Skip to contents

In this version considers only the case with two modes. Could be generalised. Requires the package MatchIt.

Usage

estim_counterfactuals(Yobs, modes, X, ...)

Arguments

Yobs

vector of the observed outcomes. (numeric vector of size N the size of the population).

modes

vector of the selected mode of each unit. The first mode of the protocol is defined as "m1", the second as "m2" (character vector or factor of size N).

X

design matrix. The rows corresponding to the non-respondents can contain NA (numeric matrix with N rows and p columns).

...

arguments for the function MatchIt::matchit.

Value

for each mode a vector of size N containing for each respondent its outcome or its estimated counterfactual depending on if it answered by the mode or the other. Equal to NA for the non-respondents (numeric matrix of dimension (N,2)).