How Political Identity Shapes Customer Satisfaction

Abstract

This article examines the effect of political identity on customers’ satisfaction with the products and services they consume. Recent work suggests that conservatives are less likely to complain than liberals. Building on that work, the present research examines how political identity shapes customer satisfaction, which has broad implications for customers and firms. Nine studies combine different methodologies, primary and secondary data, real and hypothetical behavior, different product categories, and diverse participant populations to show that conservatives (vs. liberals) are more satisfied with the products and services they consume. This happens because conservatives (vs. liberals) are more likely to believe in free will (i.e., that people have agency over their decisions) and, therefore, to trust their own decisions. The authors document the broad and tangible downstream consequences of this effect for customers’ repurchase and recommendation intentions and firms’ sales. The association of political identity and customer satisfaction is attenuated when belief in free will is externally weakened, choice is limited, or the consumption experience is overwhelmingly positive.

As political opinions grow more divided (Pew Research Center 2017), questions about how political identity shapes customer behaviors and marketplace outcomes are garnering more attention from researchers and marketers alike. While the marketing literature increasingly documents how political identity impacts preconsumption preferences for products and services (e.g., Han et al. 2019; Ordabayeva and Fernandes 2018; Winterich, Zhang, and Mittal 2012), less is known about how political identity shapes postconsumption behaviors and outcomes. To that end, recent work documents that conservatives (vs. liberals) are less likely to file customer complaints (Jung et al. 2017). Yet it is unclear whether this reflects conservatives’ (vs. liberals’) higher levels of satisfaction with products and services they consume. This is because, although dissatisfaction may precede complaint behavior, it does not result in customer complaints in most cases (e.g., Bearden and Oliver 1985; Day and Landon 1977). Thus, understanding how political identity shapes customer satisfaction—which has broad implications for repurchase intention, recommendation likelihood, and sales (e.g., Anderson and Mittal 2000; Anderson and Sullivan 1993)—is important for marketing theory and practice.

The present research examines the effect of political identity on customer satisfaction. Customer satisfaction refers to customers’ postconsumption judgment about a product or a service, taking the entire product or service consumption experience into account (e.g., Anderson and Fornell 1993; Oliver 2014; Spreng, Mackenzie, and Olshavsky 1996). We propose that a conservative political identity leads to higher customer satisfaction than a liberal political identity. We argue this happens because conservatives (vs. liberals) hold a stronger belief in free will (i.e., that individuals exercise agency and control over their actions). Stronger belief in free will increases conservatives’ (vs. liberals’) trust in their decisions, which increases their satisfaction with products and services they decide to buy and consume. We identify theoretically and practically relevant boundary conditions of this effect and downstream consequences for customers (repurchase and recommendation intentions) and firms (sales).

Our research advances theory by bridging the previously unrelated literature streams on political identity and customer satisfaction to identify a relevant effect of political identity on customer satisfaction and to outline its psychological mechanism, consequences, and boundaries. Our work extends the literature on political identity by identifying an essential postconsumption consequence of political identity—customer satisfaction—which goes beyond customer complaints and has broad implications such as repurchase, recommendation, and sales. Our findings similarly advance the literature on customer satisfaction by identifying political identity as its critical driver and by delineating factors that can systematically shift its effect.

This article offers useful insights for executives and managers aiming to improve customer satisfaction and thus improve repurchase, recommendations, and sales. First, firms across categories and industries should consider customers’ political identity to anticipate the satisfaction levels of distinct segments. Second, we identify practical factors that marketers can leverage to systematically understand and shift customer satisfaction. Third, we delineate the tangible implications of conservative (vs. liberal) segments’ higher satisfaction levels for subsequent customer behaviors and sales. Marketing executives can implement these practical insights, as customers’ political identity can be readily observed in behaviors such as voting and choice of news channels or inferred from customers’ location (Jung and Mittal 2020).

Next, this article describes the theoretical foundations for our predictions and nine studies examining them. They combine controlled experiments with analyses of real-world satisfaction data from the field, different product and service categories (actual prior purchases, real-time consumption experiences, airport services, health care insurers, business-to-business [B2B] suppliers, and restaurant reviews), and large-scale satisfaction measurement by different firms.

Theoretical Framework

Role of Political Identity in Postconsumption Customer Behavior

While prior research documents how political identity influences individuals’ preconsumption preferences and choices (e.g., Han et al. 2019; Ordabayeva and Fernandes 2018), less is known about the role of political identity in customers’ postconsumption judgments and behaviors, including their satisfaction with the products and services they consume. Jung et al. (2017) documented a negative association between political conservatism (vs. liberalism) and customer complaints. However, it is unclear whether this implies that conservatives are generally more satisfied than liberals. Although some may view complaints as an expression of dissatisfaction (Alicke et al. 1992; McGraw, Warren, and Kan 2015), not all dissatisfactory consumption experiences lead to complaints, and relatively few dissatisfied customers actually complain (e.g., Bearden and Oliver 1985; Day and Landon 1977). Instead of complaining, most dissatisfied customers tend to choose alternative brands (Morel, Poiesz, and Wilke 1997) because complaining is costly, requiring time, resources, and social capital (e.g., McGraw, Warren, and Kan 2015; Zhang, Feick, and Mittal 2014), and is often not expected to yield favorable outcomes for customers (e.g., Day and Landon 1977; Day 1984). In short, customer complaints may only partly reflect customers’ satisfaction or dissatisfaction. Instead, customer satisfaction is a broader phenomenon with wide-ranging implications such as repurchase and recommendations. However, research on conservatives’ and liberals’ complaining behavior neither considers factors that mitigate the effect of political identity nor examines the consequences of complaints for firm outcomes such as sales.

Addressing this gap, we posit that a conservative (vs. liberal) identity increases customer satisfaction, with tangible consequences for firms, and that this effect is mitigated by several moderators. Next, we delineate the process through which political identity impacts satisfaction.

Belief in Free Will and Trust in Decisions Drive Conservatives’ (vs. Liberals’) Higher Customer Satisfaction

We propose that individuals’ political identity shapes their customer satisfaction through belief in free will and their subsequent trust in their decisions. Our conceptualization builds on two distinct relationships between (1) political identity and perceptions of agency as well as (2) attributions of responsibility and satisfaction, discussed separately in prior research.

Belief in free will

Belief in free will is the extent to which individuals think that people have control over their actions (Carey and Paulhus 2013). This belief stems from individuals’ desire to view their life, destiny, and outcomes as under their control, rather than predetermined (Baumeister and Monroe 2014). Its validated measurement scale includes items such as “people have complete free will” and “people have complete control over the decisions they make” (Paulhus and Carey 2011).

We argue that belief in free will is systematically linked to political identity by building on research that relates political identity to perceptions of agency and locus of control (Skitka 1999). Studies report that conservatives are more likely than liberals to perceive that individuals have agency and control over their outcomes (Gupta and Wowak 2017; Tetlock 2000). Therefore, conservatives attribute outcomes such as inequality (Pratto and Cathey 2002) and unemployment (Skitka and Tetlock 1992) to internal, individual actions (e.g., hard work and discipline) rather than external circumstances. Consequently, we expect a conservative (vs. liberal) political identity to be associated with a stronger belief in free will. In doing so, we conceptualize political identity and free-will belief as theoretically distinct constructs, consistent with the differences between them discussed in previous work. Indeed, political identity is a salient and visible facet of individuals’ identity that is formed early in life (Fraley et al. 2012) and that is linked to a constellation of motives and beliefs, of which free-will belief is just one part (Jost, Federico, and Napier 2009). In contrast, free-will beliefs are not as salient or visible, and they require a deeper understanding of how the world works (Paulhus and Carey 2011). We validate the distinctiveness of political identity and free-will beliefs in Study 3 and a follow-up posttest.

While prior political identity research has not considered that free-will beliefs resulting from differences in political identity may systematically drive satisfaction, there may be some support for this proposition in studies on customer satisfaction. These studies report that satisfaction is influenced by customers’ attribution of the outcome of a consumption experience to self or to factors outside of self (e.g., Tsiros, Mittal, and Ross 2004), which echoes the notion of perceived agency and locus of control captured by free-will beliefs. However, free-will beliefs diverge from internal (vs. external) attributions of responsibility and locus of control studied in customer satisfaction research in an important way. Specifically, attributions and locus of control are largely determined by the nature of consumption outcomes and experiences: customers are less likely to perceive responsibility and control over consumption outcomes (with implications for satisfaction) that are negative, unstable (i.e., vary across similar consumption experiences due to factors such as product perishability), or uncontrollable by self (i.e., due to salient external conditions like the weather or other participating parties) (e.g., Tsiros, Mittal, and Ross 2004; Weiner 2000). In contrast, free-will beliefs are more stable and less dependent on specific circumstances or outcomes (Zheng, Van Osselaer, and Alba 2016). Indeed, the literature on political identity and its associated beliefs indicates that specific individual experiences, including stressful and unexpected ones, do not fundamentally change conservatives’ and liberals’ beliefs; instead, they often motivate people to justify and stick to their worldviews and ideological positions (e.g., Cutright et al. 2011; Jost and Hunyady 2002). Further, free will is directly linked to decision making as it reflects the capacity to make choices aligned with one's goals. Consistent with our theorization, a pilot study confirmed that free-will beliefs and attributions of responsibility or control, although positively correlated, are distinct constructs (Web Appendix A). Thus, compared with attributions and control from customer satisfaction research, free-will beliefs can more systematically and fundamentally capture the aspect of individuals’ conservative (vs. liberal) identities (a largely stable and inherent individual trait) that leads them to be consistently more satisfied with consumption experiences. More specifically, we posit that conservatives’ (vs. liberals’) stronger free-will beliefs drive their higher levels of satisfaction because they enhance individuals’ trust in their decisions.

Trust in decisions

Prior decision-making research shows that individuals strive to make decisions that are satisfactory and to perceive their decisions as sound (Brownstein 2003; Schrift, Netzer, and Kivetz 2011). This motivation to trust one's own decisions and to view the decisions as good and sound stems from individuals’ fundamental desire to view their actions and abilities in a positive light (e.g., Bandura 2010).

Perceiving higher (vs. lower) personal agency and control over one's decisions, which is the case for individuals with high (vs. low) belief in free will, may enhance individuals’ trust in their decisions. This is because perceiving agency over an action enhances individuals’ beliefs of their efficacy and expected ability to successfully complete the action (Bandura 1980), and perceiving personal control over a decision or action increases individuals’ confidence in the outcome of the decision or action (Botti and McGill 2010; Simon and Spiller 2016). These insights suggest that conservatives’ (vs. liberals’) stronger belief in free will may enhance their trust in the decisions they make (i.e., increase their trust that they can make sound decisions).

We argue that greater trust in one's own decisions may drive conservatives’ (vs. liberals’) higher customer satisfaction. This may occur because cumulative customer satisfaction reflects a holistic evaluation of individuals’ entire consumption experience with a product or a service (Anderson and Fornell 1993; Oliver 2014). Because this holistic evaluation requires the recollection and retrospective reconstruction of prior consumption experiences, it may be influenced by subjective factors (Pieters and Swick 1993). We argue that political identity is one such factor.

Research on motivated reasoning shows that individuals’ perceptions and evaluations of events and experiences are often shaped by their motivations and views (e.g., Fiske and Taylor 1991). Individuals selectively process and interpret information to support their initial beliefs and attitudes (e.g., Zuckerman et al. 1995). Motivated reasoning plays an important role in customers’ assessment of their satisfaction with products and services they consume. Customers are more likely to selectively overlook adverse information and report higher satisfaction with products and services about which they have positive initial beliefs (e.g., Ahluwalia, Burnkrant, and Unnava 2000; Woodruff, Cadotte, and Jenkins 1983). Drawing on this reasoning, we posit that individuals’ motivation to trust their decisions and to view their decisions as sound may lead them to confirm and justify their initial decision to consume a product or service by selectively processing information about the product or service experience. As such, conservatives’ (vs. liberals’) higher attributions of agency for decisions due to stronger free-will belief and their resulting higher trust in their decisions is likely to result in higher customer satisfaction. Thus,

H₁: Customers with a conservative political identity report a higher level of customer satisfaction than customers with a liberal political identity.

H₂: The association between political identity and customer satisfaction is mediated by conservatives’ (vs. liberals’) stronger belief in free will and higher trust in their decisions.

In addition to the proposed role of belief in free will and trust in decisions, we test potential alternative mediating mechanisms that have been linked to political identity in extant ideology research. In Studies 1 and 2b, we test whether conservatives’ (vs. liberals’) higher trust in their consumption decisions and satisfaction levels could result from their greater need for closure and certainty (Jost, Glaser, et al. 2003), stronger perception of the market as fair (Jost, Blount, et al. 2003), greater engagement in heuristic over deliberative processing (Miller, Krochik, and Jost 2009), stronger experience of processing fluency (Northey and Chan 2020), or higher cognitive dissonance or dissonance reduction tendencies (Vraga 2015). In Study 1, we also consider the potential role of customers’ expectations about consumption outcomes and commitment to their consumption decisions based on prior customer satisfaction research (Oliver 1986).

Moderators

Consistent with the process-by-moderation logic, we posit that the association between political identity and customer satisfaction may be mitigated when individuals’ belief in free will is attenuated or when their trust in their decisions is increased. To that end, we examine the moderating effects of attenuated free-will beliefs and choice availability, which we argue impact individuals’ free-will perceptions. We also examine the moderating role of valence of the consumption experience, which we argue impacts individuals’ trust in their decisions.

Attenuation of belief in free will

Although free will represents a largely stable worldview, it can also be experimentally manipulated (Alquist, Ainsworth, and Baumeister 2013; Vohs and Schooler 2008). Specifically, belief in free will can be temporarily changed by presenting individuals with scientific evidence about the lack of free will in real-life events and outcomes and by subsequently prompting individuals to recall instances when they lacked free will in their own life (Alquist, Ainsworth, and Baumeister 2013; Vohs and Schooler 2008). This manipulation reduced individuals’ susceptibility to correspondence bias (i.e., likelihood to credit other people's behaviors to personal, rather than situational, factors)—similar to the effect found at inherent low levels of free-will beliefs (Genschow, Rigoni, and Brass 2017). Furthermore, the effects of political identity on customer behavior can be mitigated when conservatives’ and liberals’ relevant motives and beliefs are experimentally muted (Ordabayeva and Fernandes 2018). Extending this logic, experimentally weakening belief in free will may lower conservatives’ satisfaction and attenuate the link between political identity and satisfaction.

• H₃: The association between political identity and customer satisfaction is attenuated when belief in free will is externally weakened (vs. not).

Choice availability

The extent to which individuals can exercise free will may be influenced by choice availability at the time of purchase. Individuals perceive autonomy over their choices when choice is actually available to them (Botti and McGill 2010). They also perceive more (less) autonomy and agency over their consumption decisions when there are more (fewer) product options from which they can choose (Reibstein, Youngblood, and Fromkin 1975). Reduced choice availability may directly lower individuals’ perceptions of and ability to exercise free will over their consumption decisions. This may attenuate the link between political identity and customer satisfaction by lowering the satisfaction level of conservatives who have a higher baseline free-will belief than liberals.

• H₄: The association between political identity and customer satisfaction is attenuated when choice availability is low (vs. high).

Valence of consumption experience

A positive (vs. negative) consumption experience may moderate the phenomenon by increasing individuals’ trust in their decisions. A positively valenced experience with a task (i.e., a positive outcome or process) increases individuals’ sense of efficacy and confidence in their ability to successfully complete the task by providing favorable feedback about their performance (Lyubomirsky, King, and Diener 2005; Podsakoff and Farh 1989). Similar insights emerge from customer satisfaction studies, which report that positive (vs. negative) consumption experiences may boost individuals’ perceptions of their stable personal qualities such as good taste (Weiner 2000). Therefore, we expect a positive (vs. negative) consumption experience to boost individuals’ trust in their ability to make sound decisions, increasing customer satisfaction. We do not expect the valence of the experience to move individuals’ belief in free will—as discussed previously, free-will belief is less dependent on variations in specific outcomes or experiences (e.g., Zheng, Van Osselaer, and Alba 2016). We test the moderating effect of valence of the experience on individuals’ trust in decisions and free-will beliefs in Study 4. Importantly, because liberals (vs. conservatives) have a lower baseline level of trust in their decisions, we expect a positive (vs. negative) experience to increase decision trust and subsequent customer satisfaction more for liberals than conservatives. We do not expect a negative consumption experience to lower conservatives’ decision trust in a similar way because conservatives are more strongly motivated to justify their outcomes and experiences, including negative ones (Jost and Hunyady 2002; Jung et al. 2017). Thus, we expect,

• H₅: The association between political identity and customer satisfaction is attenuated when customers have a positive (vs. negative) consumption experience.

Downstream Consequences

Customer satisfaction is essential to firms because it has tangible behavioral and financial consequences, including customers’ recommendation behavior, repurchase behavior, and firms’ sales (Mittal et al. 2021; Morgan and Rego 2006; Oliver 2014). First, satisfied customers are more likely than dissatisfied customers to engage in positive word of mouth and recommend a product or a service, which directly lowers a firm's customer acquisition cost (Anderson 1998). Second, increased customer satisfaction increases customers’ likelihood to repurchase the brand (Anderson 1994), which increases customer retention and lowers the cost of serving current customers (Anderson and Mittal 2000). Indeed, satisfied customers are insulated from competitor offerings because satisfied customers find it riskier to switch brands (Anderson and Sullivan 1993). Consequently, prior research has shown a strong association between customer satisfaction and sales (Mittal et al. 2021; Morgan and Rego 2006). Building on these findings, we predict that conservatives’ (vs. liberals’) higher customer satisfaction will lead to their higher recommendation and repurchase intentions as well as higher sales for firms.

• H₆: Conservatives’ (vs. liberals’) higher customer satisfaction is associated with higher (a) recommendation intention, (b) repurchase intention, and (c) sales.

Empirical Plan

We report nine studies that utilize multiple methodologies, primary and secondary data, real and hypothetical behavior, different samples, and diverse product and service categories. The first five studies utilize controlled experiments, and the last four studies use field data. We isolate the effect of political identity by consistently controlling for key demographic covariates (age, gender, income), which could be linked to identity-related beliefs.

Study 1 identifies the effect of measured political identity on customer satisfaction with an actual prior purchase and tests the underlying process as well as alternative explanations. Studies 2a and 2b control for the content of the customer experience to establish robustness. Study 2a examines the effect of manipulated political identity and its underlying process, and Study 2b further tests this process against alternative explanations. Studies 3 and 4 examine the moderating effects of attenuated free-will beliefs and valence of the consumption experience.

Studies 5a–5d utilize real-world settings and downstream consequences to extend generalizability and managerial relevance. The studies test the link between political identity and customer satisfaction using secondary data, including users’ online ratings of restaurants (Study 5a), travelers’ satisfaction with an airport (Study 5b), patients’ satisfaction with health care insurers (Study 5c), and customer satisfaction with B2B suppliers (Study 5d). These studies also test the moderating effect of choice availability (Study 5a) and examine managerially relevant downstream outcomes—repurchase intentions (Studies 5c and 5d), recommendation intentions (Study 5c), and sales (Study 5d). Four additional studies reported in the Web Appendix provide additional evidence of generalizability.

Study 1: Identifying Phenomenon and Underlying Process

Study 1 examines the link between political identity and customer satisfaction and tests the psychological process. We predicted that conservatives would report higher levels of customer satisfaction than liberals (H₁) because of their stronger belief in free will and higher trust in their decisions (H₂). We compared this process to alternative constructs that have been linked to political identity in extant research: need for closure, fluency, fair market ideology, dissonance reduction, cognitive dissonance, and heuristic processing. We also probed additional processes that could be at play: decision commitment and expectations about decision outcomes.

Method

A total of 412 U.S. adults completed the study on Amazon Mechanical Turk (MTurk) for a small payment (M_age = 38 years; 30.1% female).

Political Identity

In Study 1 (and in Studies 2b–4), we used a single-item scale (Jost 2006; “Please locate yourself on the following scale of political orientation”; 1 = “extremely liberal,” and 9 = “extremely conservative”; M = 6.75, SD = 2.16). Web Appendix B provides the distribution of political identity scores in Studies 1, 2b–4, and 5d which measured individual-level political identity and considers the potential implications of the distribution for our results.

Customer satisfaction

Following Tsiros and Mittal (2000), participants recalled a product that they had purchased in the past. To ensure that the purchase was sufficiently significant and memorable and to make subsequent product evaluations relevant, the product needed to be purchased within the past two years for at least $30. Participants reported how much they had paid for the product (M = $695.00, SD = $6,391.21) and how many months ago they had made the purchase (M = 12.02, SD = 7.25).

Following Oliver (2014), we measured postconsumption customer satisfaction in two ways. Participants indicated how satisfied they were with the product (1 = “very dissatisfied,” and 11 = “very satisfied”; M = 9.00, SD = 1.87) and rated their overall satisfaction taking the entire product experience into account (1 = “extremely dissatisfied,” and 7 = “extremely satisfied”; M = 5.85, SD = .95). We standardized and averaged the two items into an index (r = .63).

Psychological process

Participants indicated their trust in their decisions (three items; e.g., “I can trust my ability to make good decisions”; 1 = “not at all,” and 7 = “very much”; α = .74; M = 5.85, SD = .85) as well as their belief in free will (seven items; Paulhus and Carey 2011; e.g., “People have complete control over the decisions they make”; 1 = “strongly disagree,” and 7 = “strongly agree”; α = .88; M = 5.65, SD = .92). Web Appendix C provides the scales and summary statistics for these two constructs as well as alternative mechanisms.

Alternative explanations

We measured alternative mechanisms from the literature: need for closure, fluency, fair market ideology, dissonance reduction, cognitive dissonance, and deliberative (vs. heuristic) processing. We also measured decision commitment and expectations about decision outcomes. Web Appendix C lists the scales and summary statistics. The alternatives and trust in decisions were measured, in random order, before free-will beliefs.

Demographics

Participants indicated their gender, age, and income (1 = “under $20,000,” and 15 = “$150,000 or above”; M = 5.14, SD = 2.68).

Results

In all studies, the regression analyses consistently controlled for three demographics—age, gender, income—that could be potentially related to political identity. Table 1 reports the regression coefficient estimates for all studies. As we describe subsequently, the results replicated without the demographic control variables in all studies. Web Appendix D provides the participant screening criteria used in each study. Because Study 1 used measures (rather than manipulations) of our core constructs, no participant was screened from the analyses.

Table 1.

Regression Coefficient Estimates (Unstandardized) for Customer Satisfaction.

Predictor	Study 1	Study 2a	Study 2b	Study 3	Study 4	Study 5a Column A	Study 5a Column B	Study 5b	Study 5c	Study 5d
Political identitya	.10***	.14**	.07**	.02	.07***	.10***	.15***	.10***	.39**	.02*
Ageb	−.005	.02***	.01	−.01*	.005	.03***	.03***	.005	−.08***	—
Genderb	−.04	−.10	.13*	.12	−.10	3.53	3.71	−.01	−.09	—
Incomeb	.03**	.002	−.01	.005	−.01	.01	.01	−.07***	.03	—
Moderatorc	—	—	—	.04	.41***	—	−.002	—	—	—
Political identity × Moderator	—	—	—	.06**	−.07***	—	.06***	—	—	—

*p ≤ .10.

**p ≤ .05.

***p ≤ .01.

^a

Political identity is a conservative (vs. liberal) political identity. In Studies 1, 2b, and 4, it is a mean-centered single-item measure (1 = “extremely liberal,” and 9 = “extremely conservative”). In Study 2a, it is a manipulation (−1 = liberal, +1 = conservative). In Study 3, it is a mean-centered single-item measure (1 = “very left-wing,” and 9 = “very right-wing”; set by lab administration at the beginning of the lab session).

^b

Age, gender (−1 = male, +1 = female), and income are participant demographics in Studies 1–4 and inferred county demographics (median age, % male, mean income per capita) in Study 5a. The scales for demographics in Studies 5b and 5c are provided in Web Appendices H and K. Demographics were missing in the B2B manager survey used in Study 5d.

^c

Moderator is manipulated free-will belief (−1 = low, 1 = high) in Study 3, manipulated valence of experience (−1 = negative, 1 = positive) in Study 4, and number of alternatives within a five-mile radius of restaurant in Study 5a.

Notes: The dependent variable (customer satisfaction) is a standardized (z-score) mean of two items (1 = “very dissatisfied,” and 11 = “very satisfied”; 1 = “extremely dissatisfied,” and 7 = “extremely satisfied”) in Studies 1–4, an online rating (one to five stars) in Study 5a, item from a field survey in Studies 5b–5d (Web Appendices H, K, and N). Studies 5a–5d included additional controls that are relevant and standard in each real-world context. For complete results, see Web Appendices D, I, L, and O.

Customer satisfaction

We regressed the customer satisfaction index on mean-centered political identity (with mean-centered age, gender, and income as controls). We mean-centered political identity in all studies in which it was measured for consistency even though some studies do not include an interaction term. The coefficient of political identity was significant and positive (b = .10, t = 4.77, p < .001, standardized β = .23): conservative (vs. liberal) political identity positively predicted customer satisfaction. In this and the remaining studies, the results replicated without the demographic control variables (b = .09, t = 4.40, p < .001, β = .21) and when controlling for product price and purchase timing (b = .09, t = 4.62, p < .001, β = .21). In addition, in this study and Studies 3 and 4, which utilized the purchase recall task, two research assistants coded the hedonic (vs. utilitarian) and positive (vs. negative) nature of the recalled purchases using dichotomous scales. Political identity did not impact these purchase characteristics (hedonism: β = .01, Wald χ² = .08, p = .782, exp(β) = 1.01; positivity: β = −.02, Wald χ² = .04, p = .843, exp(β) = .98), and the results held after controlling for these characteristics (b = .09, t = 4.62, p < .001, β = .22). Therefore, the control variables are not discussed any further.

Psychological process

Three separate regressions showed that (1) political identity positively predicted free-will beliefs (b = .13, t = 6.39, p < .001, β = .30), (2) free-will beliefs positively predicted trust in decisions (b = .71, t = 25.00, p < .001, β = .77), and (3) trust positively predicted customer satisfaction (b = .62, t = 14.43, p < .001, β = .59). A serial mediation analysis (Model 6 in PROCESS 3 with 5,000 bootstraps and 95% bias-corrected confidence intervals [CI]; Figure 1) with political identity as the independent variable, customer satisfaction as the dependent variable, and free-will beliefs and trust as the two mediators (in this order) revealed a significant serial indirect effect of political identity through the two mediators (ab = .0457, SE = .0130, 95% CI = [.0237, .0751]), but not through each mediator alone (free-will beliefs: 95% CI = [−.0098, .0395]; trust: 95% CI = [−.0219, .0069]). A serial mediation with the reverse order of the two mediators was not significant (95% CI = [−.0045, .0200]), nor was a serial mediation with political identity as the mediator and free-will belief as the independent variable (95% CI = [−.0171, .0046]).

Figure 1.

Studies 1, 2a, and 2b: psychological process.

Alternative explanations

We ran a competing mediation analysis to test if conservatives’ (vs. liberals’) higher need for closure, processing fluency, fair market ideology, dissonance reduction, cognitive dissonance, or heuristic (vs. deliberative) processing—instead of free-will beliefs—could drive their higher decision trust and subsequent higher satisfaction (Model 80 in PROCESS 3, where free-will beliefs compete with the six alternatives). We also ran a competing mediation analysis to test if two additional competing constructs (commitment to decisions and expectations of decision outcomes)—instead of decision trust—could lead conservatives’ (vs. liberals’) stronger free-will beliefs to boost their satisfaction (Model 81 in PROCESS 3, where trust competes with the two constructs). Table 2 summarizes the findings of these analyses.

Table 2.

Studies 1 and 2b: Competing Alternative Mediators (95% Bias-Corrected CIs of Indirect Effects).

Mediators	Study 1 Free-Will Competitors	Study 1 Trust Competitors	Study 2b Free-Will Competitors
Hypothesized
MED 1: Free-will belief	[.0203, .0699]	—	[.0061, .0400]
MED 2: Decision trust	—	[.0025, .0503]	—
MED 1 (Free-Will Belief) Competitors
Need for closure	[−.0022, .0235]	—	[−.0017, .0067]
Cognitive dissonance	[−.0073, .0005]	—	[−.0009, .0073]
Dissonance reduction	[−.0046, .0101]	—	[−.0007, .0082]
Deliberative processing	[−.0059, −.0002]	—	[−.0053, .0007]
Fluency	[−.0018, .0023]	—	[.0009, .0148]
Fair market ideology	[−.0001, .0165]	—	[−.0056, .0140]
Additional MED 2 (Trust) Competitors
Commitment	—	[−.0016, .0421]	—
Expectations	—	[−.0009, .0565]	—

Notes: MED = mediator. For measurement scales, see Web Appendix C.

None of the alternative constructs yielded a significant mediation except a serial mediation path through deliberative processing and decision trust (ab = −.0024, SE = .0015, 95% CI = [−.0059, −.0002]), which was still markedly weaker than the proposed serial mediation path through free-will beliefs and decision trust (ab = .0419, SE = .0127, 95% CI = [.0203, .0699]). We conclude that the proposed process (involving free-will beliefs and decision trust) explained the effect better than alternative constructs and specifications.

Discussion

Study 1 confirms H₁’s prediction that conservatives have higher customer satisfaction than liberals. This is explained by conservatives’ (vs. liberals’) stronger belief in free will which increases their decision trust (H₂). Study 1 also shows that the proposed process through free-will belief and trust in decisions explains the phenomenon better than several alternative explanations.

Next, Studies 2a and 2b examine the robustness of the phenomenon by testing the effect of political identity and its underlying mechanism while controlling for the actual customer experience. In addition, Study 2a tests the causal effect of political identity by manipulating it, and Study 2b further examines alternative explanations.

Studies 2a and 2b: Effect of Political Identity and Underlying Process Controlling for Actual Customer Experience

Studies 2a and 2b examine the effect of political identity while experimentally controlling for the actual customer experience to rule out the role of conservatives’ (vs. liberals’) distinct purchases in driving the effect. Studies 2a and 2b also test the psychological process. Study 2a examines the process underlying the effect of manipulated political identity, whereas Study 2b further tests alternative explanations from extant research.

Study 2a: Method

Study 2a tests the causal effect of political identity on customer satisfaction and the underlying process in the context of a controlled customer experience. A total of 291 U.S. adults (M_age = 37 years; 47.8% female) were recruited on MTurk for a small payment.

Manipulated political identity

After indicating their demographics (same measures as in Study 1), participants engaged in a task that manipulated political identity which had been validated in prior research (Han et al. 2019; Ordabayeva and Fernandes 2018). In the conservative (vs. liberal) condition, participants recalled a prior interaction with someone who was more liberal (vs. conservative) than them. Comparing oneself to a more liberal (vs. conservative) individual should lead participants to perceive themselves as more conservative (vs. liberal). Afterwards, to ensure that participants paid attention, they were asked to indicate what they had been instructed to recall (multiple choice).

Psychological process

Participants filled abbreviated versions of the free-will belief and trust in decision scales from Study 1 in randomized order (trust in decisions: “I can trust my ability to make good decisions” and “I trust my capacity to do what is right”; r = .68, M = 5.87, SD = .99; free-will beliefs: “People have complete control over the decisions they make” and “People have complete free will”; r = .64, M = 5.35, SD = 1.83; seven-point scales). We abbreviated the scales to minimize participant fatigue after the political identity manipulation (e.g., Drolet and Morrison 2001).

Customer satisfaction

To more rigorously control for the nature and content of the customer experience, participants were invited to watch a how-to instructional video, instead of recalling a past product experience. Instructional videos were very relevant and popular during the COVID-19 pandemic (Wiers 2020), during which the study was conducted. To ensure choice availability, participants chose from five instructional videos of similar length and content (instructions to fold clothing, seal food packaging, work out from home, clean home, and preserve and peel common food ingredients).

After watching the video, participants completed the two-item scale of satisfaction with the video (as in Study 1; r = .79), which we again standardized and averaged. Reassuringly, the political identity manipulation did not impact the choice of the video (p = .715), nor did controlling for video choice impact the findings. Seventy-seven participants failed the recall task attention check and were screened out, leaving 214 usable observations (the screened out participants were not analyzed and the number of participants who failed the attention check did not differ between the political identity conditions; p = .791).

Study 2a: Results

Pretests

We conducted two pretests to check the effectiveness of the manipulation. The first pretest (N = 338 applying the same attention check as in the main study; M_age = 39 years; 41.4% female) confirmed that the manipulation moved individuals’ perceptions of their political identity on the single-item scale from Study 1 (M_con = 6.36, SD = 2.35; M_lib = 5.77, SD = 2.60; t = 2.95, p = .003, d = .32) as well as Tetlock’s (2000) dichotomous scale (% conservative: M_con = 61.88%, SD = 48.72%; M_lib = 46.63%, SD = 50.03%; χ² = 7.88, p = .005). The second pretest (N = 325 with the same attention check; M_age = 38 years; 44% female) showed that the manipulation moved participants’ free-will beliefs (M_con = 5.44, SD = .97; M_lib = 5.22, SD = 1.00; t = 2.01, p = .045, d = .23) and trust in decisions (M_con = 5.84, SD = .96; M_lib = 5.62, SD = 1.03; t = 2.05, p = .041, d = .23) (using full scales from Study 1).

Customer satisfaction

In the main study, an analysis of variance on satisfaction with political identity as a fixed factor revealed higher customer satisfaction in the conservative (vs. liberal) condition (M_con = .16, SD = .89; M_lib = −.15, SD = .98; F(1, 212) = 5.76, p = .017, η² = .03), in support of H₁. Controlling for demographics yielded consistent findings (p = .025; Table 1), as did controlling for participants’ choice of the video (p = .021).

Psychological process

The conservative (vs. liberal) political identity manipulation increased free-will beliefs (b = .16, t = 1.95; p = .052, standardized β = .13; M_con = 5.50, SD = 1.00; M_lib = 5.20, SD = 1.32; F(1, 212) = 3.82, p = .052, η² = .02), free-will beliefs positively predicted trust in decisions (b = .32, t = 6.07, p < .001, β = .38), and trust in decisions positively predicted customer satisfaction (b = .23, t = 3.69, p < .001, β = .25). A serial mediation analysis (Model 6) revealed a significant serial indirect effect of manipulated political identity through free-will beliefs and trust (ab = .0069, SE = .0050, 95% CI = [.0004, .0227]) (see Figure 1).

Study 2a supports the causal effect of manipulated political identity on customer satisfaction and the underlying mechanism. An additional study reported in Web Appendix E replicated the effect of manipulated political identity on customer satisfaction with an actual recalled prior purchase. Study 2b utilizes the same controlled customer experience as Study 2a to further compare the proposed process against salient alternative explanations.

Study 2b: Method

A total of 202 U.S. adults (M_age = 36 years; 42.1% female) completed the study on Prolific Academic for a small payment.

Political identity

After providing demographic information, participants completed the single-item scale of political identity from Study 1 (Jost 2006; M = 5.35, SD = 2.02).

Customer satisfaction

As in Study 2a, participants consumed an instructional video and indicated their satisfaction with it on a two-item scale, which was standardized and averaged for the analysis. As in Study 2a, political identity did not impact participants’ choice of a video (p = .740), and controlling for video choice in the analysis did not change the results.

Psychological process

Each participant indicated—in random order—their trust in their decisions (α = .90; M = 5.87, SD = .93) and then their free-will beliefs (α = .89; M = 5.26, SD = 1.14). We used the full scales from Study 1 (see Web Appendix C).

Alternative explanations

We examined alternative explanations from the literature: need for closure, processing fluency, fair market ideology, dissonance reduction, cognitive dissonance, and deliberative processing. Participants completed scales for these constructs from Study 1 after completing the scales of our core mediators, trust and free-will beliefs (all scales and summary statistics are provided in Web Appendix C). The order of the alternative mediators was randomized.

Study 2b: Results

Customer satisfaction

A regression of the customer satisfaction index on mean-centered political identity (with mean-centered age, gender, and income as controls) revealed a significant positive coefficient of political identity (b = .07, t = 2.01, p = .045, standardized β = .14) (which remained significant without the controls: b = .07, t = 2.07, p = .040, β = .14): conservative (vs. liberal) political identity positively predicted customer satisfaction.

Psychological process

As in Studies 1 and 2a, political identity positively predicted free-will beliefs (b = .23, t = 6.12, p < .001, β = .40), free-will beliefs positively predicted trust in decisions (b = .40, t = 7.91, p < .001, β = .49), and trust positively predicted customer satisfaction (b = .33, t = 4.78, p < .001, β = .32). A serial mediation analysis (Model 6) revealed a significant serial mediation effect of political identity through free-will beliefs and trust in decisions (ab = .0304, SE = .0113, 95% CI = [.0125, .0565]) in this order, but not in reverse order (95% CI = [−.0060, .0053]). The mediation through free-will beliefs alone (95% CI = [−.0366, .0335]) or decision trust alone (95% CI = [−.0399, .0077]) was not significant, and a serial mediation with free-will beliefs as the independent variable and political identity as the mediator was not significant either (95% CI = [−.0278, .0057]).

Alternative explanations

To test alternative mechanisms, we conducted the same competing serial mediation analysis as in Study 1 (Model 80 in PROCESS 3), in which free-will beliefs competed with the alternatives as the serial mediator. The results revealed a significant serial indirect effect through free-will beliefs and trust in decisions (ab = .0201, SE = .0086, 95% CI = [.0061, .0400]), as we theorized. None of the alternative serial mediators was significant except processing fluency, which yielded a much weaker indirect effect (ab = .0067, SE = .0036, 95% CI = [.0009, .0148]) than free-will beliefs. Once again, the proposed process through free-will beliefs and trust in decisions explained the phenomenon better than the alternatives (see Table 2 and Figure 1).

Discussion

Studies 2a and 2b tested the robustness of the phenomenon in complementary ways. Both studies isolated the effect of political identity on customer satisfaction while controlling for the actual customer experience. Further, Study 2a corroborated the causal nature of the effect of political identity and the underlying mechanism, whereas Study 2b tested the proposed mechanism against alternatives. Taken together, Studies 1, 2a, and 2b identified a robust link between political identity and customer satisfaction (H₁) across different customer experiences (recalled as well as real-time controlled) and showed that customers’ free-will beliefs and trust in decisions explain this link across settings and experiences (H₂).

Studies 1 and 2b uncovered two potential (but weak) alternatives (processing fluency and deliberative processing) to free-will beliefs in mediating the effect of measured political identity. Therefore, we ran a follow-up posttest to check if manipulating political identity impacts these alternatives in the same way it impacts free-will beliefs and trust (N = 322 applying the same attention check as in the main study and pretests; M_age = 38 years; 45% female). Reassuringly, political identity did not affect fluency (p = .412) or deliberative processing (p = .253).

Studies 3 and 4 examine the boundary conditions of the effect. Study 3 tests the moderating role of externally attenuating free-will beliefs, and distinguishes free-will beliefs from political identity. Study 4 tests the moderating role of valence of the customer experience.

Study 3: Moderating Role of Attenuated Belief in Free Will

Study 3 tests the process by attenuating free-will belief through a validated free-will belief manipulation (Alquist, Ainsworth, and Baumeister 2013; Vohs and Schooler 2008). The effect of political identity on customer satisfaction should be attenuated when belief in free will is weakened (H₃). We also measured political identity before and after the free-will manipulation to test whether manipulating free-will belief would shift political identity. A lack of such a shift in measured political identity would suggest that the two constructs are distinct.

Method

A total of 442 U.S. undergraduate students (M_age = 23 years; 52% female) participated for course credit.

Premeasure of political identity

The study was part of an hour-long online lab session containing multiple surveys. At the beginning of the session, participants provided demographics and indicated their political identity (1 = “very left-wing” to 9 = “very right-wing,” set by the lab administration in the introductory survey at the beginning of the session; M = 4.53, SD = 1.71).

Manipulated belief in free will

In the main study, participants completed the free-will manipulation (Alquist, Ainsworth, and Baumeister 2013; Vohs and Schooler 2008). In the low (vs. high) free-will condition, participants read an article about research showing that behavior is determined by situational factors (vs. personal choice). Next, they saw 10 statements on the importance of free will (e.g., “I demonstrate my free will every day when I make decisions”). They selected one statement and described an instance in their life when that statement rang the most false (vs. most true) for them. Participants then completed two manipulation check items about free will (“People have complete free will” and “I have free will”; 1 = “strongly disagree,” and 7 = “strongly agree”; r = .61; M = 5.46, SD = 1.23).

Postmeasure of political identity

Participants indicated their political identity on the same scale as at the beginning of the session (M = 4.67, SD = 1.79). The measures of political identity taken before and after the free-will manipulation were strongly correlated (r = .87).

Customer satisfaction

Following Study 1, participants completed the product-recall task and the two-item customer satisfaction scale (r = .72). During the research session, 62 participants did not follow the instructions or experienced technical issues, leaving 380 usable observations. The screened participants were not analyzed.

Results

Manipulation check and political identity

An analysis of variance on the free-will manipulation-check scale revealed a significant effect of the free-will manipulation (F(1, 379) = 35.69, p < .001, η² = .09). Participants in the high-free-will condition perceived themselves as having more free will than participants in the low-free-will condition (M_{high free will} = 5.83, SD_high = 1.02; M_{low free will} = 5.11, SD_low = 1.30). Reassuringly, the manipulation did not impact the postmeasure of political identity (F(1, 379) = .04, p = .834, η² <.001), nor did it change measured political identity after versus before the manipulation (F(1, 379) = .14, p = .707, η² <.001). This indicates that the free-will manipulation moved participants’ free-will beliefs, but not their political identity. This pattern of results provides evidence for discriminant validity between free will and political identity.

Customer satisfaction

We regressed the satisfaction score on mean-centered political identity, the free-will manipulation (−1 for low and +1 for high), and their interaction (with demographic controls). The main effects of political identity (premeasure) (b = .02, t = .61, p = .540, standardized β = .03) and free will (b = .04, t = .82, p = .412, β = .04) were nonsignificant. However, as predicted, and shown in Figure 2, the political identity (premeasure) × free will interaction was significant (b = .06, t = 2.31, p = .022, β = .12), in support of H₃. A conservative (vs. liberal) political identity positively predicted customer satisfaction when belief in free will was high (b = .08, t = 1.96, p = .052, β = .15), but not when belief in free will was low (b = −.05, t = −1.41, p = .162, β = −.10). A Johnson–Neyman analysis showed that the moderation occurred because weakened free-will beliefs lowered satisfaction among conservatives (who scored 5.65 or higher on the 1-to-9 political identity scale; b = .10, t = 1.97, p = .050, 95% CI = [.00, .20]), but it did not impact satisfaction among liberals (who scored below 5.65).

Figure 2.

Study 3: moderating effect of manipulated belief in free will.

The results replicated with the postmeasure of political identity (interaction: b = .07, t = 2.49, p = .013, β = .13; effect of political identity in the high-free-will condition: b = .08, t = 1.94, p = .054, β = .15 and in the low-free-will condition: b = −.05, t = −1.44, p = .153, β = −.11).

Discussion

Study 3 supports H₃ showing that the association between political identity and customer satisfaction was attenuated when belief in free will was experimentally weakened. Importantly, the free-will manipulation did not impact political identity, and the results were identical with the pre- and postmeasures of political identity. This confirmed the distinctiveness of free-will beliefs from political identity.

Discriminant Validity

To further establish discriminant validity between political identity and free-will beliefs, we conducted a posttest (N = 309 MTurkers; M_age = 41 years; 56% female). We measured free-will beliefs using seven items from Study 1 (Paulhus and Carey 2011; α = .86; M = 5.30, SD = 1.07). We measured political identity using a three-item scale (expanded) from the main studies: “Please locate yourself on the following scale of political orientation”; 1 = “extremely liberal/left-wing/Democratic,” and 9 = “extremely conservative/right-wing/Republican”; α = .88; M = 4.69, SD = 2.13). A confirmatory factor analysis showed the average variance extracted for each construct (.73 for political identity and .48 for belief in free will) was greater than the squared correlation of the two constructs (.24), establishing discriminant validity (Fornell and Larcker 1981). Next, Study 4 probes the role of trust in one's decisions by testing the moderating effect of valence of the consumption experience.

Study 4: Moderating Role of Positive (vs. Negative) Experience

Study 4 tests the moderating effect of a positive (vs. negative) consumption experience. Because a positive (vs. negative) experience should boost individuals’ trust in their decisions, we expected it to attenuate the effect of political identity (H₅). We tested this prediction by examining the impact of individuals’ actual positive (vs. negative) experience with a product.

Method

We recruited 412 participants (M_age = 37 years; 44% female) on MTurk.

Political identity

We used the single-item scale from Study 1 (M = 6.35, SD = 2.39).

Valence of experience

Participants engaged in the purchase recall task from Study 1. We manipulated the valence of the experience. In the positive (vs. negative) experience condition, participants described an actual past purchase “that turned out well (vs. poorly) and ended up being an excellent (vs. terrible) decision.”

Customer satisfaction

Participants completed the two-item scale from Study 1 (r = .84).

Results

Pretest

A separate pretest (N = 201; M_age = 36 years; 45.3% female) checked the effectiveness of the valence manipulation. Following the manipulation, participants completed the scales of free-will beliefs and trust in decisions from Study 1 in random order. Next, they rated the valence of the recalled consumption experience (1 = “definitely bad/negative,” and 7 = “definitely good/positive”).

First, as intended, the results confirmed that the positive (vs. negative) condition induced a recall of a consumption experience that was significantly positive (vs. negative) in valence (M_pos = 6.33, SD = .80; t-test of difference from the midpoint = 26.21, p < .001; M_neg = 3.43, SD = 1.99; t-test of difference from the midpoint = −2.55, p = .013). Second, the manipulation significantly moved participants’ decision trust (M_pos = 5.89, SD = .89; M_neg = 5.54, SD = 1.11; t = 2.42, p = .016, d = .34), but not their free-will beliefs (M_pos = 5.50, SD = .93; M_neg = 5.45, SD = .91; t = .35, p = .726, d = .05).

Customer satisfaction

We regressed the satisfaction score on mean-centered political identity, the experience valence manipulation (−1 for negative and +1 for positive), and their interaction (with demographic control variables). The results revealed significant main effects of political identity (b = .07, t = 4.25, p < .001, standardized β = .18) and experience valence (b = .41, t = 10.09, p < .001, β = .43): customer satisfaction was higher among conservatives (vs. liberals) and after a positive (vs. negative) consumption experience. Importantly, as predicted, and shown in Figure 3, the interaction of political identity and valence was significant (b = −.07, t = −3.89, p < .001, β = −.17). A conservative (vs. liberal) political identity increased customer satisfaction when the experience was negative (b = .14, t = 4.37, p < .001, β = .29), but not when it was positive (b = .007, t = .45, p = .657, β = .03), in support of H₅. A positive (vs. negative) experience boosted customer satisfaction more for liberals (bottom of the political identity scale: b = .78, t = 7.82, p < .0001) than for conservatives (top of the scale: b = .23, t = 3.81, p = .0002).

Figure 3.

Study 4: moderating effect of valence of consumption experience.

Discussion

Study 4 confirms that an overwhelmingly positive experience can systematically increase satisfaction among liberals and weaken the effect of political identity on satisfaction (H₅). This finding further corroborates the role of decision trust in driving the focal phenomenon, because a positive (vs. negative) consumption experience heightens individuals’ trust in their decisions. Practically, this result suggests that devising strategies to create very positive experiences for customers can pay off as it can win over segments which are chronically less satisfied (liberals). In the absence of such strategies, accounting for the baseline (main) effect of political identity would be crucial for companies whose customers’ typical consumption experiences are ambiguous with few overwhelmingly positive experiences.

Having established the phenomenon, its underlying process, and boundary conditions from controlled experiments in Studies 1–4, we examine the generalizability of the phenomenon as well as its tangible downstream consequences (H₆) in field settings in Studies 5a–5d.

Studies 5a–5d: Customer Satisfaction in the Field, Moderator, and Downstream Consequences

Studies 5a–5d examined the link between political identity and customer satisfaction in real-world settings using secondary data on customers’ satisfaction with restaurants (using online ratings) in Study 5a, airport services in Study 5b, health care insurers in Study 5c, and B2B suppliers in Study 5d. We also test the moderating role of choice availability (H₄) and examine the downstream consequences including customers’ behavioral intentions and firms’ sales (H₆).

Study 5a: Online Ratings of Restaurants, Moderating Role of Choice Availability

Study 5a uses customers’ online ratings of restaurants as a proxy of their postconsumption satisfaction. It tests the moderating role of choice availability by including the number of options available to customers within the vicinity of the restaurants that they rated. The effect of political identity on customer satisfaction is expected to be stronger when there are more (vs. fewer) alternatives to choose from (H₄).

Customer satisfaction

We used the Yelp Open Dataset (www.yelp.com/dataset), which contained 6.7 million ratings of .2 million businesses from 1.6 million consumers between 2004 and 2018. We analyzed customers’ ratings of restaurants (one to five stars) to capture customer satisfaction during the 2008, 2012, and 2016 presidential election cycles (we did not analyze 2004 because it only included 13 observations). This resulted in 746,825 ratings of 28,762 restaurants from 300,026 customers in 27 counties that had at least five observations.

Political identity

Consistent with Han et al. (2019), we used county-level U.S. presidential election data from the MIT Election Data + Science Lab (https://electionlab.mit.edu/) to infer a customer's political identity. As a proxy of customers’ political identity, we examined the proportion of votes cast for Republican versus Democratic presidential candidates in each election cycle in each county. We used the difference of these proportions as a measure of political identity, where a more positive (negative) score reflected a more conservative (liberal) political identity. A customer's county of residence was inferred from the county of the restaurant the customer rated. For a customer who rated multiple restaurants (N = 115,265) operating in different counties, the customer's county of residence was inferred as the county in which the customer provided more than half of their ratings.

Choice availability

We used the geographic coordinates of each restaurant, and counted the number of other restaurants within a five-mile radius from the restaurant location (which customers could presumably choose from). The results held with a ten-mile radius.

Control variables

First, we included per capita county-level income, percentage of female residents, and median age from the U.S. Census Bureau to mirror the demographic controls in our experiments. Second, we included controls typically used with online rating data (e.g., Chen and Lurie 2013). These included the number of views and the number of “useful,” “funny,” and “cool” votes each review received, the number of ratings the customer provided, the number of months the customer had been on Yelp, the number of fans the customer has, the customer's Yelp Elite status, the number of ratings the restaurant received, and whether the restaurant was open.

Study 5a: Results

To account for unobserved heterogeneity across customers, restaurants, and counties, we used a random-effects model:

$$\begin{array}{rl}\mathrm{Ratin}{\mathrm{g}}_{\mathrm{ijkt}}=& {\beta }_0+{\beta }_1\mathrm{Political}\mathrm{Identit}{\mathrm{y}}_{\mathrm{kt}}\\ & +{\beta }_2\ln (\mathrm{Number}\mathrm{of}\mathrm{alternative}{\mathrm{s}}_{\mathrm{jkt}})\\ & +{\beta }_3\mathrm{Political}\mathrm{Identit}{\mathrm{y}}_{\mathrm{kt}}\times \ln (\mathrm{Number}\mathrm{of}\mathrm{alternative}{\mathrm{s}}_{\mathrm{jkt}})\\ & +{\gamma }^{\mathrm{\prime }}\mathbf{R}\mathbf{a}\mathbf{t}\mathbf{i}\mathbf{n}\mathbf{g}\mathbf{C}\mathbf{o}\mathbf{n}\mathbf{t}\mathbf{r}\mathbf{o}\mathbf{l}{\mathbf{s}}_{\mathrm{ijkt}}+{\delta }^{\mathrm{\prime }}\mathbf{U}\mathbf{s}\mathbf{e}\mathbf{r}\mathbf{C}\mathbf{o}\mathbf{n}\mathbf{t}\mathbf{r}\mathbf{o}\mathbf{l}{\mathbf{s}}_{\mathrm{ijkt}}\\ & +{\theta }^{\mathrm{\prime }}\mathbf{R}\mathbf{e}\mathbf{s}\mathbf{t}\mathbf{a}\mathbf{u}\mathbf{r}\mathbf{a}\mathbf{n}\mathbf{t}\mathbf{C}\mathbf{o}\mathbf{n}\mathbf{t}\mathbf{r}\mathbf{o}\mathbf{l}{\mathbf{s}}_{\mathrm{j}}+{\psi }^{\mathrm{\prime }}\mathbf{C}\mathbf{o}\mathbf{u}\mathbf{n}\mathbf{t}\mathbf{y}\mathbf{C}\mathbf{o}\mathbf{n}\mathbf{t}\mathbf{r}\mathbf{o}\mathbf{l}{\mathbf{s}}_{\mathrm{k}}\\ & +\sum _{\mathrm{y}}{\tau }_{1\mathrm{y}}\mathrm{Yea}{\mathrm{r}}_{\mathrm{t}}+\sum _{\mathrm{m}}{\tau }_{2\mathrm{m}}\mathrm{Mont}{\mathrm{h}}_{\mathrm{t}}+{\nu }_{\mathrm{i}}+{\upsilon }_{\mathrm{j}}+{\omega }_{\mathrm{jk}}+{\epsilon }_{\mathrm{ijkt}},\end{array}$$

(1)

for customer i, restaurant j, county k, at time t, where ${\nu }_{\mathrm{i}}\sim \mathrm{N}({0,\sigma }_{\nu }^2)$ , ${\upsilon }_{\mathrm{j}}\sim \mathrm{N}({0,\sigma }_{\upsilon }^2)$ , ${\omega }_{\mathrm{jk}}\sim \mathrm{N}({0,\sigma }_{\omega }^2)$ , and ${\epsilon }_{\mathrm{ijkt}}\sim \mathrm{N}({0,\sigma }_{\epsilon }^2)$ . Table 1 shows the main results and Figure 4 maps the interaction. The full results of Studies 5a–5d are reported in Web Appendices F, K, N, and Q, respectively. A model with only the main effects revealed a significant positive effect of conservative (vs. liberal) political identity on customer satisfaction (b = .10, z = 3.72, p < .001; Table 1, Study 5a Column A; H₁). In support of H₄, a model with the interaction (that offered a superior fit: χ²(1) = 4.83, p  = .028) yielded a significant interaction effect (b = .06, z = 3.31, p < .001; Table 1, Study 5a Column B). As shown in Figure 4, the simple effect of conservative (vs. liberal) political identity became significant and positive, and this positive slope increased as choice availability increased (measured as the number of alternatives). The results were robust to alternative measures of political identity and choice availability (Web Appendix G). The main effect of political identity replicated using online ratings for a single brand of restaurant with multiple locations (Cheesecake Factory; Web Appendix H). The moderating effect of choice availability was replicated in a lab study that manipulated choice availability (Web Appendix I).

Study 5b: Travelers’ Satisfaction with Airport Services

Study 5b examined travelers’ satisfaction with the San Francisco International Airport (SFO). We analyzed the publicly available 2010–2018 Customer Survey conducted by SFO on an annual basis (https://www.flysfo.com/media/customer-survey-data). On average, 3,185 customers rated SFO each year.

Customer satisfaction

We used customers’ overall rating of SFO (1 = “unacceptable,” and 5 = “outstanding”). Web Appendix J provides several other satisfaction measures used.

Political identity

Following Study 5a, we used customers’ zip code information to infer the political identity of the county in which they reside. ¹ We used county-level presidential election data for 2012 and 2016 to calculate the difference in the proportions of total votes cast for Republican versus Democratic presidential candidates.

Control variables.

In addition to customers’ age, gender, and income (as in all experiments), we controlled for customers’ experiences with specific aspects of the airport (security and screening, finding one's way around the airport, departure time, encountering a specific problem) as well as the survey method (in-person interview vs. self-administered). We also included state- and year-fixed effects. After we removed observations with missing values, the final data set contained 9,970 surveys spanning seven years (2011, 2013–2018). Note that the 2010 and 2012 surveys omitted zip code data and could not be used.

Study 5b: Results

We ran the following regression model for customer i in county j at year t:

$$\mathrm{Satisfactio}{\mathrm{n}}_{\mathrm{ijt}}={\beta }_0+{\beta }_1\mathrm{Political}\mathrm{Identit}{\mathrm{y}}_{\mathrm{jt}}+{\gamma }^{\mathrm{\prime }}\mathbf{C}\mathbf{o}\mathbf{n}\mathbf{t}\mathbf{r}\mathbf{o}\mathbf{l}{\mathbf{s}}_{\mathrm{ijt}}+{\epsilon }_{\mathrm{ijt}}.$$

(2)

Table 1 shows that conservative (vs. liberal) travelers reported higher satisfaction with the airport (b = .10, t = 3.39, p  < .001, β = .04), which supports H₁ (see Web Appendix K for full results). When we analyzed whether conservative (vs. liberal) travelers were more satisfied with the airport's specific facilities and services (e.g., restaurants, retail shops and concessions, information booths), the findings were unchanged (see Web Appendix L).

Study 5c: Patients’ Satisfaction with Health Care Insurers, Impact on Recommendation and Repurchase Likelihood

Study 5c tests whether conservative (vs. liberal) patients are more satisfied with their insurers (H₁) and are subsequently more likely to repurchase (i.e., renew) and recommend their insurance plan (H₆). We utilized proprietary data where 6,000 patients in the United States rated their health care insurer, from a pool of 47 insurers, in 2015–2016.

Customer satisfaction

Patients rated the extent to which the insurer met their expectations (1 = “has fallen short of your expectations,” and 10 = “exceeded your expectations”). Web Appendix M provides the additional measures used in Study 5c.

Political identity

We used participants’ state of residence to infer their political identity from the 2016 presidential election data using the approach in Studies 5a and 5b.

Recommendation and repurchase likelihood

Patients rated their likelihood to recommend the insurer to family and friends (0 = “not at all likely,” and 10 = “extremely likely”) and to renew their plan with the insurer (1 = “definitely would not renew,” and 10 = “definitely would renew”).

Control variables

We controlled for participants’ age, gender, and income. In addition, we included their health condition, relationship with the health insurer (tenure, spending, nonemergency and emergency medical decisions), and insurer fixed effects in the model. After removing observations with missing variables, the final data set included 4,930 surveys.

Study 5c: Results

We specified the following model for customer i, in state j, rating health insurer k:

Figure 4.

Study 5a: effect of political identity on customer satisfaction at different levels of number of options.

$$\mathrm{Satisfactio}{\mathrm{n}}_{\mathrm{ijk}}={\beta }_0+{\beta }_1\mathrm{Political}\mathrm{Identit}{\mathrm{y}}_{\mathrm{j}}+{\theta }^{\mathrm{\prime }}\mathbf{C}\mathbf{o}\mathbf{n}\mathbf{t}\mathbf{r}\mathbf{o}\mathbf{l}{\mathbf{s}}_{\mathrm{ijk}}+{\epsilon }_{1\mathrm{ijt}},$$

(3)

$$\mathrm{Recommen}{\mathrm{d}}_{\mathrm{ijk}}={\gamma }_0+{\gamma }_1\mathrm{Political}\mathrm{Identit}{\mathrm{y}}_{\mathrm{j}}+{\gamma }_2\mathrm{Satisfactio}{\mathrm{n}}_{\mathrm{ijk}}+{\omega }^{\mathrm{\prime }}\mathbf{C}\mathbf{o}\mathbf{n}\mathbf{t}\mathbf{r}\mathbf{o}\mathbf{l}{\mathbf{s}}_{\mathrm{ijk}}+{\epsilon }_{2\mathrm{ijt}},$$

(4)

$$\mathrm{Rene}{\mathrm{w}}_{\mathrm{ijk}}={\delta }_0+{\delta }_1\mathrm{Political}\mathrm{Identit}{\mathrm{y}}_{\mathrm{j}}+{\delta }_2\mathrm{Satisfactio}{\mathrm{n}}_{\mathrm{ijk}}+{\psi }^{\mathrm{\prime }}\mathbf{C}\mathbf{o}\mathbf{n}\mathbf{t}\mathbf{r}\mathbf{o}\mathbf{l}{\mathbf{s}}_{\mathrm{ijk}}+{\epsilon }_{3\mathrm{ijt}}.$$

(5)

As the errors in Equations 3–5 may be correlated, we estimated the three equations using seemingly unrelated regressions (Zellner 1962). As shown in Table 1, conservative (vs. liberal) political identity increased customer satisfaction (b = .39, z = 2.27, p = .023, β = .03), in support of H₁. Satisfaction, in turn, increased customers’ likelihood to recommend the insurer (b = .99, z = 113.85, p < .001, β = .85) and to renew the plan (b = .84, z = 80.06, p < .001, β = .75; Web Appendix N). Thus, H₆ was supported. Political identity impacted customers’ likelihood to recommend (indirect effect = .39, p = .023) and repurchase from the provider through satisfaction (indirect effect = .33, p = .024; the standard error was computed using the delta method). The results remained unchanged with alternative satisfaction measures (Web Appendix O).

Study 5d: Satisfaction with B2B Suppliers, Impact on Repurchase and Actual Sales

Study 5d examines whether conservative (vs. liberal) managers at B2B firms report higher satisfaction with their supplier firm (H₁). It also investigates two tangible outcomes of satisfaction for firms: likelihood to repurchase from the supplier and suppliers’ sales (H₆). We analyzed data on manager satisfaction and demographics from a research collaborative that conducted national monthly surveys of B2B managers (January 2017–June 2019). We obtained and matched actual sales data for publicly listed B2B suppliers from Compustat. We obtained the survey data from Mittal et al. (2021).

Customer satisfaction

Satisfaction with B2B supplier was measured as 1 = “extremely dissatisfied,” and 7 = “extremely satisfied.” Web Appendix P provides additional measures.

Political identity

A five-point scale was used (1 = “strong Republican,” 2 = “moderate Republican,” 3 = “Independent,” 4 = “moderate Democrat,” and 5 = “strong Democrat”). The scale was reverse-coded so that higher scores reflected a more conservative identity.

Repurchase likelihood

Likelihood to use the supplier for their next project was measured as 1 = “extremely unlikely,” and 7 = “extremely likely.”

Sales

We used the natural logarithm of B2B suppliers’ quarterly sales from Compustat.

Repurchase likelihood

Control variables. For satisfaction and repurchase behavior, we controlled for the annual purchase of each manager's firm from the supplier, their involvement with the supplier, tenure, sector in which the supplier operates, and year and month fixed effects. For sales, we controlled for the supplier's financial leverage, liquidity, and sector, industry competitive intensity, industry demand instability, and year and month fixed effects. The final data set included 6,910 surveys for 646 suppliers operating in 42 two-digit Standard Industrial Classification industries.

Study 5d: Results

We ran the following models for customer i, supplier j operating in industry k, at time t:

$$\begin{array}{r}\mathrm{Satisfactio}{\mathrm{n}}_{\mathrm{ijkt}}={\beta }_0+{\beta }_1\mathrm{Political}\mathrm{Identit}{\mathrm{y}}_{\mathrm{ijkt}}+{\theta }^{\mathrm{\prime }}\mathbf{C}\mathbf{o}\mathbf{n}\mathbf{t}\mathbf{r}\mathbf{o}\mathbf{l}{\mathbf{s}}_{1\mathrm{ijkt}}\\ +{\nu }_{1\mathrm{k}}+{\epsilon }_{1\mathrm{ijkt}},\end{array}$$

(6)

$$\mathrm{Repurchas}{\mathrm{e}}_{\mathrm{ijkt}}={\gamma }_0+{\gamma }_1\mathrm{Satisfactio}{\mathrm{n}}_{\mathrm{ijkt}}+{\omega }^{\mathrm{\prime }}\mathbf{C}\mathbf{o}\mathbf{n}\mathbf{t}\mathbf{r}\mathbf{o}\mathbf{l}{\mathbf{s}}_{2\mathrm{ijkt}}+{\nu }_{2\mathrm{k}}+{\epsilon }_{2\mathrm{ijkt}},$$

(7)

$$\ln (\mathrm{Sale}{\mathrm{s}}_{\mathrm{ijkt}})={\delta }_0+{\delta }_1\mathrm{Satisfactio}{\mathrm{n}}_{\mathrm{ijkt}}+{\psi }^{\mathrm{\prime }}\mathbf{C}\mathbf{o}\mathbf{n}\mathbf{t}\mathbf{r}\mathbf{o}\mathbf{l}{\mathbf{s}}_{3\mathrm{ijkt}}+{\nu }_{3\mathrm{k}}+{\epsilon }_{3\mathrm{ijkt}},$$

(8)

where ${\nu }_{1\mathrm{k}}\sim \mathrm{N}({0,\sigma }_{\nu 1}^2)$ , ${\nu }_{2\mathrm{k}}\sim \mathrm{N}({0,\sigma }_{\nu 2}^2)$ , and ${\nu }_{3\mathrm{k}}\sim \mathrm{N}({0,\sigma }_{\nu 3}^2)$ are industry-random effects. We estimated the three equations together using seemingly unrelated regressions.

As shown in Table 1, the results supported H₁: the association of a conservative (vs. liberal) political identity and customer satisfaction was positive (b = .02, z = 1.72, p = .086). In support of H₆, customer satisfaction was positively associated with repurchase behavior (b = 1.05, z = 113.72, p < .001; Web Appendix Q) and sales (b = .17, z = 12.95, p < .001; Web Appendix Q). Thus, a one-unit increase in conservative (vs. liberal) political identity led to a .02-unit lift in repurchase likelihood (p = .068) and a .30% lift in sales (p = .088; standard errors of the indirect effects were computed using the delta method).

Discussion

Studies 5a–5d establish the generalizability and practical relevance of the link between political identity and customer satisfaction across several real-world settings (restaurant reviews, airport, health insurers, B2B services) and populations (real diners, travelers, patients, and managers of B2B firms). These studies also provide further evidence of the tangible consequences of the phenomenon for firms’ financial outcomes including sales.

General Discussion

This article uncovers a novel and consequential effect of political identity on customer satisfaction: conservatives are consistently more satisfied than liberals with the products and services they consume. Evidence from lab and field settings provides robust support for this effect with various participant populations (students, adults on MTurk, diners, travelers, patients, and managers) and hypothetical and real behaviors across various product and service categories. This effect emerges because of conservatives’ (vs. liberals’) stronger belief in free will, which, in turn, increases their trust in the decisions they make. We outline and test theoretically and practically relevant moderators, and we link our results to tangible downstream consequences such as customer repurchase and recommendation intentions and firm sales. The standardized regression coefficients obtained for the effect of political identity on the two-item combined standardized scale of satisfaction ranged between .14 and .29 in Studies 1–4. While the size of the effect varied to some extent across consumption settings (recalled, real-time experiences) and operationalizations of political identity (measured, manipulated), its emergence across contexts, categories, and populations is robust.

Theoretical Implications

This article brings together literature streams on political identity and customer satisfaction that were previously studied separately within a single nomological framework to delineate systematic differences in customer satisfaction among liberals and conservatives, with tangible implications for customers and firms.

Supplementing marketing studies on political identity that typically examine how political identity shapes individuals’ preconsumption choices and judgments (e.g., Han et al. 2019; Ordabayeva and Fernandes 2018; Winterich, Zhang, and Mittal 2012), we show that political identity systematically impacts individuals’ postconsumption satisfaction with the products and services they actually consume. Our work thereby expands the current limited understanding of how political identity shapes postconsumption behavior by going beyond customer complaints, identifying the unique mechanism (i.e., belief in free will and trust in decisions) behind conservatives’ and liberals’ satisfaction levels (and, potentially, their complaints), outlining relevant boundary conditions, and documenting the wide-ranging implications for customer behaviors and firm outcomes. Our research similarly advances the literature on customer satisfaction (e.g., Anderson, Fornell, and Lehmann 1994; Anderson and Sullivan 1993; Mittal et al. 2021; Oliver 2014) by identifying a novel, robust, and consequential driver of customer satisfaction (i.e., political identity), and by uncovering factors that can systematically shift the satisfaction levels of different segments. Whereas prior research has mainly focused on product performance in conceptualizing and predicting customer satisfaction, our research shows the importance of also including customers’ political identity.

Our findings offer interesting opportunities for future research. First, future studies could examine how political identity may impact additional consumption phenomena shaped by customers’ free-will beliefs, trust in decisions, and satisfaction. Examples include susceptibility to the endowment effect, the “mere possession” effect, and similar biases stemming from perceptions of products that individuals own and consume (Kahneman, Knetsch, and Thaler 1991). It may be the case that conservatives, who are more satisfied with the products they buy and consume than liberals, may be more susceptible to developing positive cognitions about and strong attachment to these products, resulting in stronger endowment and mere ownership effects. It may also be the case that conservatives are less likely to discard, resell, and recycle products they own because of their higher satisfaction. These can be tested in future research.

Second, we examine several potential alternative explanations including need for closure, processing fluency, heuristic processing, cognitive dissonance, dissonance reduction, fair market ideology, commitment to decisions, and expectations about decision outcomes. Yet the role of additional mechanisms in driving the phenomenon may need further examination. For example, the robustness of our results when controlling for product price, hedonic (vs. utilitarian) and positive (vs. negative) product characteristics (Studies 1, 3, and 4), as well as the chosen product option within a controlled option set (Studies 2a and 2b), indicates that political identity-based differences in satisfaction go beyond potential differences in the type, monetary value, or quality of products that individuals consume. Likewise, the robust phenomenon identified across various commercial (e.g., products one bought) and noncommercial (e.g., videos one watched online) stimuli suggests that conservatives’ (vs. liberals’) higher satisfaction goes beyond their general positive attitudes toward business. How the distinct motivations of conservatives and liberals may interact to shape customer satisfaction is an open question in need of more research.

Third, future research may examine additional boundary conditions of political-identity-based differences in customer satisfaction. For example, when customers do not choose the products and services themselves (e.g., when they receive gifts or delegate the choice to others), their perception of exercising free will over consumption decisions may be weakened, attenuating the political identity–satisfaction link. A similar attenuating effect could emerge when customers follow others’ recommendations such as from friends and family or algorithms. Further, the dimension (economic, social) along which individuals self-identify as being conservative or liberal may matter (Everett 2013). Given the sociocultural nature of conservatives’ (vs. liberals’) stronger concern about freedom, our effect could be more pronounced for social (vs. economic) conservatives. Future research could test these possibilities.

Managerial Implications

Because customers’ political identity can be readily observed and inferred from available proxies, our findings are immensely useful to practitioners. First, companies could use political identity to anticipate their customers’ postconsumption satisfaction and downstream behaviors when predicting and managing satisfaction levels and its outcomes such as sales, recommendation, and repurchase behavior in readily observable customer segments.

Second, understanding practical contexts that attenuate or strengthen differences in customer satisfaction and subsequent outcomes based on political identity can be useful to managers. For example, managers may expect the link between political identity and satisfaction to be stronger in categories and industries with higher levels of competition and option availability and in contexts that may involve unpleasant or negative consumption experiences (e.g., health care, air travel, or utilities). Conversely, this link may be weaker in less competitive industries and categories involving highly positive consumption experiences. Our findings also suggest the need for including political identity more systematically in studies that aim to identify factors that impact the link between customer satisfaction and downstream consequences.

Third, managers can use these results to better assess and measure the true underlying satisfaction levels of different customer groups and enhance customer satisfaction among different customer groups. For example, to further enhance the satisfaction in conservative segments, companies could highlight the availability of many options as well as customers’ choice freedom and agency at the time of purchase (Study 5a). This could be done, among other avenues, through tailored messages that highlight choice availability and freedom (e.g., “You have many options, and we are proud that you chose us”). In contrast, managers can enhance customer satisfaction in liberal segments by finding ways to increase customers’ trust in their consumption choices. This could be done by proactively crafting positive consumption experiences (Study 4) and, potentially, by highlighting the extent of customers’ prior category experience or expertise in communication messages. Corroborating the latter possibility, two follow-up studies (Web Appendix R) showed that customer satisfaction among liberals systematically increases when customers recall their extensive (vs. limited) experience with the product category or when customers’ perceptions of their category expertise is heightened through the framing of a firm's coupon promotion as being offered to the customer for being one of the brand's most loyal customers and for their long history and experience with the brand. Companies could thus remind customers of their existing expertise (e.g., “Your usage behavior shows that you are in the top 90% of all users”) or frame customer–firm relationship milestones to increase customers’ perceptions of their knowledge of the product or brand (e.g., “You have been with us longer than most of our customers”).

Fourth, conservatives are more satisfied than liberals, controlling for objective value or quality. Marketers using product ratings to gauge products’ objective performance should control for customers’ political identity in their analysis of product ratings.

Finally, this article offers useful public policy implications. Specifically, conservatives’ (vs. liberals’) higher satisfaction level with different experiences (including negative ones) suggests that communications about public policies may mobilize support in conservative audiences if they invoke or encourage anticipation of postexperience satisfaction levels. In addition, conservatives’ greater perception of free will and personal responsibility may be activated to motivate action. These insights may be useful when devising communication messages seeking to increase conservatives’ participation in initiatives such as COVID-19 vaccination, which are often less popular in conservative (vs. liberal) segments.

Supplemental Material

sj-docx-1-jmx-10.1177_00222429211057508 - Supplemental material for How Political Identity Shapes Customer Satisfaction

Supplemental material, sj-docx-1-jmx-10.1177_00222429211057508 for How Political Identity Shapes Customer Satisfaction by Daniel Fernandes, Nailya Ordabayeva, Kyuhong Han, Jihye Jung and Vikas Mittal in Journal of Marketing