Consultants often treat polling as an amulet that protects candidates from the evil campaign gods. If the numbers are good, it means everything will be fine come Election Day.
Pollsters, meanwhile, often feel isolated from a campaign’s decision making, especially if the general consultant has an outsized role. If left unchecked, tension between a campaign’s general consultant and its pollster can derail a candidate.
Here’s how consultants can overcome their misconceptions about surveying and better work with their pollster.
Polling Isn’t A Magic Amulet
Polls are primarily there to improve campaign decision making and often pay for themselves by leveraging campaign resources. Take the time to reflect on the decisions you’re going to make after the results come back. Then work backwards from that list as you contemplate all the mechanics of polling: sample size, sampling frame, and questionnaire, et cetera. If the poll results don’t inform your decisions, then you’ve got nothing.
Let the Pollster Write the Questions
A consultant’s first impulse is to send the pollster an outline of the questionnaire, or worse: a list of questions. Suppress that impulse. Pollsters are experts when it comes to writing questions. They understand that survey respondents must hear the question, digest it, form an answer and then communicate it — all while wondering what the interviewer is really up to.
Exploit Appended Data
The best survey samples come from voter files containing appended data. Take the time to learn what variables might come along with the sample at no extra charge, and insist that your pollster use that data during analysis. As you contemplate survey design, think through how that appended data might inform your subsequent decisions.
Use Contemporary Analytics to Perform Data Analysis
Static survey products such as toplines and crosstabs have been a staple of data presentations since PowerPoint was introduced in 1990. Now, those tools are becoming passé. Schopenhauer’s genius comes from seeing patterns in the data that your competitor doesn’t see. The job of the pollster and consultant is to visualize those patterns. Interactive analytics can make that happen, and the software is friendlier than you might think.
Questions measure variables. Variables fall into three categories: behaviors such as party registration, union membership, choices among alternatives, or voting; Opinions such as ideology, party loyalty, and priorities; and demographics such as age, gender, and ethnicity.
Traditional crosstabs, for the most part, break down behaviors by a limited set of demographics. This is inadequate when trying to visualize illusive insights. Today’s analytics contain tools that examine all of the possible relationships between and within the three categories of variables. The strongest drivers of voting behavior, for example, don’t tend to be demographic, but psychographic and behavioral. That requires analytic tools not found in a binder full of crosstabs.
Use Analytics To Design Better Questionnaires
Improved analytics are also driving innovation in questionnaire design. Pollsters and consultants who rely on improved analytics can design better questionnaires. Lacking space to catalog all of the possibilities, let me provide one example. With improved analytics your questionnaire can anticipate relationships between entire blocks of questions and behavioral variables. You can isolate those respondents who moved in your direction from a first ballot test to a second. Then you can see how that subset of the sample reacted to the full block of messages.
In short, your task is to see what your competitor cannot see. That’s campaign genius.
Val R. Smith has been a pollster and market researcher for 35 years. He is currently the Director of Research at SmithJohnson Research and a Principal in Data Analysis and Display, LLC.