Motivated Independence? Implicit Party Identity Predicts Political Judgments Among Self-Proclaimed Independents

Abstract

Reporting an Independent political identity does not guarantee the absence of partisanship. Independents demonstrated considerable variability in relative identification with Republicans versus Democrats as measured by an Implicit Association Test (IAT; M = 0.10, SD = 0.47). To test whether this variation predicted political judgment, participants read a newspaper article describing two competing welfare (Study 1) or special education (Study 2) policies. The authors manipulated which policy was proposed by which party. Among self-proclaimed Independents, those who were implicitly Democratic preferred the liberal welfare plan, and those who were implicitly Republican preferred the conservative welfare plan. Regardless of the policy details, these implicit partisans preferred the policy proposed by “their” party, and this effect occurred more strongly for implicit than explicit plan preference. The authors suggest that implicitly partisan Independents may consciously override some partisan influence when making explicit political judgments, and Independents may identify as such to appear objective even when they are not.

Keywords

identity implicit cognition partisanship political ideology political judgment

People sometimes say one thing but do another. In politics, party affiliations have a pervasive impact on judgment (Campbell, Converse, Miller, & Stokes, 1960; Cohen, 2003; Kaplan, Freedman, & Iacoboni, 2007), but a large proportion of the electorate report being “Independent.” An Independent identity denotes the absence of partisan influence and implies objective political decision making. How do Independents rise above the fray and remain uninfluenced by political parties? We suggest that they do not.

Current Understanding of Political Independence

As a group, U.S. Independents’ political positions and voting, on average, are moderate—somewhere between the two dominant political party platforms (Lewis-Beck, Norpoth, Jacoby, & Weisberg, 2008). However, this may imply a more nonpartisan description than Independents deserve. A sizable portion of Independents qualify as “Independent leaners” who will, when pressed, acknowledge that they feel closer to either the Democratic or Republican party (Greene, 1999; Keith et al., 1992). In their voting patterns, Independent leaners are almost indistinguishable from their respective partisan blocs, even though they decline to identify as party members (Dennis, 1988; Keith et al., 1992). As such, some partisanship can be identified in Independents through more intensive questioning. In this article, we show that even more partisanship can be identified implicitly—without relying on Independents’ ability or willingness to report it.

Understanding the motivations and political behavior of Independents has practical and theoretical value. Since 1990, the percentage of the electorate who identify as Independent in the American National Election Studies (2010) ranged from 34% to 40%. In fact, Independents have outnumbered Republicans since the mid-1960s, and in some years have outnumbered Democrats. Many researchers treat Independents as “error” in the investigation of political behavior, and often exclude nonpartisans from data sampling or analysis (e.g., Cohen, 2003; Kaplan et al., 2007) or only use them as comparisons to avowed partisans, implicitly or explicitly granting them nonpartisan status (e.g., Garst & Bodenhausen, 1996; Smith, Ratliff, & Nosek, 2012). Understanding Independent identity offers an opportunity for a more comprehensive understanding of political behavior. Because Independent voters are often the “swing” vote, identifying variation in partisanship among Independents has the potential to improve party-based political prediction. Party cues play an important role in political behavior, often shaping the expression of political values (Goren, Federico, & Kittilson, 2009) and policy preferences (Cohen, 2003; Rahn, 1993), especially when policy information is scant (Bullock, 2011). Furthermore, studying Independents provides a unique opportunity to understand the conditions under which implicit and explicit identities diverge, and the role each plays in judgment and behavior.

Self-Proclaimed Independents May Possess Implicit Partisan Identities

Identities are adopted for a variety of reasons that serve social and personal purposes (Brewer, 1991; Hogg, Terry, & White, 1995). Independents may self-identify as such because they want to convey the social values of objectivity and the ability to think about issues based on the facts rather than just going with a group. In a pilot test, we surveyed 362 self-identified Independents at Project Implicit (https://implicit.harvard.edu/) about their reasons for identifying as Independent. Two statements denoting objectivity were among the most strongly endorsed of 35 reasons: “I prefer to think for myself rather than feel like I need to support a party line” and “I say ‘independent’ because I come to my political positions by thinking objectively” (see Table 1 for average endorsement of the full list of reasons). By comparison, far less than half endorsed political apathy (I don’t care about politics) and political hostility (I hate politics; I hate politicians). An item indicating moderation (One party is too liberal for me, and the other party is too conservative) was generally not endorsed, providing further evidence that the popular hypothesis that most Independents are moderates is incorrect, or at least incomplete. Independence is desired perhaps because of the United States’ individualist culture (Triandis, Bontempo, Villareal, Asai, & Lucca, 1988), where autonomy of thought and action is culturally valued.

Table 1.

Reasons for Identifying as Independent

	Average	% agree	% disagree
I vote for the person, not the party	5.9	88	4
I prefer to think for myself rather than feel like I need to support a party line	5.8	87	4
My views do not fit into any one ideology or party platform	5.4	79	6
I’m independent by nature	5.3	77	6
I say “independent” because I come to my political positions by thinking objectively	5.3	75	7
I value freedom from party membership	5.2	69	5
Political parties are inherently flawed	5.2	71	11
My beliefs overlap with multiple parties so it doesn’t make sense to identify with just one	5.2	74	13
I don’t like to label myself	5.0	67	16
Both parties represent interest groups rather than concerns that I care about	4.9	63	14
I do not like political parties	4.8	62	16
Neither party represents the people	4.8	59	20
My personal beliefs do not overlap enough with any party	4.7	59	19
Being an Independent makes me feel good	4.6	43	11
I strongly identify with being an Independent	4.5	46	21
I don’t feel I have a voice in current state of politics	4.4	52	32
I prefer one party on most issues, but the other party on at least one issue that is really important to me	4.2	48	27
I feel an affiliation with others who are Independent	4.2	37	23
One party is too liberal for me, and the other party is too conservative	4.1	38	33
I hate politicians	4.0	39	37
I support people who reject the current political system	3.9	31	33
I say “independent” to voice my displeasure with the current political system	3.8	35	42
I hate politics	3.7	40	49
I regularly alternate voting for candidates from different parties	3.7	30	41
The primary parties are too conservative for me, so I identify with neither	3.6	46	22
I used to affiliate with a party, but have since changed my identification	3.5	48	34
Being a member of a political party distracts from being a good citizen of the country	3.5	49	28
I don’t care about politics	3.4	30	57
I do not understand the party positions clearly enough to decide whether to be a member	3.3	31	56
I say “independent” to avoid being dragged into political activity or debates	3.3	29	57
I prefer to vote for candidates from minor parties	3.2	10	50
The primary parties are too liberal for me, so I identify with neither	3.1	13	56
I say “independent” instead of identifying with my preferred party so that I don’t offend some other people in my life	2.5	13	73
I say “independent” instead of identifying with my preferred party because I don’t want others to know my political beliefs	2.5	11	76
I like to vote for the underdog regardless of what party the person belongs to	2.3	4	79

Note: 1 = strongly disagree to 7 = strongly agree.

Independents may strive for objectivity and nonpartisanship in political matters, but identifying as Independent and being Independent are two different things. After all, people are reluctant to see bias in their own decision making (Ehrlinger, Gilovich, & Ross, 2005; Pronin, Lin, & Ross, 2002), and behavior is influenced by automatic processes that occur without conscious awareness, intention, or control (Bargh, 1994; Greenwald & Banaji, 1995). Indirectly measured implicit identities reflect associations between concepts (e.g., Democrat or Republican) and the self (Greenwald et al., 2002), and may be distinct from self-reported, explicit identities. Self-proclaimed Independents may, as a group, show substantial variation in their implicit political identities because of differential experience with politics and political parties, and that implicit identity might, in turn, influence political judgment.

Implicit evaluations have been shown to predict political judgment and behavior beyond that accounted for by explicit evaluations (Nosek, Graham, & Hawkins, 2010). Implicit candidate and party evaluations added incremental predictive validity over explicit evaluations and voting intention for predicting voting behavior (Friese, Bluemke, & Wanke, 2007; Rocatto & Zogmeister, 2010). Among a group of undecided voters, implicit candidate evaluations predicted later vote choice (Arcuri, Castelli, Galdi, Zogmaister, & Amadori, 2008; Study 2) and implicit evaluations of a particular policy predicted later explicit evaluations of the policy (Galdi, Acuri, & Gawronski, 2008). Whereas undecided voters may avoid committing on a single policy or particular candidate, Independents avoid committing between political identities more generally. Thus, demonstrating implicit partisan identities among Independents could have substantial impact on predicting political behavior and broaden understanding of divergent identities. We hypothesized that Independents would demonstrate variation in implicit identification with Democrats versus Republicans, and their implicit party identity would predict political judgment along party lines.

Study 1

Method

Participants

Eighteen hundred sixty-five (67% female; M _age = 30.03, SD _age = 12.31) U.S. citizens volunteered, consented, and completed the study at the Project Implicit website (https://implicit.harvard.edu).¹ The racial composition of the sample was 77% White, 7% Black or African American, 7% biracial or multiracial, 3% Asian, and 5% Other. Nine percent reported their ethnicity as Hispanic or Latino, 83% as non-Hispanic, and 8% Unknown or Other. Twelve percent had no college, 50% had some college or an associate’s degree, 18% had a bachelor’s degree, and 21% had some graduate school or an advanced degree (percentages do not always add up to 100% due to rounding).

Materials

Newspaper article with welfare plans

A mock newspaper article presented competing welfare plans as an amendment to a state’s welfare program (Cohen, 2003; Smith et al., 2012). The Umbrella Aid Plan was generous and provided US$976 a month to families with two children with full Medicaid coverage for 8 years. The Comprehensive Assistance Plan was stringent and provided US$300 a month to families with two children with partial Medicaid coverage for 1.5 years (see Appendix A for full article).

Explicit measures

Participants reported explicit welfare plan preference on a 7-point scale ranging from “I strongly prefer the Comprehensive Assistance Plan to the Umbrella Aid Plan” (−3) to “I strongly prefer the Umbrella Aid Plan to the Comprehensive Assistance Plan” (3). Political ideology was assessed on a “Strongly Conservative” (−3) to “Strongly Liberal” (3) scale with two separate items for social and economic ideology. Participants selected their party identification from the following options: Democrat, Republican, Independent—I don’t identify with either party, Libertarian, Green, Other, Don’t know. Participants who selected Independent were asked “If you had to choose between Democrats and Republicans, how would you identify your political affiliation?” Reponses were reported on a 7-point scale ranging from “Strongly Republican” (−3) to “Strongly Democratic” (3) with Independent as the neutral point.²

Implicit measures

Implicit plan preference was assessed with an Implicit Association Test (IAT; Greenwald, McGhee, & Schwartz, 1998) using the procedure recommended by Nosek, Greenwald, and Banaji (2007). It contained two short practice blocks of categorizing items from the concept categories Umbrella Aid Plan (stimuli: Umbrella Aid, Full Medicaid coverage, and US$976/month) and Comprehensive Assistance Plan (stimuli: Comprehensive Assistance, Partial Medicaid coverage, and US$300/month), or just the evaluative categories Good (Joy, Love, Peace, Wonderful, Pleasure, Glorious, Laughter, and Happy) and Bad (Agony, Terrible, Horrible, Nasty, Evil, Awful, Failure, and Hurt). Then, in two combined-response blocks (20 trials; 40 trials), participants simultaneously categorized stimuli from one concept category (Umbrella Aid Plan) and one evaluation category (Good) using a response key (“i”) and the other concept category (Comprehensive Assistance Plan) and evaluation category (Bad) using a different response key (“e”). Next, a practice block of 40 trials reversed the response keys of the concept categories. In the next two critical blocks (20 trials; 40 trials), the concept categories shared the same response key with the opposite evaluation (e.g., Umbrella Aid Plan and Bad shared the “e” response key, Comprehensive Assistance Plan and Good shared the “i” response key). The party identification IAT (Graham, Haidt, & Nosek, 2009; Lindner & Nosek, 2009) used the concept categories Democrats (left wing, liberal, Barack Obama, and Democrat) and Republicans (right wing, conservative, George Bush, and Republican), and identification categories Self (Mine, Myself, Self, I, and My) and Other (They, Their, Them, Theirs, and Other).

The response key assignments for the combined-response blocks were randomized between subjects. An IAT D score reflected the recommended scoring algorithm for the assessed response latencies by condition (Greenwald, Nosek, & Banaji, 2003). Positive values indicated an implicit preference for the generous plan compared with the stringent plan and an implicit identification with Democrats relative to Republicans, reflecting faster response latencies when Umbrella Aid Plan and Good shared a response key and Democrats and Self shared a response key.

Follow-up questions and attention checks

Two follow-up questions asked participants to estimate on a scale from “did not contribute at all” (1) to “contributed a great deal” (5) the extent to which (a) the specific details of the proposals and (b) their own personal philosophy about the role of government in social issues, contributed to their preference for the welfare policies. The third item assessed how much the political party affiliated with the welfare proposals influenced their preference and was answered on a scale from “no influence” (1) to “a great deal of influence” (5). Two study-specific attention checks measured what party proposed each plan and a third required participants to report which plan was more generous.³ Also, a modified Instructional Manipulation Check (IMC; Oppenheimer, Meyvis, & Davidenko, 2009) assessed how carefully participants read study instructions.

Design and Procedure

Visitors to Project Implicit registered to be members of the participant pool, completed a demographics questionnaire, and were randomly assigned to the present study from a pool of studies. Participants were randomly assigned to read one of only two newspaper articles: (a) an article where Democrats proposed the generous plan and Republicans proposed the stringent plan or (b) an article where Republicans proposed the generous plan and Democrats proposed the stringent plan. Other than the party name and partisan designation of the supporting politicians, the articles were identical. Next, they rated their implicit and explicit preference for the two plans in a randomized order. Participants then completed the follow-up questions, then the attention checks, followed by the political ideology and identity questions and party identity IAT in a randomized order. The IMC was counterbalanced at the beginning or end of the study.

Results

Descriptive statistics

Removing participants who failed the study-specific attention checks or IMC did not change the statistical significance of the main analyses, so all participants were retained.⁴ On average, participants’ political ideology was moderate on economic issues (M = 0.06, SD = 1.67) and slightly liberal on social issues (M = 0.88, SD = 1.90), and they implicitly identified with Democrats relative to Republicans (M = 0.18, SD = 0.51). Participants preferred the generous plan to the stringent plan both explicitly (M = 0.14, SD = 2.02) and implicitly (M = 0.20, SD = 0.46), and explicit and implicit plan preference were correlated, r(1865) = .46, p < .0001. See Table 2 for descriptive statistics for each political group.

Table 2.

Descriptive Statistics for Studies 1 and 2

	Study 1 (N = 1,865)			Study 2 (N = 1,911)
	Dems (n = 593)	Inds (n = 625)	Reps (n = 304)	Dems (n = 600)	Inds (n = 617)	Reps (n = 339)
Economic political ideology	1.07 (1.36)	−0.08 (1.43)	−1.45 (1.29)	1.23 (1.27)	−0.14 (1.50)	−1.67 (1.23)
Social political ideology	1.90 (1.37)	0.76 (1.75)	−0.96 (1.81)	2.00 (1.31)	0.95 (1.72)	−1.37 (1.62)
Implicit party identity	0.48 (0.39)	0.17 (0.45)	−0.36 (0.38)	0.48 (0.38)	0.10 (0.47)	−0.49 (0.39)
Explicit plan preference	0.82 (1.91)	0.09 (1.96)	−0.88 (1.95)	0.63 (1.99)	0.23 (1.93)	−0.17 (1.93)
Implicit plan preference	0.28 (0.46)	0.20 (0.45)	0.06 (0.48)	0.19 (0.39)	0.16 (0.39)	0.10 (0.38)

Note: Dems = Democrats; Inds = Independents; Reps = Republicans. Higher numbers reflect identification as liberal and Democratic compared with conservative and Republican. Standard deviations are in parentheses. In Study 1, higher numbers reflect preference for the generous Umbrella Aid Plan compared with the stringent Comprehensive Assistance Plan. In Study 2, higher numbers reflect preference for mainstreaming (Integrated Classrooms Plan) versus special programs (Special Programs Plan).

Partisans’ plan preference

As seen in Figure 1, Democrats preferred the generous welfare plan more when it was proposed by Democrats (M_exp = 1.02, SD_exp = 1.86; M_imp = 0.42, SD_imp = 0.40) than when it was proposed by Republicans (M_exp = 0.62, SD_exp = 1.95; M_imp = 0.13, SD_imp = 0.46) both explicitly, t(591) = 2.53, p = .012, d = 0.21, and implicitly, t_{satterthwaite}(573) = 8.27, p < .0001, d = 0.69. Likewise, Republicans liked the generous welfare plan more (or disliked it less) when it was proposed by Republicans (M_exp = −0.54, SD_exp = 2.03; M_imp = 0.15, SD_imp = 0.46) than by Democrats (M_exp = −1.26, SD_exp = 1.77; M_imp = −0.04, SD_imp = 0.48) both explicitly, t(302) = −3.29, p = .001, d = −0.38, and implicitly, t(302) = −3.41, p = .001, d = −0.39.

Figure 1.

The effect of partisan influence for explicit (A) and implicit (B) plan preference in Study 1

Independents’ plan preference

By definition, Independents explicitly decline to identify as party members. However, there was substantial interindividual variability in implicit party identity (SD = 0.45). We conducted two regression analyses with explicit and implicit plan preference as the dependent variables and the proposing party (−.5 represented the condition where Republicans proposed the generous plan and .5 represented the condition where Democrats proposed the generous plan), implicit party identity, and their interaction as independent variables. The dependent variables are on different scales, so unstandardized coefficients should be interpreted in terms of their respective distributions. CI’s represent 95% confidence intervals.

Independents did not show a main effect of proposing party predicting either explicit or implicit plan preference (ps > .15). However, a main effect of implicit party identity emerged for both explicit, b = 1.18, CI = [0.83, 1.52], t(548) = 6.73, p < .0001, and implicit, b = 0.17, CI = [0.09, 0.25], t(548) = 4.35, p < .0001, plan preference. Implicitly Democratic Independents preferred the generous welfare plan to the stringent welfare plan, and vice versa for implicitly Republican Independents. Moreover, an interaction between proposing party and implicit party identity predicted explicit, b = 1.08, CI = [0.39, 1.77], t(548) = 3.08, p = .002, and implicit, b = 0.60, CI = [0.44, 0.75], t(548) = 7.53, p < .0001, plan preference. Implicitly Republican Independents explicitly preferred the plan that was proposed by Republicans, and implicitly Democratic Independents preferred the plan that was proposed by Democrats (d = 0.26; Figure 2A). Implicit plan preference showed the same pattern, with an effect magnitude more than twice as large (d = 0.64; Figure 2B). These results illustrate two influences of implicit party identity—Independents prefer policies that are ideologically aligned with their implicit identity (implicit Republicans prefer the more stringent welfare plan and Democrats prefer the more generous plan regardless of which party proposed it), and Independents prefer policies that are proposed by the party with which they identify implicitly.⁵

Figure 2.

Regression analysis predicting explicit (A) and implicit (B) preference for the welfare policies from implicit party identity and the party proposing the plan for Independents in Study 1

Pure Independents’ plan preference

We repeated the analysis above on the 214 (34%) “Pure Independents,” who reported no inclination toward either major party even after being pressed to do so. A main effect emerged for proposing party on explicit plan preference, b = 0.64, CI = [0.05, 1.22], t(182) = 2.16 p = .032, but not implicit plan preference, b = 0.07, CI = [−0.07, 0.20], t(182) = 0.99, p = .325. Explicitly, Pure Independents preferred the generous plan more when it was proposed by Democrats (M = 0.52, SD = 1.96) than when it was proposed by Republicans (M = −0.19, SD = 1.95). Also, there was a main effect of implicit party identity for both explicit, b = 1.24, CI = [0.60, 1.88], t(182) = 3.83, p = .0002, and implicit, b = 0.17, CI = [0.02, 0.31], t(182) = 2.29, p = .023, plan preference. Implicitly Democratic Pure Independents preferred the generous welfare plan, and implicitly Republican Pure Independents preferred the stringent welfare plan. However, for Pure Independents, there was no interaction between the proposing party and implicit party identity for explicit plan preference, b = 0.27, CI = [−1.01, 1.54], t(182) = 0.41, p = .682. For implicit plan preference, the interaction was moderately sized but not significant, b = 0.29, CI = [0.00, 0.58], t(182) = 1.95, p = .053.

Incremental predictive validity

Another approach to illustrate the value of implicit party identity is to test whether it has incremental predictive validity over the existing self-report measures of political identity—explicit party identification, political ideology, and Independents’ leaning party membership. Also, participants reported how much the party proposing the plan influenced their plan preference, and this is important to control for as an Independent might decline to identify with either party but still believe that one party is better at dealing with a particular political issue than the other. Can implicit party identity predict variation in political judgment beyond all of these explicit measures?

To answer this question, we conducted two hierarchical regression analyses with explicit and implicit plan preference as the dependent variables on the full sample of Independents. The independent variables were simultaneously entered into the regression analyses: main effects of proposing party, self-rating of party influence, “leaning” party identification, and self-reported political ideology. See Table 3 for coefficients for each term in the models. Adding the main effect of implicit party identity accounted for an additional 1.4% of the variance in explicit plan preference, b = 0.65, CI = [0.27, 1.03], t(542) = 3.34, p = .001, and 1.0% of the variance in implicit plan preference, b = 0.12, CI = [0.03, 0.22], t(542) = 2.62, p = .009. We then added the two-way interactions between proposing party and self-rating of party influence, proposing party and “leaning” party identification, proposing party and self-reported political ideology, and finally, the interaction between proposing party and implicit party identity. This added an additional 0.7% of variance explained for explicit plan preference, b = 0.80, CI = [0.04, 1.56], t(538) = 2.08, p = .038, and 6.2% of additional variance explained for implicit plan preference, b = 0.56, CI = [0.39, 0.74], t(538) = 6.29, p < .0001. Even after accounting for self-reported political leanings and ideology, as well as the influence of the proposing party on judgment, implicit party identity still predicted preferences for the welfare plans—both as a main effect indicating a policy preference (for stringent or generous welfare plans) and as an interaction with the proposing party indicating partisan influence on judgment.

Table 3.

Incremental Predictive Validity Results for Independents From Studies 1 and 2

	Study 1 (n = 625)		Study 2 (n = 617)
	Explicit plan preference	Implicit plan preference	Explicit plan preference	Implicit plan preference
Model terms	SD = 2.02	SD = 0.46	SD = 1.97	SD = 0.39
Proposing party	0.41** [0.11, 0.70]	0.07* [0.00, 0.14]	0.35* [0.04, 0.66]	0.07* [0.01, 0.13]
Self-rating of party influence	−0.01 [−0.18, 0.16]	−0.02 [−0.06, 0.02]	0.19* [0.00, 0.37]	0.04 [0.00, 0.07]
Leaning party membership	0.33**** [0.18, 0.47]	0.04* [0.01, 0.08]	0.03 [−0.12, 0.18]	0.02 [−0.01, 0.05]
Political ideology	0.32**** [0.16, 0.47]	0.05* [0.01, 0.08]	0.00 [−0.16, 0.15]	0.01 [−0.02, 0.04]
R² for explicit main effects	13.2%	5.0%	1.6%	2.5%
Implicit party identity	0.65*** [0.27, 01.03]	0.12** [0.03, 0.22]	0.14 [−0.25, 0.53]	0.07 [−0.01, 0.15]
R² for all main effects	14.6%	6.0%	2.4%	2.9%
Proposing party × Self-rating of party influence	0.53** [0.16, 0.90]	0.07 [−0.02, 0.16]	0.50* [0.11, 0.90]	0.07 [0.00, 0.15]
Proposing party × Leaning party membership	0.39* [0.07, 0.71]	0.07 [0.00, 0.15]	0.69**** [0.38, 1.00]	0.09** [0.03, 0.15]
Proposing party × Political ideology	−0.08 [−0.41, 0.25]	0.06 [−0.02, 0.14]	−0.28 [−0.60, 0.04]	0.07* [0.00, 0.13]
R² for explicit interactions	17.4%	9.4%	7.2%	9.0%
Party proposer × Implicit party identity	0.80* [0.04, 1.56]	0.56**** [0.39, 0.74]	0.35 [−0.41, 1.11]	0.18* [0.03, 0.33]
R² for full model	18.0%	15.6%	7.3%	9.9%

Note: All bs are unstandardized regression coefficients and should be interpreted in the context of their individual distributions, which are indicated with SDs for the outcome variables for each study. Confidence intervals (±2 SE) are given in brackets. Study 1 (welfare plan) was designed to elicit both a main effect and interaction with party proposer of implicit party identity predicting both outcomes. Study 2 (education plans) was designed to elicit an interaction, but not a main effect.

p < .05. **p < .01. ***p > .001. ****p > .0001.

Awareness of partisan influence

Participants reported that the details of the welfare plans (M = 3.60, SD = 1.13) and their personal philosophy about the role of government in social issues (M = 3.18, SD = 1.21) influenced their preference for the welfare plans. However, participants reported that the party that proposed the policy had little influence on their preference (M = 1.92, SD = 1.08). In fact, the modal response for all political groups was no influence (Democrats: 37%; Independents: 62%; Republicans: 44%). To examine the accuracy of these reports among Independents, we tested whether awareness moderated the interaction between proposing party and implicit party identity. Because awareness had a significant skew with many people reporting “no influence,” we coded it as a categorical variable (−.5 for no influence and .5 for a little, some, quite a lot, and a great deal of influence) and ran two separate models, predicting both explicit and implicit plan preference. The three-way interaction between proposing party, implicit party identity, and awareness was not a significant predictor of either explicit or implicit plan preference (ps > .2). For explicit plan preference, the main effect of implicit party identity, b = 1.10, CI = [0.73, 1.47], t(542) = 5.81, p < .0001, and the interaction between proposing party and implicit party identity, b = 1.25, CI = [0.50, 1.99], t(542) = 3.30, p = .001, remained significant.⁶ Likewise, for implicit plan preference, the main effect of implicit party identity, b = 0.17, CI = [0.08, 0.25], t(542) = 3.90, p = .0001, and the interaction between proposing party and implicit party identity remained significant, b = 0.59, CI = [0.43, 0.76], t(542) = 6.92, p < .0001. These results suggest that Independents were influenced by the parties who proposed the policies, but did not report awareness of this influence.

Study 2

Overview

Study 1 showed that self-reported Independents demonstrated variation in implicit party identification, and this variation predicted political judgment along party lines. Although Pure Independents’ implicit party identities predicted their policy positions along ideological lines, they avoided using the proposing party to guide their policy preference—at least explicitly. This could indicate that Pure Independents are able to override or otherwise avoid some partisan influence on judgment. In Study 1, one plan was a generous, liberal plan, and the other was a stringent, conservative plan, and these provided ideological cues for evaluating the plans. If the ideological difference in the plans is removed or considerably reduced, then the only role that implicit party identity should be able to play is in producing an interaction with the party proposing the policy. To address this possibility and simultaneously replicate the key effects in another policy domain, we created an article that focused on a different policy debate—education for children with disabilities—that has less defined partisan positions than welfare.

Method

Participants

Nineteen hundred eleven (66% female; M_age = 29.23, SD_age = 11.82) participants completed the study on Project Implicit.⁷ The racial composition of the sample was 77% White, 7% Black or African American, 6% biracial or multiracial, 3% Asian, and 7% Other. Nine percent reported their ethnicity to be Hispanic or Latino, 86% as non-Hispanic or Latino, and 6% Other or Don’t know. Thirteen percent had no college, 47% had some college or an associate’s degree, 18% had a bachelor’s degree, and 22% had some graduate school or an advanced degree.

Materials and Procedure

The procedure was identical to Study 1, but the materials were edited to propose two education plans.⁸ The Special Programs Plan proposed that children with disabilities receive individualized instruction in special classrooms separate from mainstream classrooms. The Integrated Classrooms Plan proposed that children with disabilities be educated in mainstream classrooms alongside children without disabilities (see Appendix B for full article). Again, the only difference between conditions was the labels “Democrats” and “Republicans” and the partisan designation of the politicians (e.g., “D” or “R”). The plan preference IAT retained the evaluation categories Good and Bad, but the concept categories were changed to Special Programs (special programs, separate classrooms, and one-on-one instruction) and Integrated Classrooms (integrated classrooms, mainstream classrooms, and inclusive activities). In the follow-up questions, “welfare” was replaced with “education” and an extra follow-up question was added that asked whether participants preferred either special programs or mainstream classrooms prior to the study.⁹

Results

Descriptive statistics

Removing participants who failed the study-specific attention checks or the IMC did not change the statistical significance of any main findings, and thus all participants were retained.¹⁰ Higher numbers reflect preference for mainstreaming (Integrated Classrooms Plan) compared with special programs (Special Programs Plan). On average, participants were slightly liberal on social issues (M = 0.85, SD = 1.98) and moderate on economic issues (M = 0.05, SD = 1.73) and implicitly identified with Democrats slightly more than Republicans (M = 0.11, SD = 0.55). Participants slightly preferred the mainstreaming plan to the special programs plan both explicitly (M = 0.25, SD = 1.97) and implicitly (M = 0.15, SD = 0.39), and explicit and implicit plan preference were correlated, r(1911) = .39, p < .0001. The difference between Democrats and Republicans on preference for the education plans (d_exp = 0.39, p < .0001; d_imp = 0.23, p = .001) was about half the size of the partisan difference in preference for the generous and stringent welfare plans from Study 1 (d_exp = 0.84, p < .0001; d_imp = 0.44, p < .0001), suggesting that the special education plans were less partisan than the welfare plans. See Table 2 for descriptive statistics for each political group.

Partisans’ plan preference

Democrats preferred the mainstreaming plan when it was proposed by Democrats (M_exp = 1.05, SD_exp = 1.82; M_imp = 0.29, SD_imp = 0.36) more than when it was proposed by Republicans (M_exp = 0.20, SD_exp = 2.06; M_imp = 0.10, SD_imp = 0.40) both explicitly, t_{satterthwaite}(589) = 5.35, p < .0001, d = 0.44, and implicitly, t(598) = 6.07, p < .0001, d = 0.50. Similarly, Republicans preferred the mainstreaming plan when it was proposed by Republicans (M_exp = 0.48, SD_exp = 1.84; M_imp = 0.19, SD_imp = 0.36) more than when it was proposed by Democrats (M_exp = −0.78, SD_exp = 1.81; M_imp = 0.02, SD_imp = 0.37) both explicitly, t(337) = −6.37, p < .0001, d = −0.69, and implicitly, t(337) = −4.24, p < .0001, d = −0.45.

Independents’ plan preference

Identical analyses from Study 1 revealed no main effects of proposing party for either explicit or implicit plan preference (ps > .15). A main effect of implicit party identity emerged for implicit, b = 0.10, CI = [0.03, 0.16], t(574) = 2.91, p = .004, but not explicit, b = 0.19, CI = [−0.14, 0.52], t(574) = 1.14, p = .253, plan preference. Critically, interactions between proposing party and implicit party identity emerged for both explicit, b = 1.03, CI = [0.36, 1.69], t(574) = 3.05, p = .002, and implicit, b = 0.33, CI = [0.20, 0.46], t(574) = 5.04, p < .0001, plan preference. Independents who were implicitly Democratic preferred the mainstreaming plan more when it was proposed by Democrats than when it was proposed by Republicans, and Independents who were implicitly Republican showed the opposite tendency. As in Study 1, the interaction effect was larger for implicit (d = 0.42; Figure 3B) than explicit (d = 0.25; Figure 3A) plan preference.

Figure 3.

Regression analysis predicting explicit (A) and implicit (B) preference for the education policies from implicit party identity and the party proposing the plan for Independents in Study 2

Pure Independents’ plan preference

The same analysis was repeated in the restricted sample of Pure Independents (n = 204; 33%). No main effects of proposing party or implicit party identity emerged for either explicit or implicit plan preference (ps > .2). A significant interaction between proposing party and implicit party identity emerged for implicit, b = 0.36, CI = [0.10, 0.62], t(187) = 2.74, p = .007, but not explicit, b = 0.73, CI = [−0.67, 2.13], t(187) = 1.03, p = .304, plan preference. Pure Independents who were implicitly Democratic implicitly preferred the mainstreaming plan when it was proposed by Democrats more than when it was proposed by Republicans, and vice versa for Pure Independents who were implicitly Republican.

Incremental predictive validity

As in Study 1, we conducted two hierarchical regression analyses to determine whether implicit party identity added incremental predictive validity for predicting plan preference above and beyond explicit measures (see Table 3 for coefficients for each model term). For explicit plan preference, implicit party identity revealed no main effect or interaction with proposing party (ps > .36) with all the other effects already present in the models. The story was quite different for implicit plan preference. The independent variables were entered simultaneously into the regression analyses: main effects of proposing party, self-rating of party influence, “leaning” party identification, and self-reported political ideology. Adding the main effect of implicit party identity accounted for an additional 0.4% of the variance, b = 0.07, CI = [−0.01, 0.15], t(564) = 1.77, p = .078. We then added the two-way interactions between proposing party and self-rating of party influence, proposing party and “leaning” party identification, and proposing party and self-reported political ideology. Finally, we added the interaction between proposing party and implicit party identity, which added an additional 0.9% of variance explained, b = 0.18, CI = [0.03, 0.33], t(560) = 2.37, p = .018. Implicit party identity predicted political judgment above and beyond variation accounted for by self-rating of proposing party influence, political ideology, and assessments of “party leaning”—the present gold standard for identifying “closet partisanship” among Independents.

Awareness of partisan influence

As in Study 1, both partisans and Independents reported that the proposing party did not influence their plan preference. For each party, the modal response to the party influence item was no influence (Democrats: 44%; Independents: 68%; Republicans: 44%). Identical analyses to Study 1 revealed nonsignificant three-way interactions predicting both explicit and implicit plan preference (ps > .07). The two-way interaction between proposing party and implicit party identity remained a significant predictor of both explicit plan preference, b = 1.10, CI = [0.39, 1.81], t(567) = 3.04, p = .003, and implicit plan preference, b = 0.34, CI = [0.20, 0.48], t(567) = 4.78, p < .0001. Despite making judgments based on their party affiliation, Independents were unable, or perhaps unwilling, to report that they were influenced by their implicit party identities.

Replication Studies

Three replication studies using the special education plans echoed the findings of Studies 1 and 2. In the first replication, Independents’ (n = 249) implicit party identity interacted with proposing party to predict implicit plan preference, b = 0.44, CI = [0.23, 0.65], t(222) = 4.22, p < .0001, but not explicit plan preference, b = 0.02, CI = [−1.02, 1.06], t(222) = 0.03, p = .974. Similarly, for incremental predictive validity, the interaction between implicit party identity and proposing party significantly predicted implicit plan preference, even after accounting for self-reported leaning party membership, party influence, and political ideology, b = 0.46, CI = [0.21, 0.71], t(214) = 3.57, p < .001. However, the interaction did not significantly predict explicit plan preference, b = −0.17, CI = [−1.45, 1.11], t(214) = −0.26, p = .794, when the self-reported variables were already in the model.

In the second replication study, Independents’ (n = 289) implicit party identity interacted with proposing party to predict both explicit plan preference, b = 1.36, CI = [0.38, 2.34], t(261) = 2.74, p = .007, and implicit plan preference, b = 0.37, CI = [0.16, 0.57], t(261) = 3.54, p = .001. The interaction between implicit party identity and proposing party provided incremental predictive validity for implicit plan preference, b = 0.31, CI = [0.07, 0.55], t(250) = 2.57, p = .011, but not explicit plan preference, b = 0.92, CI = [−0.21, 2.05], t(250) = 1.61, p = .109.

In the previous studies, the education plans and preference for those plans always appeared prior to assessment of implicit party identity. This order was intended to avoid Independents’ awareness that their political identity was of particular interest in their subsequent evaluation of the plans. However, given our interpretation, having the implicit measures at the end of the session implicitly assumes that implicit party identity would not be affected by the experimental manipulations. To assess whether the key effect was dependent on task order, in the third replication of Study 2, we manipulated whether the education policy preferences or political party identity was assessed first. Among Independents (n = 394), implicit party identity interacted with proposing party to predict both explicit plan preference, b = 0.79, CI = [0.06, 1.52], t(370) = 2.13, p = .034, and implicit plan preference, b = 0.27, CI = [0.11, 0.42], t(370) = 3.33, p = .001. Task order (coded −.5 for article first, .5 for party identity first) did not moderate this effect as evidenced by the nonsignificant three-way interaction between task order, implicit party identity, and proposing party on explicit plan preference, b = −0.40, CI = [−1.89, 1.09], t(366) = −0.53, p = .597, and implicit plan preference, b = −0.21, CI = [−0.53, 0.11], t(366) = −1.32, p = .188. Even with the nonsignificant three-way interaction in the models, the two-way interaction between implicit party identity and proposing party remained significant predictors of both explicit plan preference, b = 0.84, CI = [0.09, 1.58], t(366) = 2.21, p = .027, and implicit plan preference, b = 0.28, CI = [0.12, 0.44], t(366) = 3.44, p < .001. The interaction did not provide significant incremental predictive validity for explicit or implicit plan preference, but the results were in the expected direction as occurred in the other studies (see Table 4).

Table 4.

Summary of Key Results for Independents From Studies 1 and 2 and Replication Studies 3, 4, & 5

	Study 1	Study 2	Study 3	Study 4	Study 5
Explicit DV	SD = 2.02	SD = 1.97	SD = 1.95	SD = 1.94	SD = 1.95
Implicit DV	SD = 0.46	SD = 0.39	SD = 0.40	SD = 0.39	SD = 0.40
Independents	n = 625	n = 617	n = 249	n = 289	n = 394
Implicit party identity main effect
Explicit DV	1.18**** [0.83, 1.52]	0.19 [−0.14, 0.52]	−0.04 [−0.56, 0.48]	0.44 [−0.05, 0.93]	0.18 [−0.19, 0.54]
Implicit DV	0.17**** [0.09, 0.25]	0.10** [0.03, 0.16]	−0.03 [−0.13, 0.07]	0.04 [−0.06, 0.14]	0.09* [0.01, 0.17]
Incremental predictive validity for implicit party identity main effect
Explicit DV	0.65*** [0.27, 1.03]	0.14 [−0.25, 0.53]	−0.22 [−0.85, 0.41]	0.73* [0.17, 1.29]	0.02 [−0.41, 0.45]
Implicit DV	0.12** [0.03, 0.22]	0.07 [−0.01, 0.15]	0.06 [−0.07, 0.19]	0.13* [0.01, 0.25]	0.05 [−0.04, 0.14]
Party proposer × Implicit party identity interaction
Explicit DV	1.08** [0.39, 1.77]	1.03** [0.36, 1.69]	0.02 [−1.02, 1.06]	1.36** [0.38, 2.34]	0.79* [0.06, 1.52]
Implicit DV	0.60**** [0.44, 0.75]	0.33**** [0.20, 0.46]	0.44**** [0.23, 0.65]	0.37*** [0.16, 0.57]	0.27** [0.11, 0.42]
Incremental predictive validity for party proposer × Implicit party identity interaction
Explicit DV	0.80* [0.04, 1.56]	0.35 [−0.41, 1.11]	−0.17 [−1.45, 1.11]	0.92 [−0.21, 2.05]	0.37 [−0.48, 1.21]
Implicit DV	0.56**** [0.39, 0.74]	0.18* [0.03, 0.33]	0.46*** [0.21, 0.71]	0.31* [0.07, 0.55]	0.13 [−0.05, 0.31]
Pure independents	n = 214	n = 204	n = 89	n = 74	n = 171
Implicit party identity main effect
Explicit DV	1.24*** [0.60, 1.88]	0.06 [−0.64, 0.76]	−0.43 [−1.40, 0.53]	0.87 [−0.22, 1.96]	0.32 [−0.28, 0.93]
Implicit DV	0.17* [0.02, 0.31]	0.08 [−0.05, 0.21]	−0.07 [−0.26, 0.11]	0.22* [0.00, 0.44]	0.12 [−0.02, 0.25]
Party proposer × Implicit party identity interaction
Explicit DV	0.27 [−1.01, 01.54]	0.73 [−0.67, 2.13]	−0.09 [−2.02, 1.84]	1.09 [−1.10, 3.27]	0.34 [−0.87, 1.54]
Implicit DV	0.29 [0.00, 0.58]	0.36** [0.10, 0.62]	0.53** [0.16, 0.91]	0.56* [0.12, 1.00]	0.04 [−0.22, 0.31]

Note: DV = dependent variable. All bs are unstandardized regression coefficients and should be interpreted in the context of their individual distributions, which are indicated with SDs for the outcome variables for each study. Confidence intervals (±2 SE) are given in brackets. Study 1 (welfare plan) was designed to elicit both a main effect and interaction with party proposer of implicit party identity predicting both outcomes. Studies 2 to 5 (education plans) were designed to elicit an interaction, but not a main effect.

p < .05. **p < .01. ***p > .001. ****p > .0001.

Three replication studies affirmed the key effects from Studies 1 and 2 (see Table 4 for summary statistics on all five studies). Independents’ implicit party identity interacted with proposing party to predict plan preference, and more strongly for implicit than explicit plan preference. One effect did not replicate in the third replication study—an interaction between implicit identity and proposing party predicting implicit plan preference for Pure Independents. Suggestive evidence, elaborated in the “General Discussion,” indicates that Pure Independents’ political judgments might contrast away from their implicit party identity when made aware that their political identity is of interest, and assimilate to their implicit party identity otherwise—as occurred for Independents in general.

General Discussion

Independents are not as independent as they say they are. Self-proclaimed Independents show considerable variation in their implicit party identities and make partisan political judgments in line with those implicit identities. In both studies, implicitly Democratic Independents preferred policies proposed by Democrats and implicitly Republican Independents preferred policies proposed by Republicans, and more strongly for implicit than explicit plan preference. In short, people who otherwise reported being nonpartisan formed positions on novel policies that conformed to their implicit party identities. Furthermore, partisans and Independents—but Independents especially—reported that they were not influenced by the parties proposing the policies, indicating partisan influences on judgment that people are unwilling or unable to report.

Differential Prediction of Implicit and Explicit Plan Preference

The influence of implicit party identity was particularly pronounced on implicit policy evaluations compared with explicit policy evaluations. Among Pure Independents, implicit party identity interacted with proposing party to predict only implicit, and not explicit, plan preference (significantly in Study 2 and just shy of conventional statistical significance in Study 1, p = .053). One interpretation is procedural—implicit measures are better at predicting implicit measures than explicit measures because of parallel procedural details. However, in line with related research, we suggest that the explanation is more substantive than procedural.

Dual process models in social psychology (Gawronski & Bodenhausen, 2006; Strack & Deutsch, 2004) hold that implicit identity reflects associative processes, whereas explicit identity reflects both associative and propositional processes. Independents’ explicit identities are partially formed with a conscious goal to be political citizens who are not swayed by partisan positions. However, implicit identities are the result of associative processes linking the self with one or the other political party. Such associations may emerge through a variety of experiences that are unrelated to the beliefs and intentions for social identification, such as early socialization, exposure to political advertisements or campaigns, or the partisan positions of friends or coworkers.

There is some evidence that implicit processes exert less influence on judgments and behavior if people are motivated to control their behavior (Fazio, 1990). Independents in general, and Pure Independents in particular, are motivated to be independent and, according to our initial survey, to be objective thinkers who are unmoved by party platforms and endorsements. As such, the motivations may have produced weaker relations between implicit party identity and explicit plan preferences. This suggests that Pure Independents have some capacity to subvert party influence—just as their deliberate self-identification would suggest. However, over time, the ability to deliberately correct for such influence may fade as the details of the learning episode are forgotten. Ranganath and Nosek (2008) found that participants who were initially able to avoid applying implicit evaluations to explicit judgment were not able to avoid the influence after a few days passed. Without the available memory details of the learning event, all that may remain to drive judgment is the associative relationship between implicit party identity and the proposed policies (Smith et al., 2012).

If suppression or reinterpretation processes are involved in partially reducing the influence of implicit party identity, then perhaps such processes can be adapted and practiced so that they nullify partisan influence. Finding such evidence would have theoretical implications in showing the boundary conditions of deliberate strategies for redirecting or altering the effects of implicit processes, and practical implications in providing decision makers with strategies to use when impartiality, objective thinking, or avoiding the influence of group memberships or party loyalties is at a premium. For instance, could a judge use a mental partisan-cleansing exercise prior to gaveling her court into session that would align her mental operations with her intentions and occupational responsibility—to be an impartial arbiter of the law?

Increasing Predictive Validity of Political Judgment

Accounting for Independents’ interests, loyalties, and likely voting can be enhanced by incorporating implicit measures into assessment. In both studies, we found that even after accounting for the influence of proposing party and explicit factors (leaning party membership and reported party influence on judgment), implicit party identity predicted implicit plan preference along party lines. Predictive validity for explicit plan preference was also increased by accounting for implicit party identity, though weakly and less consistently than implicit plan preference. Furthermore, to our knowledge, this is the first demonstration of predicting political judgment among Pure Independents. This demonstrates incremental predictive validity of political judgment for implicit party identity above and beyond explicit predictors. Already, implicit measures assessing racial attitudes—the IAT and Affective Misattribution Procedure (AMP; Payne, Cheng, Govorun, & Stewart, 2005)—have appeared in the American National Election Studies. Also, the Brief IAT (Sriram & Greenwald, 2009) and the AMP can be administered in just a few minutes, improving the potential applicability in large-scale survey contexts. Implicit measurement in studies of political opinion with representative samples will enhance the understanding of implicit measurement in general and specifically about partisanship among Independents.

Further Speculation About Pure Independents’ Resistance to Partisan Influence

The third replication study demonstrated that the order of task presentation did not moderate the key results for the full sample of Independents. This suggests that Independents’ implicit party identity does not shift as a function of making political judgments and, likewise, that the political judgments are not altered by first measuring implicit party identity. However, the restricted sample of Pure Independents (n = 171) in the third replication study did appear to be influenced by the order of the tasks—The three-way interaction between implicit party identity, proposing party, and task order was near significant for explicit plan preference, b = −2.32, CI = [−4.73, 0.09], t(152) = −1.90, p = .059, and significant for implicit plan preference, b = −0.65, CI = [−1.19, −0.11], t(152) = −2.39, p = .018. Furthermore, the original two-way interactions were rendered nonsignificant (ps > .5) when task order was added to the model, suggesting that for Pure Independents, the order of the tasks moderated the extent to which implicit party identity predicted partisan political judgment. Follow-up tests revealed that when policy preferences were assessed prior to implicit party identification, the two-way interaction between implicit party identity and proposing party was a marginally significant predictor of both explicit plan preference, b = 1.53, CI = [−0.19, 3.26], t(75) = 1.77, p = .081, and implicit plan preference, b = 0.38, CI = [−0.02, 0.78], t(75) = 1.87, p = .065, with similar effect magnitudes as the prior studies (see Table 4). However, when implicit party identity was assessed prior to encountering the policies, the two-way interaction between implicit party identity and proposing party was not a significant predictor of either explicit plan preference, b = −0.79, CI = [−2.50, 0.92], t(77) = −0.92, p = .361, or implicit plan preference, b = −0.27, CI = [−0.63, 0.09], t(77) = −1.49, p = .140. In fact, the effects were in the opposite direction.

The hint of a negative relationship when political identity is highlighted before making political judgments could illustrate Pure Independents’ extreme concern with avoiding partisan influence. That is, they might have some subjective experience of favoring the policy aligned with their implicit party identity and deliberately contrast their judgment away from that tendency in an effort to deny its influence. But, when not directly alerted to the relevance of their own political identity (i.e., when party identity is measured after making the political judgment), they might assimilate to their implicit party identity—as Independents did on the whole across studies and task orders. An alternative possibility is that Independents prefer one of the welfare or education policies, and then align their implicit party identities to match the party who proposed that preferred policy. However, that would not account for the (possible) negative relationship when implicit party identity is measured first. This speculative interpretation is highly intriguing but requires substantial follow-up research to clarify the reliability of the result and separate and test the possible explanations. In any case, it supports the evidence that Pure Independents have or recruit cognitive resources that mitigate some influence of their own implicit partisanship.

Conclusion

Political independence implies objective information processing, a normatively desirable state. This may be the reason that, in recent years, more American citizens identify as Independent than as Democrats or Republicans (Pew Research Center, 2010). However, the current results suggest that the cultural values of objectivity have not permeated deep enough into citizens’ identities to remove partisan influence from all their judgments. In politics, as in most domains, “who we are” and “who we say we are” are not the same thing.

Footnotes

Appendix A

Appendix B

Authors’ Note

Nosek is an officer and Hawkins is a consultant of Project Implicit, Inc., a nonprofit organization that includes in its mission “To develop and deliver methods for investigating and applying phenomena of implicit social cognition, including especially phenomena of implicit bias based on age, race, gender, or other factors.” Data are available at .

Declaration of Conflicting Interests

The author(s) declared no potential conflicts of interests with respect to the research, authorship, and/or publication of this article.

Funding

The author(s) disclosed receipt of the following financial support for the research, authorship, and/or publication of this article: This project was supported by a gift from Project Implicit.

Notes

References

The American National Election Studies Guide to Public Opinion and Electoral Behavior. (2010). Party identification 7-point scale 1952-2004. Stanford University and the University of Michigan [producers and distributors]. Available from www.electionstudies.org

Arcuri

Castelli

Galdi

Zogmaister

Amadori

(2008). Predicting the vote: Implicit attitudes as predictors of the future behavior of decided and undecided voters. Political Psychology, 29, 369-387.

Bargh

J. A.

(1994). The four horsemen of automaticity: Awareness, intention, efficiency, and control in social cognition. In Wyer

R. S.

Srull

T. K.

(Eds.), Handbook of social cognition (2nd ed., pp. 1-40). Hillsdale, NJ: Erlbaum.

Brewer

M. B.

(1991). The social self: On being the same and different at the same time. Personality and Social Psychology Bulletin, 17, 475-482.

Bullock

J. G.

(2011). Elite influence on public opinion in an informed electorate. American Political Science Review, 105, 496-515.

Campbell

Converse

P. E.

Miller

W. E.

Stokes

D. E.

(1960). The American voter. Chicago, IL: University of Chicago Press.

Cohen

G. L.

(2003). Party over policy: The dominating impact of group influence on political beliefs. Journal of Personality and Social Psychology, 85, 808-822.

Dennis

(1988). Political independence in America, part I: On being an independent partisan supporter. British Journal of Political Science, 18, 77-109.

Ehrlinger

Gilovich

Ross

(2005). Peering into the bias blind spot: People’s assessments of bias in themselves and others. Personality and Social Psychology Bulletin, 31, 680-692.

10.

Fazio

R. H.

(1990). Multiple processes by which attitudes guide behavior: The MODE model as an integrative framework. In Zanna

M. P.

(Ed.), Advances in experimental social psychology (Vol. 23, pp. 75-109). New York, NY: Academic Press.

11.

Friese

Bluemke

Wanke

(2007). Predicting voting behavior with implicit attitude measures: The 2002 German parliamentary election. Experimental Psychology, 54, 247-255.

12.

Galdi

Acuri

Gawronski

(2008). Automatic mental associations predict future choices of undecided decision-makers. Science, 321, 1100-1102.

13.

Garst

Bodenhausen

G. V.

(1996). “Family values” and political persuasion: Impact of kin-related rhetoric on reactions to political campaigns. Journal of Applied Social Psychology, 26, 1119-1137.

14.

Gawronski

Bodenhausen

G. V.

(2006). Associative and propositional processes in evaluation: An integrative review of implicit and explicit attitude change. Psychological Bulletin, 132, 692-731.

15.

Goren

Federico

C. M.

Kittilson

M. C.

(2009). Source cues, partisan identities, and political value expression. American Journal of Political Science, 53, 805-820.

16.

Graham

Haidt

Nosek

B. A.

(2009). Liberals and conservatives rely on different sets of moral foundations. Journal of Personality and Social Psychology, 96, 1029-1046.

17.

Greene

(1999). Understanding party identification: A social identity approach. Political Psychology, 20, 393-403.

18.

Greenwald

A. G.

Banaji

M. R.

(1995). Implicit social cognition: Attitudes, self-esteem, and stereotypes. Psychological Review, 102, 4-27.

19.

Greenwald

A. G.

Banaji

M. R.

Rudman

L. A.

Farnham

S. D.

Nosek

B. A.

Mellott

D. S.

(2002). A unified theory of implicit attitudes, stereotypes, self-esteem, and self-concept. Psychological Review, 109, 3-25.

20.

Greenwald

A. G.

McGhee

D. E.

Schwartz

J. L. K.

(1998). Measuring individual differences in implicit cognition: The Implicit Association Test. Journal of Personality and Social Psychology, 74, 1464-1480.

21.

Greenwald

A. G.

Nosek

B. A.

Banaji

M. R.

(2003). Understanding and using the Implicit Association Test: I. An improved scoring algorithm. Journal of Personality and Social Psychology, 85, 197-216.

22.

Hogg

M. A.

Terry

D. J.

White

K. M.

(1995). A tale of two theories: A critical comparison of identity theory with social identity theory. Social Psychology Quarterly, 58, 255-269.

23.

Kaplan

J. T.

Freedman

Iacoboni

(2007). Us versus them: Political attitudes and party affiliation influence neural responses to faces of presidential candidates. Neuropsychologia, 45, 55-64.

24.

Keith

B. E.

Magleby

D. B.

Nelson

C. J.

Orr

Westlye

M. C.

Wolfinger

R. E.

(1992). The myth of the independent voter. Berkeley: University of California Press.

25.

Lewis-Beck

M. S.

Norpoth

Jacoby

W. G.

Weisberg

H. F.

(2008). The American voter revisited. Ann Arbor: University of Michigan Press.

26.

Lindner

N. M.

Nosek

(2009). Alienable speech: Ideological variations in the application of free-speech principles. Political Psychology, 30, 67-92.

27.

Nosek

B. A.

Graham

Hawkins

C. B.

(2010). Implicit political cognition. In Gawronski

Payne

B. K.

(Eds.), Handbook of implicit social cognition (pp. 548-564). New York, NY: Guilford.

28.

Nosek

B. A.

Greenwald

A. G.

Banaji

M. R.

(2007). The Implicit Association Test at age 7: A methodological and conceptual review. In Bargh

J. A.

(Ed.), Social psychology and the unconscious: The automaticity of higher mental processes (pp. 265-292). New York, NY: Psychology Press.

29.

Oppenheimer

D. M.

Meyvis

Davidenko

(2009). Instructional manipulation checks: Detecting satisficing to increase statistical power. Journal of Experimental Social Psychology, 45, 867-872.

30.

Payne

B. K.

Cheng

C. M.

Govorun

Stewart

B. D.

(2005). An inkblot for attitudes: Affect misattribution as implicit measurement. Journal of Personality and Social Psychology, 89, 277-293.

31.

Pew Research Center. (2010). Independents oppose party in power . . . again. Retrieved from http://pewresearch.org/pubs/1739/independent-voters-typology-2010-midterms-favor-republicans-conservative

32.

Pronin

Lin

D. Y.

Ross

(2002). The bias blind spot: Perception of bias in self versus others. Personality and Social Psychology Bulletin, 28, 369-381.

33.

Rahn

W. M.

(1993). The role of partisan stereotypes in information processing about political candidates. American Journal of Political Science, 37, 472-496.

34.

Ranganath

K. A.

Nosek

B. A.

(2008). Implicit attitude generalization occurs immediately, explicit attitude generalization takes time. Psychological Science, 19, 249-254.

35.

Rocatto

Zogmeister

(2010). Predicting the vote through implicit and explicit attitudes: A field research. Political Psychology, 31, 249-274.

36.

Smith

C. T.

Ratliff

K. A.

Nosek

B. A.

(2012). Rapid assimilation: Automatically integrating new information with existing beliefs. Social Cognition, 30, 199-220.

37.

Sriram

Greenwald

A. G.

(2009). The Brief Implicit Association Test. Experimental Psychology, 56, 283-294.

38.

Strack

Deutsch

(2004). Reflective and impulsive determinants of social behavior. Personality and Social Psychology Review, 8, 220-247.

39.

Triandis

H. C.

Bontempo

Villareal

M. J.

Asai

Lucca

(1988). Individualism and collectivism: Cross-cultural perspectives on self-ingroup relationships. Journal of Personality and Social Psychology, 54, 323-338.