Introduction

Public opinion polls, widely used and relatively low in analytical complexity, are often highly accurate for election forecasting (Boon, 2012; Graefe, 2014; Hillygus, 2011). However, in highly contested elections where variability in voter intention is high, a prudent polling organization may want assurance that a single poll result is not biased. Bellwether counties or cities that are known to mirror statewide preferences can help provide this assurance, quickly and inexpensively. In other words, bellwether polls are combined with polls of larger areas as sister-tests toward development of more accurate pre-election predictions. While statewide polls yield vote share percentage estimates, bellwether polls yield an up-or-down indicator of the election winner. The combination, or pooling, of results may offer more nuanced information which may be missed in a survey of the large area alone.

The key to successfully using a sister-test polling methodology is identifying the bellwether areas. While there is a very small, dated, and inconclusive literature on bellwethers, one modern-day polling organization has successfully used this approach since 2003. The Suffolk University Political Research Center (SUPRC) bellwethers with state polls correctly forecast election outcomes in 97% of trials where a clear winner could be determined (i.e., no ties). The contribution of this article is to offer a detailed view of the bellwether identification and selection process and to provide information on the robustness of this non-traditional public opinion polling method.