Mobilizing Collective Identity: Frames & Rational Individuals

See Chong (2000) and Chai (2001) for rational explanations of the origins and maintenance of group identity, which is very different from the focus here on mobilization influences on moderate identifiers.

This view is consistent with framing theory in that framing theory explains framing as well as priming using cues (Chong and Druckman 2007a). As Chong and Druckman (2007a) discuss, framing can work via accessibility or applicability, and framing and priming are theoretically indistinguishable.

The social movement literature does provide important concepts for understanding mobilization, such as preconditions to framing movements in the form of political opportunity structures (Diani 1996; McCarthy and Zald 1996; Polletta and Ho 2006). However, the theoretical causal mechanisms have yet to be fully explored or empirically tested.

Previous work on group identity and behavior do not address the question of when group identity will have influence. Many scholars have studied whether ingroup bias leads to increased support for same-race candidates (Barreto 2007; Canon 1999; Lublin 1997; Terkildsen 1993). However, in most cases, group-based mobilization is often assumed but not explicitly accounted for. Instead, ingroup bias alone, rather than campaign effects, is predicted to drive turnout and participation. Also, rational choice approaches to group identity focus on other-regarding preferences, such as levels of altruism or group utility (Edlin 2007; Fowler and Kam 2007) or on group coordination, known as the ethical voter approach (Coate and Conlin 2004; Feddersen and Sandroni, A theory of participation in elections, Unpublished Paper, 2002). However, these approaches not only overlook mobilization effects, but also fail to predict which individuals will gain utility from the group. Therefore, they explain only the behavior of strong identifiers—the very same people who are most likely to mobilize regardless of the circumstances. A second rational choice approach to group identity argues that group leaders increase the costs for not participating as well as promising benefits to those who participate (Morton 1987; Uhlaner 1989) or can increase mobilization efforts in close elections or elections with high stakes, which increases turnout (Shachar and Nalebuff 1999). However, these approaches fail to explain how leaders successfully mobilize group members or increase costs for not participating. In other words, we still cannot answer questions related to how or why certain mobilization strategies are successful sometimes and yet fail in other circumstances.

Strong identifiers are not included in this study since they are likely to participate regardless of mobilization, and so are unlikely targets of mobilization. However, other work suggests that strong identifiers actually respond most to the group frame in a minority context because an underdog effect is triggered, whereas they are confident enough that fellow members will turn out in parity & majority that they free-ride (Aroopala 2010).

There are also pieces that critique the “p term” in the rational choice equation, and argue that the likelihood of actually being pivotal is so small that it likely has no effect (see Aldrich 1993 for a discussion). However, the pivotality I discuss here has to do with the perception of being pivotal, as induced by parity of group size, rather than the small likelihood of actually being pivotal. I argue that the perception of being pivotal, regardless of the actual likelihood of being pivotal, increases participation, as evidenced in various studies of threshold effects (Suleiman and Rapoport 1992).

See Rabbie et al. 1989 and Karp et al. 1993 for more discussion on how expectations of reciprocity can play a large role in the decision to take costly action, particularly in a public goods environment, which is akin to voting.

These arguments are in keeping with Ethnic Conflict literature stating that individuals are more likely to take action based on group identity when they perceive benefits associated with their identity and when they receive targeted mobilization from elites geared towards enhancing group-based behavior for political gain (Green and Seher 2003; Chai 2001; Brass 2003). However, this literature does not specify conditions, such as threshold effects or empirically test for them.

This, however, is dependent upon the strength of the frame: a weak frame will likely fail without the added threshold effect, whereas a strong frame may work independently of the threshold effect as well.

As noted above and in the conclusion, the rational choice explanation that a feeling of being pivotal is induced by being near a threshold is not inconsistent with a psychological viewpoint that being near a threshold increases the interest in and perceived importance of the event. Whichever view of the psychological explanation for the observed threshold effects on behavior is taken, my central argument here remains the same: this threshold effect can help us better predict when group frames will work because it either reinforces or undermines the group frame. In other words, accounting for threshold effects and group frames will better predict successful mobilization than will group frames alone, which suggests that group identity’s role in political behavior varies according to environmental factors such as frames and instrumental factors.

As the focus here is on in-group mobilization, I do not include the option of voting for the out-group.

This payoff Table is from the instructions for the Neutral condition. The Rivalry condition replaced group names with “Your Group…” and “Rival Group…” Similarly, the entire instructions set for the game were presented in either this Neutral or Rivalry language, dependent upon the frame treatment.

The conversion rate was .37 cents for each point.

Note, however, that Levine and Palfrey (2007) have two sessions with fifty rounds each, totaling 100 rounds.

This identity assignment differs significantly from Levine and Palfrey (2007) who randomly switched subjects between Alpha and Beta identities in each round. As such, their experimental design does not allow a real test of identity-based behavior, which is the primary focus of this project.

See the sample decision screens in the Appendix.

See footnote 2 on framing theory as it relates to cues and priming.

The frame used here is a subtle frame. As Chong and Druckman (2007a) highlight, frames can vary in strength and may have differing effects based on strength. Therefore, it is quite possible that an overt frame would have even stronger effects.

Note, the size treatments here differ from those in Levine and Palfrey (2007) in two important ways. First, each subject was in only one of two main treatments in which the size of both groups was fixed: landslide majority/minority difference or toss-up (small majority/minority difference). Further, within these treatments, each subject was randomly reassigned between minority and majority in each round, and real numbers were used instead of percentages (i.e. 4 Alpha and 5 Beta). Since my focus in this project is on the individual’s perception of pivotality, which more closely approximates actual voter feelings in real elections, I use percentage of total electorate rather than raw numbers.

Statistical analysis shows that pooling the minority cases of 15–25% and 20–30% and pooling the majority cases of 75–85% and 70–80% does not affect the results.

Table 8 in the Appendix gives an example of the composition for one of the smaller groups with six total participants. In a round with six total subjects, there can be 0 Alpha that are also Blue, 1 Alpha that is also Green, 3 Beta that are also Blue, and 2 Beta that are also Green. The decision screen would show that 15–25% of the participants in that round are Alpha and 75–85% of the participants are Beta (1 Alpha and 5 Beta). Similarly, the screen would also show that 45-55% of the participants are Blue and that 45–55% of the participants in that round are Green (3 Blue and 3 Green). Please see Figs. 4 and 5 in the Appendix for sample decision screens.

The order of the groups listed on the screen (i.e. the percentages) are randomly reversed half of the time. For example, sometimes the Alpha/Beta identity is listed first and sometimes the Blue/Green identity is listed first. Please see the sample Decision Screens in the Appendix (Figs. 4, 5).

In the 1st, 20th^,, and 35th rounds, manipulation checks were given before subjects made their choice for that round asking subjects to indicate which groups they were in, which two groups determining earnings in that round, and what the relative sizes of each of the groups in that round were. The results indicate that subjects were attentive and comprehended the rules & instructions.

In a second control session, only one group identity, in addition to the respective sizes of each of the two groups, is shown on the screen. In this way, I am able to isolate whether showing the sizes of both group identities on the screen (a total of four groups) has any effect on the likelihood of voting for X.

Similar findings can be seen when the whole sample is pooled (across both identities and all group sizes).

There is some debate about whether altruism generally refers to feelings towards an entire population or towards the individual’s own group. In this context of competing groups, it seems the latter explanation is more viable as it goes against a vast literature on ingroup/outgroup behavior to argue that altruists are more likely help the outgroup at the expense of their ingroup. Additionally, altruism is included here as recent work points towards its role in turnout decisions (Fowler and Kam 2007).

Some may be concerned that the Altruism measures were taken post-treatment, however dropping Altruism from the models does not significantly affect the main findings. Altruism is included in the models, as is gender, due to its established importance by previous literature.