All in the Family: Partisan Disagreement and Electoral Mobilization in Intimate Networks—A Spillover Experiment

Photo by CBS Television – eBay item  photo front  photo back, Public Domain, https://commons.wikimedia.org/w/index.php?curid=17261926

Authors Florian Foos and Eline A. de Rooij describe their forthcoming American Journal of Political Science article, titled “All in the Family: Partisan Disagreement and Electoral Mobilization in Intimate Networks – a Spillover Experiment,” in the following blog post.

[Archie’s daughter Gloria Bunker-Stivic:] “He’s not voting anyway, so three out of four votes from this house isn’t bad.”

[Archie Bunker:] “Well, she ain’t even gonna get that, little girl, because I just changed my mind. […] Edith and I are gonna vote. So that’s gonna cancel your two votes. So from this house you get zilch.”

(All In The Family, The Election Story, 1971, Season 2 Episode 6)

Voting is a social activity. Since the voting studies of the 1940s, political scientists have emphasized the importance of informal political discussions with family, friends and acquaintances for deciding whether to vote in an election, and which party to support. An important current debate centers on the question if political disagreement fuels mobilization or if people try to avoid potentially conflictual situations by abstaining from politics. In this paper we address this question by focusing on electoral mobilization in our most intimate network: the household. We ask whether households in which individuals support the same political party (“homogeneous”) are more or less conducive to electoral mobilization than households in which individuals support rival parties (“heterogeneous”).

Household members might refrain from discussing politics with those who support other parties in order to avoid conflict. Alternatively, in particular within households, people might more easily endure political disagreement, or might even enjoy discussing politics with non-like-minded others, experiencing a sense of competitiveness. If people are conflict avoiders, electoral mobilization between household members should be more pronounced in homogeneous partisan households; while in the latter, Archie Bunker scenario, mobilization should be equally, if not more pronounced in heterogeneous households.

One of the major challenges in answering this question is that we cannot be certain that it is political discussion that causes higher levels of turnout between homogenous household members, rather than individuals having self-selected into politically like-minded and highly engaged households. To address this causal inference challenge we used the unique features of a randomized spillover experiment that we conducted in the United Kingdom in 2012, combined with a detailed database on voters’ party preferences. We used a voter list to randomly allocate one registered voter per household to one of three groups. Labour Party volunteers attempted to contact subjects in the first and the second group by phone encouraging them to vote in an upcoming election. The message to the second group was far more (Labour) partisan in tone than the message to the first group and hence increased the potential for partisan disagreement between household members. Subjects in the third (control) group received no phone call. By subsequently examining whether the turnout levels of the subjects’ household members increased after the election, we can gauge whether our mobilization messages were more likely to “spill over” through political discussion within homogeneous or heterogeneous partisan households.

Figure 1. The spillover effects of targeting subjects with a campaign message on the turnout of their household members (a – top); and of receiving high versus low intensity partisan messages on the turnout of their household members (b- bottom)

Note: 95% confidence intervals. Based on Figures 3a and 5a in the article.

We find that turnout rates are, if anything, higher among the household members of subjects in heterogeneous households than among those in homogeneous households (or in households without a partisan attachment) as a result of campaign messages generally (Figure 1a). This suggests that members in households that support different parties do not refrain from political discussions in order to avoid conflict. Figure 1b shows that turnout rates are larger still among non-experimental members of heterogeneous households when the partisan tone of the message –and thus the likelihood of discussion– is increased.

Political discussion within personal networks is widely believed to benefit the functioning of democracy by raising political interest and aiding political opinion formation, and ultimately boosting turnout. An important question touches on what kinds of networks are most beneficial. We show that within households the electoral mobilizing function of discussion is certainly not restricted to those in which individuals support the same political parties. It remains an open question whether disagreeing household members mobilize as a result of having been persuaded, or in order to cancel out the other’s vote.