Improving the Interpretation of Random Effects Regression Results
| Author | Soren Jordan,Andrew Q Philips |
| DOI | http://doi.org/10.1177/14789299211068418 |
| Published date | 01 February 2023 |
| Date | 01 February 2023 |
| Subject Matter | Professional Section: Methods |
https://doi.org/10.1177/14789299211068418
Political Studies Review
2023, Vol. 21(1) 210 –220
© The Author(s) 2022
Article reuse guidelines:
sagepub.com/journals-permissions
DOI: 10.1177/14789299211068418
journals.sagepub.com/home/psrev
Improving the Interpretation
of Random Effects Regression
Results
Soren Jordan1 and Andrew Q Philips2
Abstract
Mummolo and Peterson improve the use and interpretation of fixed-effects models by pointing
out that unit intercepts fundamentally reduce the amount of variation of variables in fixed-effects
models. Along a similar vein, we make two claims in the context of random effects models. First,
we show that potentially large reductions in variation, in this case caused by quasi-demeaning,
also occur in models using random effects. Second, in many instances, what authors claim to be
a random effects model is actually a pooled model after the quasi-demeaning process, affecting
how we should interpret the model. A literature review of random effects models in top journals
suggests that both points are currently not well understood. To better help users interested in
improving their interpretation of random effects models, we provide Stata and R programs to
easily obtain post-estimation quasi-demeaned variables.
Keywords
methodology, random effects, demeaning, interpretation
Accepted: 3 December 2021
Mummolo and Peterson (2018) discuss how fixed-effects (FE) models in panel data—
including unit-specific dummy variables to allow the intercept to shift up or down for
each unit—estimate only within-unit variance. Thus, coefficients describe only the effect
of a one-unit increase occurring within a unit, not between units. FE models have several
advantages, namely that they exclude the possibility of endogeneity caused by a correla-
tion between the included covariates and any omitted time-invariant unit effects. However,
FE complicate interpretation, since variation over time is generally appreciably smaller
than overall variation (i.e. what we would interpret with a global summary). To avoid
problematic interpretation, especially overstating the substantive effects of the model,
Mummolo and Peterson suggest accounting for this within-transformation, since it will
almost always result in smaller levels of variance for the variable of interest to consider
when constructing counterfactuals.
1Auburn University, Auburn, AL, USA
2University of Colorado Boulder, Boulder, CO, USA
Corresponding author:
Soren Jordan, Auburn University, 7080 Haley Center, Auburn, AL 36849-5412, USA.
Email: sorenjordanpols@gmail.com
1068418PSW0010.1177/14789299211068418Political Studies ReviewJordan and Philips
research-article2022
Professional Section: Methods
To continue reading
Request your trial