appendix1.tex

\section*{Appendix 1: Additional Figures and Tables}

%----------------------------------------------------------
\begin{figure}[!htbp] 
    \caption{Flow chart for treatment protocol}
    \centering
    %\captionsetup{singlelinecheck = false}
    %\caption*{\qquad \qquad (a)	Amount won as a	\qquad \qquad \qquad (b) Recovered a positive amount\\
    %\qquad \qquad percentage of amount asked}
    \includegraphics[width=1\textwidth]{Figures/A1_flowchart.png}
    \label{fig:A1_1}
\end{figure}
%----------------------------------------------------------
\begin{figure}[H] 
  \centering
  \caption{Lee bounds on treatment effects}
  \begin{minipage}[b]{0.4\textwidth}
    \includegraphics[width=\textwidth]{Figures/A1_fig5_1.png}
  \end{minipage}
  %\hfill
  \begin{minipage}[b]{0.4\textwidth}
    \includegraphics[width=\textwidth]{Figures/A1_fig5_2.png}
  \end{minipage}
  \label{fig:A1_fig5}
  \begin{figurenotes}
  Our survey measures of outcomes are subject to attrition. In the two weeks survey we have 367 attritors, this affects solved conflict (2w) and talked to lawyer. In the two months survey we have 540 attritors, which affects the variable sued. For the attritors in the two months survey who solved conflict after two weeks we impute the missing information from the 2w survey. This leaves 474 attritors for solved conflict(2m) in the information experiment and 400 in the lawyer inducement experiment. We implement lee bounds to take this attrition into account. This figure displays the bounds for the treatment effect (i.e. with respect to the control group T1 for the information experiment and A for the lawyer inducement experiment) along with their 95 percent confidence intervals. For each outcome we report 3 treatment effects. All significant results in Table 2 for T2 are robust to lee bounds. For T3 solved conflict at 2w and sued is not robust. For treatment B the effect on talked two lawyer is robust.
  \end{figurenotes}
\end{figure}
%----------------------------------------------------------
%Faltan tablas: attrition y find_lawyer