Improving Tests of Theories Positing Interaction

By William D. Berry, Matt Golder, D. Milton. 2012

520 citations recorded by the source corpus.

Subfields: Political Science and International Relations, Sociology and Political Science, Statistical and Nonlinear Physics

View Paper: https://www.semanticscholar.org/paper/6f8acdf43f18bb6fe1d2444eac658cca6a549c2f

How researchers see this paper

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

In Political Science and International Relations

1. N. Miller
2. Thomas J. Leeper
3. Peter Lorentzen
4. P. Landry

In Sociology and Political Science

1. T. Hartman
2. Ryan T McKay
3. Antón P. Martínez
4. Jilly Gibson-Miller
5. Mark Shevlin
6. Liat Levita
7. T. Stocks
8. Jamie Murphy

In Statistical and Nonlinear Physics

1. Thomas J. Leeper

Citations in the corpus

• How Much Should We Trust Estimates from Multiplicative Interaction Models? Simple Tools to Improve Empirical Practice (2016, unknown): “Although the linear regression framework is flexible enough to incorporate higher order terms of X and their interaction with D (see Kam and Franzese Jr. 2007; Berry, Golder, and Milton 2012), this is rarely done in practice.”
• Undermining Authoritarian Innovation: The Power of China’s Industrial Giants (2013, background): “In addition, following the recommendation of Berry et al. (2012) we consider what our theory suggests about the difference between these two types of cities when at the extremes of the sample.”
• Overstating and Understating Interaction Results in International Business Research (2017, unknown): “Additionally, we refer scholars to discussions about levels of measurement (Allison, 1977), symmetric interactions (Berry et al., 2012), false positive rates (Esarey & Sumner, 2016), and flexible estimation strategies (Hainmueller, Mummolo, & Xu, 2016).”
• The Political Resource Curse (2014, unknown): “In a limited dependent variable model, the coefficient on the interaction term does not quite capture the modifying effect—the modifying effect depends on the values of the other covariates (Berry, DeMeritt, and Esarey 2010; Berry, Golder, and Milton 2012; Norton, Wang, and Ai 2004).”
• Interpreting Regression Results using Average Marginal Effects with R ’ s margins (2017, background): “library("stats") m <- lm(y ~ x1 * x2 * x3) cf <- coef(m)[-1] # drop beta_0 me_x1 <- cf[1] + cf[4]*x2 + cf[6]*x3 + cf[7]*x2*x3 me_x2 <- cf[2] + cf[4]*x1 + cf[5]*x3 + cf[7]*x1*x3 me_x3 <- cf[3] + cf[5]*x2 + cf[6]*x1 + cf[7]*x1*x2”
• The Authoritarian Dynamic During the COVID-19 Pandemic: Effects on Nationalism and Anti-Immigrant Sentiment (2020, methodology): “Scholars recommend that conditional effects are calculated from the marginal effect at every observed value of the moderator; in this case, COVID-19 anxiety, to properly interpret interactions (e.g., see Berry et al., 2012; Brambor et al., 2006).”
• The Secret Success of Nonproliferation Sanctions (2014, background): “Berry, Golder, and Milton 2012 note that interaction terms are inherently symmetric (meaning that each component variable conditions the other) and therefore recommend constructing plots both ways. expected.70 Taken together, these results suggest that in the postsanctions era, the biggest…”