Autocorrelated errors explain the apparent relationship between disapproval of the US Congress and prosocial language

TL;DR

This paper shows that the perceived link between prosocial language and disapproval of US Congress is due to autocorrelated errors in the statistical model, not a true relationship.

Contribution

It identifies the importance of accounting for autocorrelation in models analyzing political language and disapproval data, correcting prior misinterpretations.

Findings

The original relationship was an artifact of model misspecification.

Autocorrelation in errors explains the apparent correlation.

Proper modeling removes the supposed link.

Abstract

Recently, it has been claimed by Frimer et al. (2015) that there is a linear relationship between the level of prosocial language and the level of public disapproval of US Congress. A re-analysis demonstrates that this relationship is the result of a misspecified model that does not account for first-order autocorrelated disturbances. A Stata script to reproduce all presented results is available as an appendix.