Gender Moderates the Influence of Self-Construal Priming on Fairness Considerations

Flinkenflogel, Nic; Novin, Sheida; Huizinga, Mariette; Krabbendam, Lydia

doi:10.3389/fpsyg.2017.00503

ORIGINAL RESEARCH article

Front. Psychol., 03 April 2017

Sec. Personality and Social Psychology

Volume 8 - 2017 | https://doi.org/10.3389/fpsyg.2017.00503

Gender Moderates the Influence of Self-Construal Priming on Fairness Considerations

$\r\nNic Flinkenflogel*$ Nic Flinkenflogel¹^*

Sheida Novin²

Mariette Huizinga¹

Lydia Krabbendam¹

¹VICI Lab, Department of Clinical, Neuro- and Developmental Psychology, VU University, Amsterdam, Netherlands
²Department of Psychology, Utrecht University, Utrecht, Netherlands

Research in social and cultural psychology has identified that self-construal, or the way the self is defined in relation to others, plays an important role in social decision-making processes. Yet it remains difficult to isolate the effect of self-construal in a comparative approach. Therefore, we used priming methodology in three studies to induce either an independent or interdependent mindset to test direct consequences on fairness considerations. Specifically, we asked whether participants would accept an unfair ultimatum game offer: a split of 10 euros, where the participant is allocated the marginal share of 3 and the proposer 7. If the participant refuses, neither gets paid. In the first study, we used the well-known similarities and differences prime. Here, activating an interdependent mindset decreased rejection of the unfair offer compared to the independent mindset and control condition, but only in females. The prime did not affect males. In the second and third study we modified our university's mission statement to instead include either independent or interdependent values. Females displayed a similar direction of effects; in males however, activating an interdependent mindset increased rejection. Taken together, the results show that whether participants accept or reject an unfair offer depends on both their gender and the self-construal prime. The results were interpreted using the distinction between relational independence that has been associated with females, and collective interdependence, that has been associated with males. Possible consequences for future studies are discussed.

Introduction

Human social behavior does not follow the rational principles of economic models. Rather than maximizing individual profit, decision-making is driven by social preferences like inequality aversion, fairness, trust, and reciprocity (Weber et al., 2004; Balliet et al., 2009; Van Lange et al., 2013). Even when people are given the option to divide a sum between themselves and a different person without any possible repercussions or loss of face, the mode of behavior is generally a fair division (Hoffman et al., 1996; Diekmann, 2004). In addition, there is marked variation both within and between individuals, indicating that decision-making in social situations does not adhere to universal principles. Part of this variation can be explained by individual attributes such as gender and age (MacPherson et al., 2002; Deakin et al., 2004; Peters et al., 2007; Eckel and Grossman, 2008; Croson and Gneezy, 2009; Balliet et al., 2011), while a substantial amount also originates in the group the individual belongs to Oosterbeek et al. (2004), Wong and Hong (2005) and Gächter et al. (2010). However, it has proven difficult to disentangle the psychological mechanisms underlying differences in decision-making behavior (Grossmann and Na, 2014). Here we focus on self-construal as a plausible underlying psychological mechanism, and use priming in three experimental studies to further elucidate variation in the way people value outcomes for self and others.

Self-construal refers to the way the self is perceived and defined in relation to others. Cross-cultural studies demonstrate that self-construals differ between cultural groups. In predominantly individualistic cultures, as for example seen in Western European countries and North America, people typically perceive themselves as unique and independent from others (Markus and Kitayama, 1991; Triandis, 1993). The self is defined mainly in terms of internal attributes such as abilities and attitudes (Markus and Kitayama, 1991), and agency and assertiveness are valued traits, reflecting the independence of action and opinion (Singhal and Nagao, 1993; Kashima et al., 1995). In contrast, in predominantly collectivistic cultures, as for example seen in East Asian countries, people typically perceive themselves as interdependent (connected to others), and the self is to a much greater degree defined in terms of group memberships, relationships to family and friends as well as social roles (Markus and Kitayama, 1991; Triandis, 1993). Similarities with others and common goals are more important than in individualistic cultures (Trafimow et al., 1991). Harmony is highly valued and conflict is avoided and even suppressed (Ohbuchi et al., 1999; Carnevale and Leung, 2001; Tinsley and Brett, 2001; Cai and Fink, 2002).

These global differences in self-construal between individualistic and collectivistic cultures do not preclude further variation between individuals within these cultures (Na et al., 2010). Gender has been put forward as a prominent source of variation in self-construal, with on average a more interdependent self-construal in females (Cross and Madson, 1997). This is associated with a greater tendency to create close, intimate relationships, to take the perspective of others, and to maintain harmony and avoid conflict. Males on the other hand construct a more independent self-construal, accompanied by a preference for autonomy and emotional detachment from others. This distinction has been further refined by Baumeister and Sommer (1997), who argued that males equally construct the self as interdependent, but seek to connect to a broader group with a shared identity (e.g., supporting sports teams, patriotism, fraternities, etc.). Where females primarily seek intimate connections in a number of close personal dyadic relationships, males orient to a larger sphere by seeking to connect with a broader group; effectively seeking a larger number of relationships, but investing less in one on one interactions. The difference in interdependence between females and males has been referred to as the relational vs. collective distinction within interdependence (e.g., Kashima et al., 1995; Gabriel and Gardner, 1999; Maddux and Brewer, 2005). This has been demonstrated in a number of studies. For example, in an online trust-dilemma game, females were more trusting toward individuals with whom they shared a relationship or the potential of a future relationship, whereas men primarily trusted individuals if they were from the same social group (Maddux and Brewer, 2005). This is in line with the notion that compared to males, females tend to foster close relationships and prioritize harmony maintenance, conflict avoidance and togetherness more (Kashima et al., 1995; Cross and Madson, 1997). Similarly, Van Vugt et al. (2007) found that in males, but not in females, cooperation in a social dilemma game was increased when group cohesion was triggered by manipulating intergroup competition between the university they attended and a rival university.

The way individuals perceive themselves in relation to others influences their cognition, emotion and motivation (Markus and Kitayama, 1991; Nisbett et al., 2001; Grossmann et al., 2012). With regard to decision-making, individuals with an independent self-construal for example may rely on a cost-benefit strategy, which maximizes the benefit of one's choice for personal gain (Weber and Hsee, 2000). In contrast, individuals with an interdependent self-construal may be inclined to use role-based strategies in decision-making, focusing on the rules and expectations that come with certain social roles with the aim of fostering social connectedness (March, 1994; Weber and Hsee, 2000; Weber et al., 2005). There is indeed some empirical evidence linking independent and interdependent self-construals with different social decision-making strategies (Howard et al., 2007; Van Prooijen et al., 2008; Emonds et al., 2011). For example, in one cross-cultural study, individuals with high levels of independent self-construals displayed more competitive and less cooperative strategies than those with high levels of interdependent self-construals, independent of cultural group (European-American and Japanese) (Oetzel, 1998). Similarly, in a study investigating the relationships between self-construal, cultural background, gender, and conflict styles, Ting-Toomey et al. (2001) found that self-construal provided a better explanation for conflict style than gender and cultural background.

Experimental social dilemmas are well-suited to investigate the motives underlying social decision-making, because their design allows for quantifying the consequences of different choices for the individual and the other person(s) involved. Social dilemmas can be defined by two characteristics: (a) individuals receive higher payoffs for making selfish choices, and (b) everyone involved receives lower payoffs if everyone makes selfish choices (Weber et al., 2004). There is abundant evidence that decision-making in social dilemmas is not only influenced by individual monetary gain, but also by the moral cost or benefit associated with the action (Levitt and List, 2007; Sanfey, 2007; Rilling et al., 2008; Tabibnia et al., 2008; Balliet et al., 2009).

A widely used social dilemma is the Ultimatum Game (e.g., Roth et al., 1991; Botelho et al., 2000; Henrich et al., 2001; Gil-White, 2004; Hoffmann and Tee, 2006; Chuah et al., 2007, 2009; Güth and Kocher, 2014). This is a two-player social dilemma where a proposer suggests a division of a certain amount of money, which the responder can either accept or reject—with rejection resulting in neither party getting paid. The Ultimatum Game is a simple game: the participants make a single choice. Yet, the choice to accept or reject is a complex interplay between motivations such as inequality aversion, negative reciprocity, and costly punishment (Güth and Kocher, 2014). It is likely that these motivations will differ between people who view themselves and their actions, goals and priorities as independent from others; compared to people that construe an interdependent self-construal, viewing themselves as more connected to others, with shared goals and priorities.

Indeed, there is evidence for cultural and gender differences in the Ultimatum Game (Solnick, 2001; Oosterbeek et al., 2004; Eckel and Grossman, 2008; Espinosa and Kovářík, 2015). With regard to culture, a meta-analysis of 37 papers from 25 countries concluded that the average rejection rate of Asian responders was higher than of responders in the U.S., albeit with large heterogeneity between studies (Oosterbeek et al., 2004). However, the review was not able to establish a link between differences in rejection rate per country and cultural constructs as individualism. With regard to gender, findings have also been inconsistent. Using an Ultimatum Game design, Solnick (2001) reported that females were less likely compared to males to accept low offers, while Eckel and Grossman (2001) found that they wore more likely to accept such offers. Such inconsistencies are likely to be due to situational constraints. In a review of six social dilemma studies, Espinosa and Kovářík (2015) concluded that females compared to males displayed more prosocial behavior in experimental designs that highlight social cues. In contrast, males, but not females showed less prosocial behavior in decisions which were taken after deliberate processing compared to more spontaneous decisions (Espinosa and Kovářík, 2015). However, studies comparing either cultural or gender groups cannot address the underlying mechanisms for why differences are found (Oyserman, 2011; Oyserman et al., 2014). Rather than comparing groups that are supposed to differ in self-construal, or measuring individuals' trait self-construal and relating that to their decision-making, a more direct approach is to use priming methods to induce interdependent or independent mindsets and investigate the downstream consequences on fairness considerations in the Ultimatum Game.

Various techniques are available to prime interdependent or independent mindsets in the laboratory. For instance, Gabriel and Gardner (1999) used pronoun circling to prime an either independent or interdependent self-construal. Instructing North American participants to circle interdependent pronouns (we, ours) in a word search describing a trip to the city, resulted in a shift of world views reflecting collectivist social values and judgments, compared to when instructed to circle independent pronouns (I, mine). Trafimow et al. (1991) primed self-construal mindsets by instructing participants to think for 2 min what made them either different from their close friends and family (the independent condition) or what makes them similar (interdependent condition). Following the Similarities and Differences With Friends and Family prime, they were asked to complete 20 open-ended statements that started with “I am.” (Kuhn and McPartland, 1954). The participants who received instructions to think about what made them different from their close others reported more self-oriented responses reflecting personal attributes (e.g., I am clever, tall, etc.), while participants who received instructions to think about what made themselves similar reported more other-oriented responses reflecting social roles or qualities (e.g., I am a daughter, I am thoughtful of other's needs).

In the following, we report three studies that investigated the downstream consequences of independent and interdependent mindsets on fairness considerations, over and above individuals' trait self-construal. Participants were first primed with an independent or interdependent mindset, and then presented with a single 7-3 offer by a proposer in an Ultimatum Game. The 7-3 offer was used because prior work shows that the offer borders between being considered fair or unfair; most participants will accept a division of 50 – 40% of the total stake, while offers of less than 20% will most likely be rejected (e.g., Güth and Kocher, 2014). The 7-3 offer balances between triggering the motivation of monetary gain, and fairness considerations: while a division of 30% can be considered “unfair,” it is also better than receiving nothing. To exclude unwanted effects from proposer characteristics, we presented the offer as coming from an anonymous other. At the end of the studies participants completed a self-construal questionnaire. We predicted that an accessible interdependent mindset would result in a lower rejection rate of the Ultimatum Game offer, because it has been shown to bring connectedness and harmony to mind compared to an accessible independent mindset, that is shown to bring uniqueness and competitiveness to mind (Utz, 2004; Kim et al., 2006; Lee et al., 2010; Mourey et al., 2013). We predicted these effects to occur over and above participants' trait self-construals, but included trait self-construal in the analyses to explore possible interaction effects. In addition, we included gender for two reasons. First, an extensive collection of previous research has pointed out the role of gender in social decision-making (e.g., Eckel and Grossman, 2008; Croson and Gneezy, 2009; Espinosa and Kovářík, 2015). Second, gender has been attributed as a key factor in determining variations in self-construal (e.g., Cross and Madson, 1997).

Study 1

In the first study we tested our predictions as detailed next by using the well-known and often used Similarities and Differences with Friends and Family task (SDFF) as the priming task (Trafimow et al., 1991). Previous research has established that this task can effectively prime self-construal (Trafimow et al., 1991; Ybarra and Trafimow, 1998; Holland et al., 2004; Oyserman and Lee, 2008; Lee and Jeyaraj, 2014). Thinking of similarities with close others decreases subjective interpersonal distance, while thinking of differences enhances such distance. It has successfully been applied in a variety of experimental setups, showing consistent moderate to high effect sizes (Oyserman and Lee, 2008). In our study participants were randomized into one of the three conditions: no-prime control, independent mindset, and interdependent mindset condition. Following the prime, they were presented with the Ultimatum Game offer: participants were asked whether they would accept or reject a 10 euro division, in which they are allocated the marginal share of 3 euro. Our predictions were threefold. First, we predicted that the participants in the interdependent condition would be more likely to accept the offer than participants in the independent condition. Second, given that the study was conducted in the Netherlands, which is classified as a typically individualistic society where people are likely to have a trait independent mindset (Hofstede et al., 1991), we expected no differences in the rejection rates between those in the no-prime control and those in the independent mindset condition. Third, we predicted that these mindset effects would be shown over and above participants' trait self-construals.

Methods

Participants

A total of 226 university students participated in the study (47% male; 69% Dutch; the other 31% consisted of a diverse and fractured population: the largest subpopulations were from Suriname (4.4%), Morocco (4.4%) Turkey (4%), Iran (2.2%), and Germany (2.2%)¹; M_age = 22.6, SD = 5.8, age range: 18–33 years). Participants were randomly assigned into one of the three conditions: independent-mindset condition (n = 75; 47% male), interdependent-mindset condition (n = 75; 48% male), and no-prime control condition (n = 76; 47% male). The study allowed advanced registration online, or walk-ins. As the experiment was conducted in Dutch, students were required to speak Dutch fluently. Active informed consent was obtained, and experimental procedures were approved by the ethical committee of the Faculty of Behavioural and Movement Sciences of the Vrije Universiteit Amsterdam (VUA). As detailed below, participants were paid for participation.

Procedure and Materials

The study consisted of a prime followed by a one-shot Ultimatum Game and self-construal questionnaire, which were all computerized. Before participation, participants filled out a consent form, which described the study as a brief investigation of decision-making. The study started with a welcome screen and brief instructions about the Ultimatum Game. Participants were explained they would be presented with a proposal for a 10 euro split from an anonymous other student, who was randomly selected from the previous participant pool. They were also told that if they accepted the offer, a division would be made according to the proposed offer on top of the base fee; if they rejected nobody would receive an additional bonus. They were then told that they would first start with a neutral task, to clear their mind before commencing with the game.

Priming

The intermittent task framed as neutral, was the SDFF prime adopted from Trafimow et al. (1991). Participants in the independent-mindset condition were asked to think and write down for 2 min what made them different from their close friends and family. Participants in the interdependent-mindset condition were asked to write what they thought made them similar. In the no-prime control condition participants proceeded directly with the Ultimatum Game.

Ultimatum game

The social dilemma was a one-shot Ultimatum Game. The dependent variable was rejection rate: participants had to decide whether to accept or reject an offer where the anonymous proposer would receive 7 euro, and the participant 3.

Self-construal scale

The Ultimatum Game was followed by a trait self-construal scale (TSC) consisting of 10 items, of which half measured trait independent self-construal (e.g., “I don't care what other people think about me, as long as I am happy”; α = 0.71) and half measured trait interdependent self-construal (e.g., “Relationships with others are more important than my own accomplishments”; α = 0.75). The scale consisted of a modified version of the Oyserman (1993) scale, translated to Dutch. Participants rated themselves on a 7-point scale (1 = does not describe me at all to 7 = describes me very well).

Finally, participants were asked to indicate their age, gender, study enrollment, and ethnicity for demographic analysis. Participants were then debriefed; they were informed that the offer was fixed, and the neutral task consisted of a prime designed to make an independent or interdependent self-construal salient in the different priming conditions. At the end of the study participants were thanked and paid. Payment consisted of a base fee of €2, with a bonus of €0.30 if they accepted the proposed offer.

Analyses Plan

A logistic regression analysis was conducted to analyze whether a model with the predictors condition (independent prime, interdependent prime and no-prime control) and gender provided a better fit at predicting the binary outcome of rejection rate (accept or reject) on the Ultimatum Game than a null (intercept-only) model. A logistic regression model with categorical predictors examines pairwise interactions of rejection rate of gender across the different conditions; only comparisons between cells are analyzed (e.g., the difference between priming interdependence vs. independence for males, or the difference between males and females in the interdependent-prime condition), and does not compute the overall effect of condition or gender as an ANOVA analysis.

Pairwise interactions were examined to analyze rejection rate of gender across the different conditions. The odds ratio are reported with 95% confidence intervals. The estimated probabilities of rejection rate were indexed by the partial logistic regression coefficients (B-values) of the means of gender within the different conditions, means of conditions within gender, and the interaction between gender and condition.

In order to examine the extent to which trait levels of independence and interdependence moderated the outcome of condition on the Ultimatum Game, a ratio score of independence and interdependence scores was computed and included as a predictor in the analysis. A higher score on TSC indicated a higher ratio of interdependence. All two- and three-way interaction effects were included in the model.

Results

Descriptives

Overall, 42.7% of the participants in the interdependent-mindset condition rejected the offer (RO) in the Ultimatum Game, compared to 58.7% of the participants in the independent-mindset condition, and 56.6% of the participants in the control condition. To examine whether participants differed in trait levels of independence or interdependence as a function of condition and gender, a two-way ANOVA was performed with TSC as the dependent variable. Only a main effect of gender was found, F_{(1, 220)} = 8.39, p = 0.004, Cohen's d = 0.39. Overall, males reported a higher ratio of independence to interdependence (M = −0.23, SD = 0.69) than females (M = –0.02, SD = 0.38). TSC did not differ between the three conditions.

The Ultimatum Game

We conducted a logistic regression including condition, gender, TSC, and all two- and three-way interaction effects as predictors (Table 1). The analysis revealed a non-significant three-way interaction between the independent and interdependent-mindset conditions with gender and TSC: Wald $χ_{(1, N = 226)}^{2}$ = 2.92, p = 0.09. The two-way interaction between gender and the independent-mindset and interdependent-mindset conditions was significant [Wald $χ_{(1, N = 226)}^{2}$ = 5.17, p = 0.02].

TABLE 1

Table 1. Study 1: Regression model for the predictors condition, gender, CO; and the two- and three-way interaction effects on rejection rate.

The results showed that females rejected less when they were primed with interdependence (RO = 33.3%) compared to when they were primed with independence (RO = 62.5%), Wald $χ_{(1, N = 226)}^{2}$ = 7.31, p = 0.007, or when they were not primed at all (RO = 57.5%), Wald $χ_{(1, N = 226)}^{2}$ = 4.23, p = 0.04. The standardized regression values indicated that females in the interdependent condition were 3.7 times less likely to reject the offer than those in the independent condition, and 2.6 times less likely to reject than those in the control condition. Females in the independent condition did not differ from those in the control condition. Males on the other hand did not respond differently based on the different priming conditions (RO_independent = 54.3%, RO_{interdependent} = 52.8%, RO_control = 57.5%) (Figure 1).

FIGURE 1

Figure 1. Study 1: Average rejection rate of males and females in the priming conditions.

The difference between males and females was only significant in the interdependent mindset condition, where males rejected more often than females, Wald $χ_{(1, N = 226)}^{2}$ = 3.81, p = 0.05. As expected, the interactions with trait self-construal were not significant.

Discussion

We predicted that participants in the interdependent mindset condition would reject the offer less than those in the independent mindset and the no-prime control condition, over and above participants' trait self-construals. Our results partially support these two predictions. While an accessible interdependent mindset indeed resulted in a lower rejection rate, this was only true for females. Male participants' rejection rate of the offer did not differ as a function of condition. These findings were not influenced by participants' trait self-construals. Moreover, despite that females reported higher trait levels of interdependent self-construal, there were no differences between females and males in the no-prime control condition. These findings suggests that participants' gender affects sensitivity to the mindset prime, rather than a priori resulting in different decisions in the Ultimatum Game.

These findings are in line with a recent review concluding that there are no differences between males and females in social dilemmas with a neutral experimental framing, while cues highlighting the social context of the dilemma lead to more prosocial behavior in females (Espinosa and Kovářík, 2015). Other results also coincide with the notion that females are generally more sensitive to the relational aspects of interdependence than males (Cross and Madson, 1997; Maddux and Brewer, 2005). In our study, thinking and writing about similarities with close others could have triggered these relational aspects of interdependence, and may therefore have had a stronger effect in females than males.

In contrast, it has been suggested that males are more sensitive to the collective aspect of interdependence (Baumeister and Sommer, 1997; Gabriel and Gardner, 1999). This collectivity refers to feeling connected to peers with a shared identity or group membership. The distinction between relational and collective interdependence has been put forward to explain gender differences in sociality in terms of the male and female motivations to connect to different spheres (Baumeister and Sommer, 1997). This difference may also affect outcomes in social dilemma's. In the following two studies we further examined if the nature of the prime may have triggered the gender difference in the effect on participants fairness considerations. As Study 1 suggested that females are more than males influenced by having relational interdependence on their mind, we tested whether males would be more influenced by a prime triggering collective interdependence in Study 2.

Study 2

We developed a new prime using a modified university mission statement. Participants read a text about collective norms of how students at the university are expected to behave among their peers. Participants were randomly assigned into one of the two conditions: independent mindset and interdependent mindset condition. Participants in the independent mindset condition read about the importance of being independent, having one's own opinion, being unique, and having personal goals. Participants in the interdependent mindset condition read about the importance of being honest, equal, social, and interpersonal. The crucial difference between this new prime and the previous Similarities and Differences prime is that in Study 2 we tap into a collective identity (that of the university to which the students belong). Following the prime, participants were again asked whether they would accept or reject a 7-3 offer in an Ultimatum Game.

Our predictions were twofold. First, we predicted that the participants in the interdependent condition would be more likely to accept the offer than participants in the independent condition. Yet, given that the interdependent prime now elicits collective rather than relational interdependence, we expected that the condition effect would be more pronounced in males than females. Second, we predicted that these mindset and gender effects would be shown over and above participants' trait self-construals. As we had established a baseline rejection rate in a no-prime control condition, we did not include a control condition in order to allocate the available population to the two priming conditions to ensure enough statistical power.