Pre-Primary Polls

Pre-primary polls are those conducted before primary elections in an attempt to measure voters' preferences.

They present a number of difficult challenges for pollsters.

Voters in primary elections can be volatile in their preferences, often because many candidates are relatively unknown and voters have limited information about them. This is especially true in early primary states in U.S. presidential campaigns, when the field of candidates is often wide and there are many lesser-known candidates struggling to gain momentum. With little information, voters can be highly susceptible to the dynamics of the campaign, and every time a candidate starts a new ad campaign or launches a micro-targeting effort, poll numbers can move in response.

Primary electorates are made up of partisans of one party (or of independents who lean toward a party) and thus tend to be relatively homogeneous as compared to the electorates in general elections. It can be easier to predict the outcomes of general elections because partisans of one party generally choose to vote for their party's candidates, making their votes relatively easy to predict. Pollsters in general elections can focus on studying how independents and swing voters will behave on Election Day. But in primary elections, voters base their decisions on a complex set of preferences that go beyond basic party affiliations and are much harder to quantify.

Primary elections also often have low or variable turnout, making turnout projections difficult. While turnout in heavily contested primaries early in the presidential cycle sometimes can rival turnout in general elections, turnout in primaries is usually much lower. Many primaries are uncontested or are not very competitive, because incumbents are rarely challenged. Thus when a competitive primary comes along, there is often little voting history that can be useful in predicting future turnout. Turnout models for general elections that place importance on past voting behavior often are not as good at predicting turnout in primaries. And when competitive primaries happen in areas that have not seen competition in a while, they often draw large numbers of people who have not voted in a primary before. The behavior of first-time voters can be difficult to predict.

U.S. election laws and party regulations related to the administration of primary elections and party caucuses vary dramatically across states, adding to the complexity of preprimary polling. For example, some states allow only voters registered with a given party to vote in that party's primary. These primaries are called “closed primaries.” Other states have “open [Page 609]primaries” that allow any voter to choose the primary in which he or she will vote. This makes it important to include political independents in pre-primary polls in open primary states. The fact that independents can change their minds at the last minute about which party's primary they will choose makes it a challenge to predict the composition of each party's electorate.

The fact that primary electorates are made up of one party's partisans also presents difficulties for sampling. Most pre-primary polls use screening techniques to determine eligibility to vote in a particular primary, and such screening techniques can be expensive to administer. In states with closed primaries and party registration, lists of registered voters can be used to draw samples of party registrants, but the quality of such lists varies dramatically by state. Lists can be out of date or lack contact information for the voters. It can also be challenging to develop appropriate weighting schemes for pre-primary polls, because it is difficult to find good estimates of the demographic profiles of a particular primary electorate that can be used as weighting benchmarks.

...