Randomized Controlled Trial:

Consider a RCT where the treatment involves providing cash subsidies to the household head of poor families to increase school enrolment of children in the household.

1. Describe a F-test that you would carry out to examine whether the treatment is uncorrelated with observable pre-treatment individual characteristics. [2 marks]
2. Describe another test that you would carry out to check whether randomization was successful. [2 marks]
3. If there is an issue with partial compliance, what is a potential solution to the problem? [2 marks]
4. Suppose you are not able to collect follow-up data from certain households following the treatment, and they end up being dropped from the data-set. What test would you carry out to check whether you should be concerned about attrition? [2 marks]
5. Suppose you are estimating the following regression model to measure the impact of the treatment on householdi, where school enrolment is a binary variable that takes on the value of 1 if a child is enrolled in school, and is 0 otherwise:

(Model 2) Is heteroskedasticity a possible concern in this regression model? [2 marks]

6.Suppose another researcher suggests that you should add gender of the household head as a control variable in the model, even though it's not correlated with treatment status. What do you expect the effect of including a 'gender' variable in model 2 on the coefficient and the standard error of the treatment variable? [2 marks]

7.Will you be able to apply a causal interpretation to the 'gender' variable? [1 marks]