes probing into this. More -research is needed to to address these important issues; and (iii) I did -not collect data on the victimization of men. More research is needed to -address these. - -\section*{Appendix}\label{appendix} -\addcontentsline{toc}{section}{Appendix} - -\phantomsection\label{tab:sample_selection} -\begin{longtable}[]{@{} - >{\raggedright\arraybackslash}p{(\columnwidth - 6\tabcolsep) * \real{0.4321}} - >{\centering\arraybackslash}p{(\columnwidth - 6\tabcolsep) * \real{0.1852}} - >{\centering\arraybackslash}p{(\columnwidth - 6\tabcolsep) * \real{0.1728}} - >{\centering\arraybackslash}p{(\columnwidth - 6\tabcolsep) * \real{0.1852}}@{}} -\caption{Sample selection for the Gender Module}\tabularnewline -\toprule\noalign{} -\endfirsthead -\endhead -\bottomrule\noalign{} -\endlastfoot -& \(1\) Wife & \(2\) Husband & \(3\) Couple \\ -Age of FR & -0.0305** (0.0126) 0.0199* (0.0118) 0.171 (0.689) 0.120 -(0.224) 0.484 (0.587) -0.0233 (0.189) 0.228 (0.213) 0.307* (0.181) --0.564*** (0.204) -0.215 (0.297) -0.0496 (0.263) 0.490** (0.195) 0.726* -(0.430) 717 & 0.0244** (0.0112) -0.00864 (0.0110) -1.142* (0.602) -0.197 -(0.245) -0.285 (0.399) 0.226 (0.218) -0.208 (0.215) 0.0468 (0.171) -0.695*** (0.189) 1.165*** (0.261) 0.312 (0.217) -0.450** (0.182) -0.700 -(0.434) 717 & -0.0124 (0.0127) 0.0176 (0.0130) \\ -\end{longtable} - -Standard errors clustered at the village level - -p \(<\) 0.1, **, p \(<\) 0.05, *** p \(<\) 0.01 - -l*3c \&\&\&\\ -\&\&\&\\ -Family MR had more land\& \& 0.0756 \& 0.679\\ -\& \& (0.0554) \& (0.476)\\ -Conflict pre-2012: HH member killed\& 0.0799 \& \& -1.494***\\ -\& (0.0588) \& \& (0.476)\\ -Conflict 2013-2014: Viol. against civilians\& 0.00885 \& -0.0184***\&\\ -\& (0.00627) \& (0.00477) \&\\ -Age of FR \& -0.00431 \& 0.00219 \& -0.0264\\ -\& (0.00602) \& (0.00448) \& (0.0349)\\ -Age of MR \& 0.00355 \& -0.00146 \& 0.0281\\ -\& (0.00435) \& (0.00390) \& (0.0320)\\ -FR completed primary education\& 0.0542 \& -0.109 \& -0.0536\\ -\& (0.0695) \& (0.0656) \& (0.672)\\ -FR completed secondary education\& -0.0147 \& 0.474***\& 0.602\\ -\& (0.129) \& (0.147) \& (1.417)\\ -MR completed primary education\& 0.0834 \& 0.0314 \& -0.347\\ -\& (0.0612) \& (0.0622) \& (0.512)\\ -MR completed secondary education\& -0.0523 \& 0.0408 \& 0.239\\ -\& (0.0703) \& (0.0764) \& (0.603)\\ -Household has a tin roof\& -0.0267 \& -0.114** \& 0.563\\ -\& (0.0537) \& (0.0509) \& (0.539)\\ -Household owns livestock\& 0.114** \& 0.0536 \& -0.235\\ -\& (0.0554) \& (0.0526) \& (0.490)\\ -territory==Uvira \& -0.0556 \& 0.295***\& 3.566***\\ -\& (0.0589) \& (0.0904) \& (1.104)\\ -territory==Fizi \& -0.0166 \& 0.404***\& -1.069\\ -\& (0.0560) \& (0.0763) \& (1.487)\\ -Project Beneficary \& -0.0854* \& 0.0492 \& 0.198\\ -\& (0.0461) \& (0.0450) \& (1.136)\\ -Constant \& 0.212 \& 0.247** \& 6.599***\\ -\& (0.133) \& (0.106) \& (1.420)\\ -Observations \& 350 \& 350 \& 350\\ - -Standard errors clustered at the village level - -p \(<\) 0.1, **, p \(<\) 0.05, *** p \(<\) 0.01 - -\phantomsection\label{refs} -\begin{CSLReferences}{1}{0} -\bibitem[\citeproctext]{ref-Albutt2017} -Albutt, Katherine, Jocelyn Kelly, Justin Kabanga, and Michael VanRooyen. -2017. {``Stigmatisation and Rejection of Survivors of Sexual Violence in -Eastern {Democratic Republic} of the {Congo}.''} \emph{Disasters} 41 -(2): 211--27. \url{https://doi.org/10.1111/disa.12202}. - -\bibitem[\citeproctext]{ref-Asadullah2018} -Asadullah, M. Niaz, and Antonio Savoia. 2018. {``Poverty Reduction -During 1990{\textendash}2013: {Did} Millennium Development Goals -Adoption and State Capacity Matter?''} \emph{World Development} 105: -70--82. \url{https://doi.org/10.1016/j.worlddev.2017.12.010}. - -\bibitem[\citeproctext]{ref-Autesserre2012a} -Autesserre, Séverine. 2012. {``Dangerous Tales: {Dominant} Narratives on -the {Congo} and Their Unintended Consequences.''} \emph{African Affairs} -111 (443): 202--22. \url{https://doi.org/10.1093/afraf/adr080}. - -\bibitem[\citeproctext]{ref-Baaz2013} -Baaz Eriksson, Maria, and Maria Stern. 2013. \emph{Sexual Violence as a -Weapon of War? {Perceptions}, Prescriptions, Problems in the {Congo} and -Beyond}. {London, UK}: {Zed Books}. - -\bibitem[\citeproctext]{ref-Babalola2014} -Babalola, Stella, Amrita Gill-Bailey, and Mathurin Dodo. 2014. -{``Prevalence and {Correlates} of {Experience} of {Physical} and {Sexual -Intimate Partner Violence} Among {Men} and {Women} in {Eastern DRC}.''} -\emph{Universal Journal of Public Health} 2 (1): 25--33. -\url{https://doi.org/10.13189/ujph.2014.020104}. - -\bibitem[\citeproctext]{ref-Bartels2013} -Bartels, Susan, Jocelyn Kelly, Jennifer Scott, Jennifer Leaning, Denis -Mukwege, Nina Joyce, and Michael VanRooyen. 2013. {``Militarized {Sexual -Violence} in {South Kivu}, {Democratic Republic} of {Congo}.''} -\emph{Journal of Interpersonal Violence} 28 (2): 340--58. -\url{https://doi.org/10.1177/0886260512454742}. - -\bibitem[\citeproctext]{ref-Bhattacharya} -Bhattacharya, Manasi, Arjun S. Bedi, and Amrita Chhachhi. 2009. -{``Marital {Violence} and {Women}'s {Employment} and {Property Status -Evcidence} from {North Indian Villages}.''} {IZA}. -\url{https://doi.org/10.1139/p87-134}. - -\bibitem[\citeproctext]{ref-Blair2012} -Blair, Graeme, and Kosuke Imai. 2012. {``Statistical {Analysis} of {List -Experiments}.''} \emph{Political Analysis} 20 (1): 47--77. -\url{https://doi.org/10.1093/pan/mpr048}. - -\bibitem[\citeproctext]{ref-Blair2014} -Blair, Graeme, Kosuke Imai, and Jason Lyall. 2014. {``Comparing and -Combining List and Endorsement Experiments: {Evidence} from -{Afghanistan}.''} \emph{American Journal of Political Science} 58 (4): -1043--63. \url{https://doi.org/10.1111/ajps.12086}. - -\bibitem[\citeproctext]{ref-Bueno2017} -Bueno, Cruz Caridad, and Errol A Henderson. 2017. {``Bargaining or -{Backlash} ? {Evidence} on {Intimate Partner Violence} from the -{Dominican Republic}.''} \emph{Feminist Economics} 23 (4): 90--116. -\url{https://doi.org/10.1080/13545701.2017.1292360}. - -\bibitem[\citeproctext]{ref-Bulte2019} -Bulte, Erwin H., and Robert Lensink. 2019. {``Women's Empowerment and -Domestic Abuse: {Experimental} Evidence from {Vietnam}.''} -\emph{European Economic Review}. -\url{https://doi.org/10.1016/j.euroecorev.2019.03.003}. - -\bibitem[\citeproctext]{ref-Chuang2019} -Chuang, Erica, Pascaline Dupas, Elise Huillery, and Juliette Seban. -2019. {``Sex, {Lies} and {Measurement}.''} - -\bibitem[\citeproctext]{ref-Corstange2009} -Corstange, Daniel. 2009. {``Sensitive Questions, Truthful Answers? -{Modeling} the List Experiment with {LISTIT}.''} \emph{Political -Analysis} 17 (1): 45--63. \url{https://doi.org/10.1093/pan/mpn013}. - -\bibitem[\citeproctext]{ref-Dave2010a} -Dave, Chetan, Catherine C. Eckel, Cathleen A. Johnson, and Christian -Rojas. 2010. {``Eliciting Risk Preferences: {When} Is Simple Better?''} -\emph{Journal of Risk and Uncertainty} 41 (3): 219--43. -\url{https://doi.org/10.1007/s11166-010-9103-z}. - -\bibitem[\citeproctext]{ref-Eckel2002} -Eckel, Catherine C., and Philip J. Grossman. 2002. {``Sex Differences -and Statistical Stereotyping in Attitudes Toward Financial Risk.''} -\emph{Evolution and Human Behavior} 23 (4): 281--95. -\url{https://doi.org/10.1016/S1090-5138(02)00097-1}. - -\bibitem[\citeproctext]{ref-Eswaran2011} -Eswaran, Mukesh, and Nisha Malhotra. 2011. {``Domestic Violence and -Women 's Autonomy in Developing Countries : Theory and Evidence.''} -\emph{Canadian Journal of Economics} 44 (4): 1222--63. -\url{https://doi.org/10.1111/j.1540-5982.2011.01673.x}. - -\bibitem[\citeproctext]{ref-Freedman2011} -Freedman, Jane. 2011. {``Explaining Sexual Violence and Gender -Inequalities in the {DRC}.''} \emph{Peace Review} 23 (2): 170--75. -\url{https://doi.org/10.1080/10402659.2011.571601}. - -\bibitem[\citeproctext]{ref-Frye2017} -Frye, Timothy, Scott Gehlbach, Kyle L. Marquardt, and Ora John Reuter. -2017. {``Is {Putin}'s Popularity Real?''} \emph{Post-Soviet Affairs} 33 -(1): 1--15. \url{https://doi.org/10.1080/1060586X.2016.1144334}. - -\bibitem[\citeproctext]{ref-Gibbs2020} -Gibbs, Andrew, Julienne Corboz, Esnat Chirwa, Carron Mann, Fazal Karim, -Mohammed Shafiq, Anna Mecagni, Charlotte Maxwell-Jones, Eva Noble, and -Rachel Jewkes. 2020. {``The Impacts of Combined Social and Economic -Empowerment Training on Intimate Partner Violence, Depression, Gender -Norms and Livelihoods Among Women: {An} Individually Randomised -Controlled Trial and Qualitative Study in {Afghanistan}.''} \emph{BMJ -Global Health} 5 (3): 1--17. -\url{https://doi.org/10.1136/bmjgh-2019-001946}. - -\bibitem[\citeproctext]{ref-HarvardHumanitarianInitiative2009} -Harvard Humanitarian Initiative. 2009. {``Characterizing {Sexual -Violence} in the {Democratic Republic} of the {Congo}: {Profiles} of -{Violence}, {Community Responses}, and {Implications} for the -{Protection} of {Women},''} 1--64. - -\bibitem[\citeproctext]{ref-Hidrobo2016} -Hidrobo, Melissa, Amber Peterman, and Lori Heise. 2016. {``The Effect of -Cash, Vouchers, and Food Transfers on Intimate Partner Violence: -{Evidence} from a Randomized Experiment in {Northern Ecuador}.''} -\emph{American Economic Journal: Applied Economics} 8 (3): 284--303. -\url{https://doi.org/10.1257/app.20150048}. - -\bibitem[\citeproctext]{ref-Hilhorst2018} -Hilhorst, Dorothea, and Nynke Douma. 2018. {``Beyond the Hype? {The} -Response to Sexual Violence in the {Democratic Republic} of the {Congo} -in 2011 and 2014.''} \emph{Disasters} 42 (S1): S79--98. -\url{https://doi.org/10.1111/disa.12270}. - -\bibitem[\citeproctext]{ref-Holbrook2010} -Holbrook, Allyson L., and Jon A. Krosnick. 2010. {``Social Desirability -Bias in Voter Turnout Reports: {Tests} Using the Item Count -Technique.''} \emph{Public Opinion Quarterly} 74 (1): 37--67. -\url{https://doi.org/10.1093/poq/nfp065}. - -\bibitem[\citeproctext]{ref-Holt2002} -Holt, Charles A, and Susan K Laury. 2002. {``Risk {Aversion} and -{Incentive Effects}.''} \emph{American Economic Review} 92 (5): -1644--55. - -\bibitem[\citeproctext]{ref-HRW2002} -Human Rights Watch. 2002. {``The War Within the War : Sexual Violence -Against Women and Girls in {Eastern Congo}.''} - -\bibitem[\citeproctext]{ref-HumanRightsWatch2009} ----------. 2009. \emph{Soldiers Who Rape, Commanders Who Condone: -{Sexual} Violence and Military Reform Int He {Democratic Republic} of -{Congo}}. \emph{Human Rights Watch Report 1-56432-510-5}. {New York}: -{Human Rights Watch}. - -\bibitem[\citeproctext]{ref-Imai2011} -Imai, Kosuke. 2011. {``Multivariate Regression Analysis for the Item -Count Technique.''} \emph{Journal of the American Statistical -Association} 106 (494): 407--16. -\url{https://doi.org/10.1198/jasa.2011.ap10415}. - -\bibitem[\citeproctext]{ref-Johnson2010} -Johnson, Kirsten, Jennifer Scott, Bigy Rughita, Michael Kisielewski, -Jana Asher, Ricardo Ong, and Lynn Lawry. 2010. {``Association of {Sexual -Violence} and {Human Rights Violations With Physical} and {Mental -Health} in {Territories} of the {Eastern Democratic Republic} of the -{Congo}.''} \emph{JAMA} 304 (5): 553--62. -\url{https://doi.org/10.1001/jama.2010.1086}. - -\bibitem[\citeproctext]{ref-Kelly2018} -Kelly, Jocelyn, Elizabeth Colantuoni, Courtland Robinson, and Michele R. -Decker. 2018. {``From the Battlefield to the Bedroom: {A} Multilevel -Analysis of the Links Between Political Conflict and Intimate Partner -Violence in {Liberia}.''} \emph{BMJ Global Health} 3 (2): 1--11. -\url{https://doi.org/10.1136/bmjgh-2017-000668}. - -\bibitem[\citeproctext]{ref-Kilonzo2009} -Kilonzo, Nduku, Njoki Ndung'u, Nerida Nthamburi, Caroline Ajema, Miriam -Taegtmeyer, Sally Theobald, and Rachel Tolhurst. 2009. {``Sexual -Violence Legislation in Sub-{Saharan Africa}: The Need for Strengthened -Medico-Legal Linkages.''} \emph{Reproductive Health Matters} 17 (34): -10--19. \url{https://doi.org/10.1016/S0968-8080(09)34485-7}. - -\bibitem[\citeproctext]{ref-Kirby2015} -Kirby, Paul. 2015. {``Ending Sexual Violence in Conflict: {The -Preventing Sexual Violence Initiative} and Its Critics.''} -\emph{International Affairs} 91 (3): 457--72. -\url{https://doi.org/10.1111/1468-2346.12283}. - -\bibitem[\citeproctext]{ref-Kohli2012} -Kohli, Anjalee, Maphie Tosha Makambo, Paul Ramazani, Isaya Zahiga, Biki -Mbika, Octave Safari, Richard Bachunguye, Janvier Mirindi, and Nancy -Glass. 2012. {``A {Congolese} Community-Based Health Program for -Survivors of Sexual Violence.''} \emph{Conflict and Health} 6 (1): 1--9. -\url{https://doi.org/10.1186/1752-1505-6-6}. - -\bibitem[\citeproctext]{ref-LaBrie2000} -LaBrie, Joseph W., and Mitchell Earleywine. 2000. {``Sexual Risk -Behaviors and Alcohol: {Higher} Base Rates Revealed Using the -Unmatched-Count Technique.''} \emph{Journal of Sex Research} 37 (4): -321--26. \url{https://doi.org/10.1080/00224490009552054}. - -\bibitem[\citeproctext]{ref-Martinsson2009} -Martinsson, Peter, Matthias Sutter, and Fredrik Carlsson. 2009. -{``Household {Decision Making} and the {Influence} of {Spouses} ' -{Income} , {Education} , and {Communist Party Membership} : {A Field -Experiment} in {Rural China}.''} {IZA}. - -\bibitem[\citeproctext]{ref-Meng2017} -Meng, Tianguang, Jennifer Pan, and Ping Yang. 2017. {``Conditional -{Receptivity} to {Citizen Participation}: {Evidence From} a {Survey -Experiment} in {China}.''} \emph{Comparative Political Studies} 50 (4): -399--433. \url{https://doi.org/10.1177/0010414014556212}. - -\bibitem[\citeproctext]{ref-DHSCongoReport} -MPSMRM, MSP, and ICF International. 2014. {``R{é}publique -{D{é}mocratique} Du {Congo Enqu{ê}te D{é}mographique} Et de {Sant{é}} -({EDS-RDC}) 2013-2014.''} {Minist{è}re du Plan et Suivi de la Mise en -{œ}uvre de la R{é}volution de la Modernit{é}}. -\url{https://doi.org/10.1787/aeo-2014-24-fr}. - -\bibitem[\citeproctext]{ref-Muller2019} -Müller, Catherine, and Jean Pierre Tranchant. 2019. {``Domestic -{Violence} and {Humanitarian Crises}: {Evidence} from the 2014 {Israeli -Military Operation} in {Gaza}.''} \emph{Violence Against Women} 25 (12): -1391--1416. \url{https://doi.org/10.1177/1077801218818377}. - -\bibitem[\citeproctext]{ref-Palermo2011} -Palermo, Tia, and Amber Peterman. 2011. {``Undercounting, Overcounting -and the Longevity of Flawed Estimates: {Statistics} on Sexual Violence -in Conflict.''} \emph{Bulletin of the World Health Organization} 89 -(12): 924--25. \url{https://doi.org/10.2471/BLT.11.089888}. - -\bibitem[\citeproctext]{ref-Peterman2011} -Peterman, Amber, Tia Palermo, and Caryn Bredenkamp. 2011. {``Estimates -and Determinants of Sexual Violence Against Women in the {Democratic -Republic} of {Congo}.''} \emph{American Journal of Public Health} 101 -(6): 1060--67. \url{https://doi.org/10.2105/AJPH.2010.300070}. - -\bibitem[\citeproctext]{ref-Peterman2018} -Peterman, Amber, Tia Palermo, Sudhanshu Handa, and David Seidenfeld. -2018. {``List Randomization for Soliciting Experience of Intimate -Partner Violence: {Application} to the Evaluation of {Zambia}'s -Unconditional Child Grant Program.''} \emph{Health Economics (United -Kingdom)} 27 (3): 622--28. \url{https://doi.org/10.1002/hec.3588}. - -\bibitem[\citeproctext]{ref-Peterson2018} -Peterson, Cora, Yang Liu, Marcie-jo Kresnow, Curtis Florence, Melissa T. -Merrick, Sarah DeGue, and Colby N. Lokey. 2018. {``Short-Term {Lost -Productivity} Per {Victim}: {Intimate Partner Violence}, {Sexual -Violence}, or {Stalking}.''} \emph{American Journal of Preventive -Medicine} 55 (1): 106--10. -\url{https://doi.org/10.1016/J.AMEPRE.2018.03.007}. - -\bibitem[\citeproctext]{ref-Porter2019} -Porter, Holly. 2019. {``Moral {Spaces} and {Sexual Transgression}: -{Understanding Rape} in {War} and {Post Conflict}.''} \emph{Development -and Change} 50 (4): 1009--32. \url{https://doi.org/10.1111/dech.12499}. - -\bibitem[\citeproctext]{ref-Post2002} -Post, Lori A., Nancy J. Mezey, Christopher Maxwell, and Wilma Novalés -Wibert. 2002. {``The {Rape Tax}: {Tangible} and {Intangible Costs} of -{Sexual Violence}.''} \emph{Journal of Interpersonal Violence} 17 (7): -773--82. \url{https://doi.org/10.1177/0886260502017007005}. - -\bibitem[\citeproctext]{ref-Raleigh2010} -Raleigh, Clionadh, Andrew Linke, Håvard Hegre, and Joakim Karlsen. 2010. -{``Introducing {ACLED}: {An} Armed Conflict Location and Event -Dataset.''} \emph{Journal of Peace Research} 47 (5): 651--60. -\url{https://doi.org/10.1177/0022343310378914}. - -\bibitem[\citeproctext]{ref-Saile2013} -Saile, Regina, Frank Neuner, Verena Ertl, and Claudia Catani. 2013. -{``Prevalence and Predictors of Partner Violence Against Women in the -Aftermath of War: {A} Survey Among Couples in {Northern Uganda}.''} -\emph{Social Science and Medicine} 86: 17--25. -\url{https://doi.org/10.1016/j.socscimed.2013.02.046}. - -\bibitem[\citeproctext]{ref-Samy2011} -Samy, Yiagadeesen, and David Carment. 2011. {``The {Millenium -Development Goals} and {Fragile States}: {Focusing} on {What Really -Matters}.''} \emph{The Fletcher Forum of World Affairs} 35 (1): 91--108. - -\bibitem[\citeproctext]{ref-Sniderman1991} -Sniderman, Paul M., Philip E. Tetlock, and Thomas Piazza. 1991. -{``Codebook for the 1991 {Race} and {Politics Survey}.''} {Berkeley}: -{University of California}. - -\bibitem[\citeproctext]{ref-Stark2017} -Stark, Lindsay, Marni Sommer, Kathryn Davis, Khudejha Asghar, Asham -Assazenew Baysa, Gizman Abdela, Sophie Tanner, and Kathryn Falb. 2017. -{``Disclosure Bias for Group Versus Individual Reporting of Violence -Amongst Conflict-Affected Adolescent Girls in {DRC} and {Ethiopia}.''} -\emph{PLoS ONE} 12 (4): 1--12. -\url{https://doi.org/10.1371/journal.pone.0174741}. - -\bibitem[\citeproctext]{ref-Steiner2009} -Steiner, Birthe, Marie T Benner, Egbert Sondorp, K Peter Schmitz, Ursula -Mesmer, and Sandrine Rosenberger. 2009. {``Sexual Violence in the -Protracted Conflict of {DRC} Programming for Rape Survivors in {South -Kivu}.''} \emph{Conflict and Health} 3 (1): 1--9. -\url{https://doi.org/10.1186/1752-1505-3-3}. - -\bibitem[\citeproctext]{ref-Tsai2019} -Tsai, Chi-lin. 2019. {``Statistical Analysis of the Item-Count Technique -Using {Stata}.''} \emph{The Stata Journal} 19 (2): 390--434. -\url{https://doi.org/10.1177/1536867X19854018}. - -\bibitem[\citeproctext]{ref-UnitedNations2015} -United Nations. 2015. {``The {Millennium Development Goals Report}.''} -\emph{United Nations}, 72. \url{https://doi.org/978-92-1-101320-7}. - -\bibitem[\citeproctext]{ref-Verwijen2016} -Verwijen, Judith. 2016. \emph{A {Microcosm} of {Militarization}: -{Conflict}, {Governance} and {Armed Mobilization} in {Uvira}, -{South-Kivu}}. {Nairobi, Kenya}: {Rift Valley Institute/Usalama -Project}. \url{https://doi.org/10.13140/RG.2.1.2995.9283}. - -\end{CSLReferences} - -\end{document} diff --git a/congogbv_paths.tex b/congogbv_paths.tex deleted file mode 100644 index a6e97d9..0000000 --- a/congogbv_paths.tex +++ /dev/null @@ -1,3 +0,0 @@ - -\newcommand{\congogbvTables}{tables} -\newcommand{\congogbvFigures}{figures} \ No newline at end of file diff --git a/congogbv_preamble.tex b/congogbv_preamble.tex deleted file mode 100644 index 8038bb8..0000000 --- a/congogbv_preamble.tex +++ /dev/null @@ -1,73 +0,0 @@ -\documentclass[10pt,a4paper,abstract=on]{scrartcl} %scrartcl from KOMA-script uses sans headers and looks less latexy than article -\usepackage[doublespacing]{setspace} -%\usepackage[capposition=top]{floatrow} %for notes under pics -%\usepackage[nomarkers]{endfloat} -\usepackage{natbib} -\usepackage[utf8]{inputenc} -%\usepackage{amsmath} -%\usepackage{amsfonts} -%\usepackage{amssymb} -\usepackage[left=2.5cm,right=2.5cm,top=3cm,bottom=3cm]{geometry} -\usepackage[pdftex, hidelinks]{hyperref} %disable if using float package -\usepackage{graphicx} -\usepackage{booktabs} -\usepackage[blocks]{authblk} %for affil -\usepackage{url} -\usepackage{tabularx} -%\usepackage{float} %allows the H positioning specifier; conflicts with hyperref -\usepackage{csvsimple} -\usepackage[nomessages]{fp} -\usepackage{xcolor} -\usepackage{enumitem} -\setlist{noitemsep} % \setlist{nosep} to reove separation around lists as well -\usepackage[flushleft]{threeparttable} -\usepackage{dcolumn} -\usepackage{lineno, blindtext} -\usepackage{pdflscape} -\usepackage[colorinlistoftodos,disable]{todonotes} %add disable option to hide notes -\setlength{\marginparwidth}{4cm} - -%citation cheat sheet: - %\ cite{key} Jones et al. (1990) - % \citet{key1,key2} Jones et al. (1990); Ladze (1986) - % \citet{key} Jones et al. (1990) - % \citet*{key} Jones, Baker, and Smith (1990) - % \citep{key} (Jones et al. 1990) - % \citep{key1,key2} (Jones et al. 1990; Ladze 1986) - % \citep*{key} (Jones, Baker, and Smith 1990) - % \citep[p.~99]{key} (Jones et al., 1990, p. 99) - % \citep[e.g.][]{key} (e.g. Jones et al., 1990) - % \citep[e.g.][p.~99]{key} (e.g. Jones et al., 1990, p. 99) - % \citeauthor{key} Jones et al. - % \citeauthor*{key} Jones, Baker, and Smith - % \citeyear{key} 1990 - % \citealt{key} Jones et al. 1990 - -%sizes cheat sheet: https://www.sascha-frank.com/latex-font-size.html -% \tiny -% \scriptsize -% \footnotesize -% \small -% \normalsize -% \large -% \Large -% \LARGE -% \huge -% \Huge - - -\begin{document} -\input{congogbv_paths} -\author{Koen Leuveld} - -\affil{Wageningen Economic Research \\ -koen.leuveld@wur.nl} - - - -\title{Sexual violence, conflict, and female empowerment} -\subtitle{Exploratory evidence from a list experiment in Eastern DR Congo} %needs scrartcl - -\maketitle - -\linenumbers diff --git a/congogbv_dataprep.do b/stata_do_files/congogbv_dataprep.do similarity index 100% rename from congogbv_dataprep.do rename to stata_do_files/congogbv_dataprep.do diff --git a/congogbv_dhs.do b/stata_do_files/congogbv_dhs.do similarity index 100% rename from congogbv_dhs.do rename to stata_do_files/congogbv_dhs.do diff --git a/congogbv_helpers.do b/stata_do_files/congogbv_helpers.do similarity index 100% rename from congogbv_helpers.do rename to stata_do_files/congogbv_helpers.do