Logo

5.1 Background Information

To demonstrate how TRMCO can be applied to political studies and what difference TRMCO makes in the current method, the author presents a replication study of the aforementioned 2007 Hellwig and Samuels article. The reason for choosing this work for replication is threefold. First, the dependent variable is the vote share of the incumbent party, which perfectly fits the distributional assumption of the truncated normal distribution and is naturally bounded between 0 and 1. Second, the dataset is available for public access and anyone can replicate the analysis. Third, the original analysis adopts the OLS model without any complication, and this makes the replication study easier.

There are two regression models in Hellwig and Samuels's article. Both models apply incumbent party's vote-share as the dependent variable, and use 13 independent variables in each regression with minor changes. The estimation is conducted with the OLS method, and robust standard errors are reported for hypothesis testing (Stata). In Model I, the major explanatory variables include Previous Vote (the incumbent party's vote share in the previous election), Economy (annual percentage change in real per capital GDP), and Trade Openness (the sum of the country's exports and imports as a proportion of its GDP). The control variables are Presidential Election (a dummy variable), Re-election (a dummy variable based on whether the incumbent president was running for re-election), Effective Number of Parties, Income (thousand of constant U.S. dollars), and four regional dummy variables Africa, Asia, Central and Eastern Europe, and Latin America and the Caribbean (using the advanced industrial democracies as the baseline category). Two remaining variables are interaction terms Economy$\times$Trade Openness and Economy$\times$Presidential Election. Model II replaces the explanatory variable Trade Openness with Capital Flows (gross private capital flows as a share of GDP), and thus, the interaction term Economy$\times$Trade Openness is replaced with Economy$\times$Capital Flows. The rest of the model specifications remain the same. The sample size is 426, including all democracies (+6 or better in Polity IV's ranking of democratic quality) in the world from 1975 to 2002.

Download [full paper] [supplementary materials] [all replication files] [technical note]