Are Leaners (Still) Closet Partisans?

• In Event: Political Identity, Partisanship, and Preferences

Abstract

When measuring party identification, the conventional wisdom is that people who initially reject a party label, but later say they lean toward one of the two parties are actually “closet partisans.” In other words, they are as likely as true partisans to exhibit stable partisan behavior over time. As a result, most analyses of public opinion and voting behavior control for party identification by including leaners as partisans. Our goal in this study is to establish whether the empirical bases for this well-established practice remain true today.

The conventional wisdom is worthy of re-examination now for several reasons. First, the number of people who refuse to select either Republican or Democrat in response to the first question in the party ID sequence has changed. The proportion of people who reject partisan labels when first asked has increased from around 30% in the early 2000s to almost 40% in 2022. The ANES surveys suggest a change of similar magnitude. If leaners are indeed closet partisans, this trend may not be of great interest (cf., Klar and Krupnikov 2016). On the other hand, it is somewhat incongruous that citizens should be rejecting these labels when levels of affective polarization are known to be extremely high, and partisan identity is now believed to be more of an all-encompassing identity, reinforced by other social divisions (Mason 2015).

A second important reason to reevaluate this practice stems from changes in survey research. Since the original conclusions about leaners were drawn, the dominant survey mode has changed. Although a few well-funded organizations such as the American National Election Study and the General Social Survey still conduct face to face interviews, and a few organizations still use telephone interviews, most survey research has moved online. This means that data collection practices have changed since the time this conventional wisdom was established.

As our results demonstrate, the online survey mode unintentionally alters the number of true independents because it is a self-administered mode of interviewing as opposed to a live interaction between the interviewer and the respondent. For example, in interactive interview modes, those who initially choose Independent or no political party are asked, “Do you lean more toward the Democrats or the Republicans?” Importantly, respondents are not explicitly offered the option that they do not lean, but it can be verbally volunteered to the face-to-face interviewer. Online, it is not possible to volunteer a response that is not offered. In order to be recorded as a “true” independent in an online survey that does not offer “Neither” option, one must skip the question entirely. Alternatively, self-administered surveys may offer the “Neither” option, but that decision—to put the response option on the screen—also has implications for the number of people who become classified as true independents. This situation becomes increasingly confusing when face-to-face and online interviews are combined into single data sets as is the case with the ANES.

Since leaners are traditionally treated as closet partisans, and the online survey mode changes who is ultimately counted as partisan versus independent, this convention is worthy of re-examination. Given the importance of party identification in controlling for stable political predispositions, contemporary analysts of political survey data need to know whether the “closet partisan” thesis still holds up as an ongoing convention given changing modes of survey administration, and the rising number of Americans who do not readily identify with a party label.

We use both cross-sectional and representative national probability panel surveys to address this question. Our focus is on the extent to which these partisan categories predict greater or lesser stability in voter preferences over time. Overall, our results thus far indicate that leaners are indeed different from partisans, and that they are different as well as from true independents.

Authors

• Chloe Ahn, University of Pennsylvania

• Diana C. Mutz, University of Pennsylvania