Not That There’s Anything Wrong with That: The Effect of Personalized Appeals on Marriage Equality Campaigns

Missouri voters approved a same-sex marriage-ban amendment August 2, 2004; Louisiana voters on September 18, 2004. Voters in Arkansas, Georgia, Kentucky, Michigan, Mississippi, Montana, North Dakota, Ohio, Oklahoma, Oregon and Utah approved their amendments on Election Day (Peterson 2004). Although pundits claimed that the presence of ballot initiatives proposing gay marriage bans led to increased turnout among conservative voters and thus contributed to the victory of Republican candidate George W. Bush, scholars have found mixed evidence as to whether or not such claims are warranted (see Smith et al. 2006; Campbell and Monson 2008).

According to the New York Times, gay marriage is supported by a majority of residents of 17 states, while as recently as 2004 it did not have majority support in any state (Gellman et al. 2010). Also, Gallup Poll data from May 3–6, 2010 found that support for gay marriage was up to 46%, significantly larger than the 27% level of support measured when Gallup first asked the question in 1996. The same poll found that in 2010, the percentage of Americans who believe homosexual relations are morally acceptable crossed the symbolic 50% line for the first time (Saad 2010).

Personal Communication with Thomas Wheatley, Organizing Director, Basic Rights Oregon, February 8, 2010.

A number of previous field experiments have been used to explore how best to increase charitable giving. Miller and Krosnick (2004) found that appeals to donate to an abortion rights group that included a “policy change threat” message were more likely to result in monetary donations than were appeals that included a “policy change opportunity” message. Frey and Meier (2004) find that individuals are more likely to contribute when informed that a majority of other individuals have already done so; Shang and Croson (2009) find that individuals are influenced by information about how much others have contributed. Karlan and List (2007) find that the rate and amount of giving is positively influenced when large matching gifts are available.

Group versus non-group sources provides another element of personalization that may complicate the premises of contact theory. Extant research in political science and in social psychology finds that group members tend to be more credible than non-group sources (Calvert 1985; Aroopala 2011). Calvert (1985) goes so far as to suggest that otherwise rational actors may even prefer biased sources provided those sources share their worldview. When a caller self-identifies as a member of the gay community, the emphasis and salience may shift to the impact of marriage equality laws on that particular individual and away from the impact on and needs of the group, the validity of the arguments presented, etc. This could lead to diffused results for several reasons. A participant may feel they are helping one person (i.e., the caller) rather than the organization or a potential donor may feel they are giving to the individual and not feel the social benefits of group membership.

Similarly, cognitive reactance, a social psychology theory, seeks to explain counter-effects of persuasion by suggesting that if one aspect (i.e., message, theme or tactic) of persuasion is over-emphasized, there is an increased likelihood that respondents will choose another alternative (Sherman et al. 2004).

The full script of the supporter identification phone bank effort can be found in Appendix C in the electronic supplemental material.

The full scripts for both the treatment and control conditions can be found in Appendix A and Appendix B in the electronic supplemental material. The treatment script provided other language in addition to personalization, including language about pride in being an Iowan and a mention of the role of the state Supreme Court in the recent marriage equality decision; thus, it is possible that those aspects of the treatment script also affected the behavior of contacted individuals. However, given that similar language regarding the role of the Supreme Court and the focus on Iowa appears elsewhere in the two scripts, we believe the coming-out aspect of the treatment paragraph is the key difference.

These contact rates are comparable to those found in field experiments in voter mobilization using live phone banks (see Gerber et al. 2008); comparable contact rates are not available for other fundraising field experiments due to the different methodologies employed (e.g., mailers, radio, or approaching random pedestrians). The different rates of contact between the two scripts reflect the pool of callers available to One Iowa. Many were straight allies and thus were not able to be assigned to the treatment script; also, not all LGBT individuals were willing to use the coming-out script. Although the organization did not keep detailed records of the sexual orientation of callers, their impression is that there were more callers assigned to the control group than to the treatment script.

We also modeled willingness to volunteer using a variable we calculated indicating the distance to each individual’s city of residence from the One Iowa headquarters in Des Moines. Consistent with the other results presented above, distance was a statistically significant negative predictor of willingness to volunteer, regardless of whether or not other control variables (age, gender, etc.) were also included in the model.

The full scripts for both the treatment and control conditions can be found in Appendix A and Appendix B in the electronic supplemental material.

We also calculated Models 2 and 3 with variables designed to be more precise than the dichotomous “Des Moines” resident variable, either indicating the population density of each individual’s home address or whether or not their home city is considered urban or rural (according to U.S. Census designations). Neither of these variables were statistically significant predictors and the results otherwise remain relatively unchanged.

Stein et al. (2000) find that in areas where Whites lived with larger proportions of Hispanics nearby (and thus were available to be in contact with them), contact had a positive effect on attitudes towards Hispanics. They find a curvilinear relationship between attitudes and the interaction of contact and context. Similarly, Overby and Barth (2002) find that citizens living in areas with higher gay populations demonstrate warmer attitudes toward homosexuals, a finding seemingly at odds with theories of contextual threat. In a subsequent study, Barth et al. (2009) find that impersonal and superficial interactions with members of an outgroup that result from their presence in the general community increase negative attitudes, and that voluntary contact (e.g., having close friends who are gays or lesbians) has a much more powerful effect on attitudes than involuntary contact (e.g., with gay or lesbian family or co-workers).

Prestwich et al. (2008) find that contact with outgroup members produces more favorable explicit attitudes and can produce fairly rapid change but that shifting implicit attitudes is a slower process that requires repeated contact. Similarly, Lemm finds that even among individuals who explicitly “disavow prejudice against gay people, bias may be present at an implicit or unconscious level” (2006: 80). She finds no relationship between contact with gay or lesbian individuals and external motivation to be non-prejudiced; this relationship is moderated by explicit attitudes but not by implicit attitudes toward homosexuals.

See Moreno and Bodenhausen (2001) regarding individuals with egalitarian beliefs and their likelihood to display discriminatory behavior when provided a socially acceptable way to do so. Hegarty et al. (2004) find that heterosexual participants have demonstrated ambivalent reactions to expressions of discomfort displayed by gay and straight targets in gay or straight social situations. These findings suggest that when an opportunity presents itself (i.e., appropriate amount of interpersonal space, length of conversation, amount of criticism leveled, evaluation of rudeness, judgment of responsibility, and simple liking or disliking), implicit forms of sexual prejudice can manifest. Individuals who claim to be supporters of gay rights and marriage equality may still hold implicit biases, biases that come into play when asked to take action.