Advanced Polling Techniques: Road to the Right Message - Strategy & Tactics - Brief Article

IT IS FAR FROM a political revelation that messaging and voter turnout win elections. However, traditional polling sometimes ignores the real science of message testing and strategy development. While many campaigns stick to the polling basics, a special election in Missouri proved the value of advanced statistical analysis. Our firm, which worked for the John Cauthorn campaign, put these analytical tools to use in a critical race that would determine control of the state Senate.

Traditionally, political pollsters concentrate on means tested topline data, such as how the candidate and their messages rate with voters grouped by income, gender, party identification, ideology and media market. We decided more was needed in this election and decided to implement some advanced statistical analysis to determine the messaging for this campaign.

Early polling in the district showed that Cauthorn was the lesser known of two candidates and was considered by most to be the underdog in the race. However, the early data also showed that as Cauthorn's name recognition improved, so would his standing with voters, provided that recognition came with the right associations.

A common campaign theory is that local campaigns should be run on local issues. This was the strategy implemented by Cauthorn's opponent early on in the campaign. Local issues, like highway repair and local education programs, dominated the debate from the onset of this special election. These were the messages that tested best when you only looked at the means.

While our early benchmark polling showed that highway repair and improving local education were the most popular issues among these rural voters, deeper statistical analysis -- all too rarely used in political polling -- showed that gun owners rights and abortion also resonated well, and more importantly, were more likely to lead to votes for Cauthorn.

By relying on means testing, we would have limited the message analysis to stated importance alone. However, by examining the data in a number of different ways, we were able to determine the derived importance of these messages. The derived importance told us that even though certain messages did not receive the highest mean importance rating, they were more important to actual vote choice. Furthermore, this analysis showed that these issues would energize Cauthorn's base without having the same result for his opponent.

While traditional topline analysis was important, we were able to use the following analytical techniques to fine tune our messages and identify those that would lead to the greatest support for our candidate:

1. CHAID Analysis. CHAID analysis examines all possible splits of the dependent variable by performing classification and segmentation analysis using significance testing. It is a method we use for profiling a candidate's base, opposition, and most importantly, the swing voters. This process allowed us to identify where the swing voters who were likely to turn out on election day lived, where they stood ideologically, as well as various demographics such as income level, education and gender. We were then able to identify the messages that resonated best with these swing voters, the results of which would turn out to be different from what means testing told us alone.

2. Standard Deviation: Standard deviation determines the range from which respondents depart from the average. By looking at the standard deviation surrounding the mean rating, we were able to determine which messages resonated most similarly among the broadest range of voters. We discovered that both the gun and abortion messages resonated most predictably amongst a relatively broad range of voters. While this was not surprising in a rural district, it did let the campaign know that they would not have to be timid in the delivery of these messages.

3. Correlations: Perhaps the most useful tests we did involved the usage of Bivariate Correlation Matrixes. These tested whether there was an association between respondents' views on the messages and their vote choice -- something means testing cannot do and an invaluable addition to this campaign strategy. We were able to identify the messages that had the most impact on vote choice for Cauthorn. While highway repair and local schools were stated as popular issues amongst the voters, they were not relative to their vote choice in either direction. However, we learned that abortion and guns did move voters and, most importantly for a special election, motivated turnout among our base and even among swing voters inclined to support our candidate.

While these analytical tools produced recommendations that differed from what was expected, the Cauthorn team trusted the data. By the time Cauthorn's opponent realized that the major issues of the campaign would be guns and abortion, Cauthorn owned those voters, and it was simply too late.

Through these advanced polling techniques, Cauthorn was able to be sure that his messages were not only influencing voters, but also motivating them to get out and vote. As a result, his win gave Republicans control of the Missouri state Senate for the first time in 60 years. While the polling was only one aspect of this upset win, the advanced analysis improved the overall utility of polling and gave us an edge we needed.

Chris Wilson is president of Wilson Research Strategies, a Republican polling firm that recently joined with Qorvis Communications with offices in D.C. and Tysons Corner, VA.

COPYRIGHT 2001 Campaigns & Elections, Inc.
COPYRIGHT 2002 Gale Group