Interpreting Ethnicity and Urbanization in Malaysia's 2013 General Election

Abstract

In this article I reinterpret Ng, Rangel, Vaithilingam, and Pillay's analysis in this issue of pro-BN voting in Peninsular Malaysia in Malaysia's 2013 general election. I show that the authors' statistical methods are inappropriate for testing whether district ethnicity predicts district-level BN vote share, and that their modeling choices result in tests of hypotheses that do not exist and cannot be derived from standard theoretical approaches to ethnic voting in Malaysia. I then provide a range of statistical evidence that supports three main conclusions: (1) ethnicity and district area (a proxy for urbanization) both predict BN vote shares at the district level, (2) neither the effect of ethnicity nor of district area can be reduced to the other, and (3) there is no interactive effect between ethnicity and urbanization. These results are in direct contradiction with the authors' results, and apply equally in Peninsular Malaysia and the entire country. I also discuss the broader issues that emerge when testing competing theories of BN vote share.