Ole-Andreas Elvik Næss, NHH

Department seminar. Ole-Andreas Elvik Næss is a PhD Candidate at NHH. He will present a paper entitled "Unbiased Treatment Effects with Interference: Theory and Evidence from a Large-Scale Voting Experiment".

Photo of Ole-Andreas Elvik Næss

Ole-Andreas Elvik Næss

Abstract

Causal inference with interference between units is diffcult, and it is particularly hard to estimate global treatment effects. The contribution of this paper is to understand when this interference will bias the estimates, and to outline a method to estimate unbiased treatment effects that can be used without making any assumptions about the form of the interference. We show that randomizing the timing of the treatment and comparing the outcome variable in a short interval before and after the treatment will lead to an unbiased treatment effect if the outcome is continuous in time.

We use this method to investigate whether a government can increase the electoral turnout with a text message. We conduct a large-scale field experiment (N = 199.618) in Norway and estimate that the encouragement increased turnout with 2-3 percentage points. Our choice of identification strategy also enables us to understand the under- lying mechanisms and the effect on other electoral outcome variables.

Read the full paper here [pdf]

Host: Tuomas Laiho

Published Dec. 20, 2018 3:13 PM - Last modified June 11, 2019 1:48 PM