Exploratory Analysis Paper

The outcome variable is the one in which we want to assess the difference between treated and control groups. As discussed in the data section our dependent variables correspond to the number of hours that children allocated to farm work, family business, domestic work, and total studying time. Separately, we fitted three models, one for each activity. However, we did not adjust a treatment effects model for domestic work since migration variables did not showed a significant association to this activity (Table 2). The treatment variable included in each model was selected on the basis of the results of an exploratory analysis conducted using OLS regression models. For models regressing on farm-work time and total studying time, the variable denoting treatment membership is the indicator of international migrant household no receiving remittances. In the case of the model on work time in a family business, membership in the treatment group was defined as those children in an international migrant household receiving remittances. Variables that we anticipate will determine the selection process (included as independent variables in eq. 2) are migration prevalence level in the community, household size and household head education.
2. Seemingly unrelated regression models of Hours Spent on Farm Work and on Family Business vs Hours studying: We employ …show more content…
Thus, we adjusted seemingly unrelated regression models (SUR), which are extensions of linear regression models that allow correlated errors between equations. SUR produces more efficient estimates than OLS by weighting the estimates by the covariance of the residuals from the individual regressions (Greene,