Impact evaluation

This app has three tabs:

Presentation: A brief description of our company
Impact Evaluation methods: We have four estimators to perform impact evaluation under different structures of the datasets
Help: A brief user guide
Our impact evaluation methods are:

Dif-in-Dif: Average treatment effects in a panel dataset where it is possible to have treatment in different periods.
Dynamic: Dynamic forecasts to build counterfactuals using Bayesian structural models.
RDD: Regression Discontinuity Design.
2SLS: Two Stage Least Squares.

Dif-in-Dif

Our users can select Dif-in-Dif from the left radio button, and then upload the dataset. You can see a visualization of the dataset, and a menu to select the category of each variable (Outcome, treatment, control, time, other).
This menu help to build the equations that are needed to perform the estimation step. This is done clicking on “Construya formula”. In addition, the user should type which column in the dataset is associated with the time, the cross sectional unit, and the first treatment period.
Then, click on “GO!”, and the outcomes are displayed. There, you can find the average treatment effects by group and period, their standard error, aggregated treatment effects: simple, selective, dynamic and calendar.
 
In addition, you can see a figure with the dynamic treatment effects (event study)
 

Dynamic

Once this option is chosen, our users should upload the dataset, and type the initial and final pre-treatment dates, and the initial and final post-treatment dates. Then, click “GO!”, and the treatment effects are displayed:

RDD

This method requires uploading the dataset, and selecting the category of each variable:
You can also see a figure with the outcomes:
 
Then, we should click on “Construya formula”. It is also required to type the discontinuity point, and click on “GO!”. We can see the treatment effects, and their standard errors:

TSLS

The users should upload the dataset, and then select the category of each variable (Outcome, endogenous, control, instrument, other).
Then, click on “Construir formula”. There are two equations: the first is to perform ordinary least squares, and the second is to perform two stage least squares.
Click on “GO!” to have the estimates, standard errors, t tests and p values. In addition, there are some diagnostics associated with the two stage least squares estimator.