Explaining Fixed Effects: Random Effects Modeling of Time-Series Cross-Sectional and Panel Data*

By Andrew Bell, Kelvyn Jones. Political Science Research and Methods, 2014

1,496 citations recorded by the source corpus.

This article challenges Fixed Effects (FE) modeling as the ‘default’ for time-series-cross-sectional and panel data. Understanding different within and between effects is crucial when choosing modeling strategies. The downside of Random Effects (RE) modeling—correlated lower-level covariates and higher-level residuals—is omitted-variable bias, solvable with Mundlak's (1978a) formulation. Consequently, RE can provide everything that FE promises and more, as confirmed by Monte-Carlo simulations, which additionally show problems with Plümper and Troeger's FE Vector Decomposition method when data are unbalanced. As well as incorporating time-invariant variables, RE models are readily extendable, with random coefficients, cross-level interactions and complex variance functions. We argue not simply for technical solutions to endogeneity, but for the substantive importance of context/heterogeneity, modeled using RE. The implications extend beyond political science to all multilevel datasets. However, omitted variables could still bias estimated higher-level variable effects; as with any model, care is required in interpretation.

Subfields: Economics and Econometrics, Political Science and International Relations, Statistics and Probability

View Paper: https://www.semanticscholar.org/paper/fe5bb5d8d6b7ac251d87bc16e75ea5889cc92425

How researchers see this paper

The researchers whose trusted networks rank it highest, strongest advocate first. Every researcher gets a different order.

In Economics and Econometrics

In Political Science and International Relations

In Statistics and Probability

Citations in the corpus

The pricing of green bonds: external reviews and the shades of green (2020, neutral): “Besides its advantages mentioned in the previous subsection, the hybrid model enables incorporating time-invariant variables (Bell and Jones 2015).”
Operational Productivity, Corporate Social Performance, Financial Performance, and Risk in Manufacturing Firms (2016, methodology): “Since our data has multiple industries, Bell and Jones (2015) recommend explicitly modeling these higher order variances to account for heterogeneity at the industry-level in addition to the ﬁrm-level.”
When Should We Use Unit Fixed Effects Regression Models for Causal Inference with Longitudinal Data? (2019, background): “…from model-based perspectives (often assuming linearity), with little attention to the causal identification assumptions (e.g., Beck 2001; Bell and Jones 2015; Clark and Linzer 2015; Wilson and Kosuke Imai is Professor of Government and of Statistics, Harvard University, 1737 Cambridge Street,…”
Bridging Methodological Divides Between Macro- and Microresearch: Endogeneity and Methods for Panel Data (2020, background): “…of the independent variable of interest and the raw variable, or (c) both demeaning and including the higher-level entity mean in the model (Bell & Jones, 2015; Certo et al., 2017; Raudenbush, 2009).11 Our takeaway is that both fixed- and random-effects models are equally viable options for…”
But You Don’t Look Like A Scientist!: Women Scientists with Feminine Appearance are Deemed Less Likely to be Scientists (2016, neutral): “A good deal of research neglects to examine divergent sources of variance, which can mask important relationships in the data (Bell and Jones 2015; Kievit et al. 2013).”