Introduction

Interracial dialogue has long been viewed as central to improving race relations and ameliorating social justice (e.g., Walsh 2006; McCoy and Sherman 1994; Schoem 2003; Nagda et al. 1999). Indeed, research has demonstrated that participants in structured dialogue programs may increase their understanding of inequality (Nagda et al. 2009), expand intergroup empathy (Sorensen 2010), enlarge cross-race networks (Wernet et al. 2003), and become racial allies to one another (Alimo 2012). However, as with any form of intergroup contact, interracial dialogue invites hindrances (e.g., intergroup anxiety, Turner et al. 2008), encounters limits (e.g., generalization of the contact, Kenworthy et al. 2005), and may even produce negative outcomes (e.g., increased prejudice, Barlow et al. 2012). Consequently, it is important to understand the dynamics of such dialogue more fully—including both the promising and perilous aspects—so that it can be used most effectively.

One challenge frequently faced in interracial dialogue is a fatigued resistance to, and resentment of, the conversation. Referred to as “shutting down” (Cargile 2010), dialogue participants can often be seen “hardening their conflicting positions and turning deaf ears to one another” (Fishman and McCarthy 2005). This resistance can take many forms, including silence, passive-aggressiveness, microinvalidations, absenteeism, and even overt hostility (Chan and Treacy 1996; Higginbotham 1996; Wong et al. 2014). In one witnessed example, a white American student responded to African American testimonies of race-based suffering by goading her classmates to “start looking at life as a pursuit for yourself as an individual and quit worrying throughout life about your race or culture!” Predictably, these words of “encouragement” were met with affirmations of why race (still) matters in the USA. Afterward, the dialogue deadlocked. As any discussion participant knows, invalidating responses and communication breakdowns (such as this) on the subject of race are not uncommon. Thus, if we hope to maximize the promise of interracial dialogue, we must understand more fully the origins, forms, and consequences of one peril in particular resistance to dialogue.

Fortunately, resistance in the context of interracial dialogue has received a good deal of attention in both critical (e.g., McLaren 1993; Hytten and Warren 2003; Marx and Pennington 2003; Giroux 1997; Gutiérrez-Jones 2001) and qualitative circles (e.g., Sue 1999; Rich and Cargile 2004; Sue et al. 2009; Miller and Donner 2000; Johnson et al. 2008; Williams and Evans-Winter 2005; Mio and Barker-Hackett 2003; McKinney 2006). Even so, it is not typically investigated using quantitative approaches. Of course, there is an enormous body of social scientific research regarding constructs and processes that can inform our understanding of resistance in the context of interracial dialogue (e.g., Rudman et al. 2001; Ensari and Miller 2006; Asbrock et al. 2012; Cikara et al. 2011; Dovidio 2013; Dovidio et al. 2010; Hewstone and Brown 1986; Lewis et al. 2000; Miller et al. 2004). Yet among these studies, relatively few have examined resistance to interracial dialogue.

As already mentioned, resistance can take many forms and is often recognized implicitly, but can be challenging to study with explicit measures. Nevertheless, several studies have quantitatively assessed resistance in the context of interracial dialogue, most typically in the form of avoidance, non-affiliative behaviors, or derogation (see Shelton et al. 2006). For example, it has been observed that high levels of interracial anxiety are associated with increased avoidance of interracial interactions (Plant 2004; Plant and Devine 2003). Once engaged in interaction, negative implicit attitudes (Dovidio et al. 2002), as well as race–topic avoidance, have been associated with the increased use of non-affiliative behaviors (Apfelbaum et al. 2008). Whether in the form of fidgeting, blinking, decreased eye contact, or increased interpersonal distance, white Americans have demonstrated greater resistance in the context of cross-race relative to same-race interactions (Trawalter and Richeson 2008). Moreover, this nonverbal resistance has been found to manifest in direct proportion to white Americans’ fear of being perceived as a racist (Goff et al. 2008), as well as beliefs about their cross-race partner’s openness (Butz and Plant 2006).

Perhaps the clearest form of resistance observed has been derogation (e.g., attributions of “complaining”). A series of studies has documented that when claimants attribute hardship to systemic discrimination rather than personal failings—as often occurs when counternarratives appear in interracial dialogue (e.g., Williams 2004)—participant/observers “blame the victim,” even in the presence of corroborating evidence (see Kaiser 2006). In one recent study of employment evaluations (Dover et al. 2014), white American participants were more likely to derogate a Latino who claimed workplace discrimination than Latino participants. However, among Latino participants, those who endorsed system-justifying beliefs (e.g., hard work equals success) were just as likely as white participants to derogate the claimant when they were told that the company had won an unspecified “diversity award.” Although not studied in the context of interracial dialogue, such results nevertheless highlight the prospect that our resistance to accepting another’s race-related testimony may be rooted in our complex relationships to the system of racial stratification.

Taken together, these studies both contribute to our understanding of resistance and also point to needed areas of research. Specifically, because resistance to interracial dialogue unfolds in complex and interactive patterns, we need to begin cultivating a robust foundation of quantitative data that is both multidimensional and attuned to the specific circumstances of this dialogue. Rather than investigating single cause–effect relationships, multiple times in tangentially related circumstances, we need more research that addresses potential interaction effects among several different variables in contexts that possess greater ecological validity (Sigley 2003). In view of this, the present study was designed to employ quantitative methods in a novel investigation of the interactive and multidimensional relationships between three concepts central to interracial dialogue: openness to the other, system threat, and prejudice.

Because human perception is undeniably dependent on categories, prototypes, and stereotypes (Bodenhausen et al. 2007), individual prejudices created by such schemas undoubtedly play a major role in fostering resistance to cross-race interactions. In addition, because the motivations behind individual prejudices are diverse, it is important to view their role in relation to circumstances that may (or may not) evoke them. Consequently, this study investigated two forms of individual prejudice in relationship to two forms of a feature central to interracial dialogue: testimonies of race-based suffering. Because system justification is chief among the motives for prejudice (Uhlmann et al. 2010), two testimonies were selected to represent different degrees of system threat so that their interaction with two types of participant prejudices could be studied. Alongside highlighting such potential interactions between prejudice and the topic of dialogue, this study was also designed to offer a multidimensional look at resistance to interracial dialogue by employing a novel measure of openness to the other extending across cognitive, behavioral, and affective domains.

Openness to the Other

One manner of understanding resistance is as the lack of openness. Openness is, of course, a concept widely applied across disparate fields. In interpersonal communication contexts, it is typically understood as the predilection toward or enactment of self-disclosure and self-expression (e.g., Papini et al. 1990; Norton and Montgomery 1982). This view of openness is, however, ill fit to interracial dialogue because here the central concern does not typically regard participant expressiveness. Instead, openness in this context involves a willingness to listen and acknowledge others. In this sense, openness is akin to the psychological construct “openness to experience”—a broad and general dimension of personality defined by several facets, including behavioral flexibility (i.e., willing to go new places), intellectual curiosity (i.e., willing to consider new ideas), and unconventional attitudes (i.e., a readiness to re-examine social values) (Costa and McCrae 1992). Despite this kinship however, openness in multicultural settings cannot be reduced to a dimension of personality because it is fundamentally relational in nature. Instead, it is perhaps best described by the concept “openness to the other” (see Fowers and Davidov 2006).

Openness to the other is a popular philosophical concept (e.g., Levinas 1981; Noddings 1995; Derrida 1978) well suited to understanding dialogue. As Gadamer explains,

In human relations the important thing is… not to overlook [the other’s] claim but to let him really say something to us. Here is where openness belongs… This openness does not exist only for the person who speaks; rather anyone who listens is fundamentally open. Without such openness to one another there is no genuine human bond (emphasis added; 1982).

Openness to the other thus shares the same sense of invitation as “openness to experience,” but it is firmly situated and fundamentally relational in nature (i.e., not a dimension of personality). Thus, provided a specific context and relational partner, openness to the other may be defined by an individual’s ability to “take seriously the truth claims that cultural group [members] make” (Fowers and Davidov 2006) while demonstrating an “willingness to engage” (Giroux 2004).

Openness to the other has previously been employed in qualitative studies (e.g., Ezzy 2010; Mihelj et al. 2011), but the question remains how to validly assess it using quantitative measures. Because it is too rich to otherwise ignore, this study proposes employing a novel, multidimensional trio of existing quantitative measures in an effort to gauge openness to the other in an ecologically valid manner: the Individualized Trust Scale (Wheeless and Grotz 1977), a word count of verbal responsiveness (Tausczik and Pennebaker 2010), and bilateral electrodermal activity (Banks et al. 2012).

The Individualized Trust Scale is herein proposed as a self-report of one’s inclination to take seriously the truth claims of another. As such, it measures an individual’s cognitive state during dialogic interaction. In turn, this self-report is supplemented by a behavioral indicator—namely a word count of verbal responsiveness (i.e., how much is actually said to another in the context of interracial dialogue). In the absence of a reliable coding scheme suited to indicate openness in dialogue, a word count of verbal responsiveness is proposed here as an initial indicator of one’s “willingness to engage.” Individuals who produce few utterances are by definition less behaviorally engaged than those who speak at length; thus, quantity of talk is a face-valid behavioral measure of engagement. Moreover, longer responses have been found to indicate greater empathy (Smith et al. 1989), increased verbal person centeredness (Burleson and Samter 1985), greater involvement (Leshed et al. 2007), and intentional social support (Galegher et al. 1998). Of course, more talk can also indicate antisocial behaviors, such as lying (Hancock et al. 2007). In the context of interracial dialogue however, behaviors anecdotally and qualitatively tied to greater resistance (and less openness) typically manifest with a fewer (rather than a greater) number of words (Higginbotham 1996; Sleeter 1996). For example, Sue et al. (2009) report that behaviors read by dialogue participants as signs of resistance included “eyeball rolling, shifting or slouching in chairs, doodling, turning red, avoiding eye contact or looking down, fidgeting, [and] becoming quiet” (emphasis added, p. 187).

Alongside cognitive and behavioral indicators, bilateral electrodermal activity (i.e., skin conductance capacity) is offered here as a third affective indicator of openness to the other. Openness in dialogue requires a stance that is relational and prosocial in nature. Because, as just discussed, much of this stance is communicated nonverbally, it is important to observe affective states as well as self-reports and expressed behavior. Moreover, because much of this affect is processed and expressed unconsciously (Banaji and Greenwald 2013), it is also important to employ a measure that has the capacity to indicate latent prosocial or antisocial affect. Bilateral electrodermal activity (EDA) holds promise of being just such a measure.

It is already well established that electrodermal measurement is excellent indicator of autonomic nervous activity (Dawson et al. 2000) and, as such, is related to nearly all affective processes (Porges 1997). Even so, the orthodox view is that its capacity to indicate emotional valence is suspect (Lang et al. 1993; Figner and Murphy 2011). Measured unilaterally, this indeed may be the case, but bilateral measurement is beginning to suggest something else.

Over the past decades, bilateral EDA has been well studied in certain psychiatric patient populations (e.g., those with depression or schizophrenia; Myslobodsky and Horesh 1978; Öhman 1981), but has otherwise remained relatively neglected. However, new technologies and theoretical developments now encourage scholars to consider lateralization in all populations. In particular, the motivation lateralization theory (Harmon-Jones 2003) and other asymmetrical models (Davidson et al. 1990; Davidson 1992, 1993; Davidson et al. 2000) suggest that left-sided neural activity is involved in approach-related emotions and right-sided activity correlates with withdrawal-related states. More particularly (and most relevant to EDA), left limbic activity has been associated with empathic concern (Singer et al. 2004; Lang et al. 2011) and right limbic activity with fear (Lanteaume et al. 2007). Because fMRI studies demonstrate that skin conductance is well integrated with the neural systems that control such motivational dispositions (Critchley et al. 2000), bilateral activity likely contains important information regarding approach and withdrawal-related affect. Specifically, because EDA is firmly directed ipsilaterally (Mangina and Beuzeron-Magina 1996) by limbic (i.e., “emotional”) brain systems—and influenced contralaterally by secondary cortical control (Sequeira and Roy 1993) mostly with regard to attention and orientation (Dawson et al. 2000)—it is likely that left- and right-sided responses are indicators of approach and withdrawal emotions, respectively. Indeed, one recent study demonstrated just that participants showed greater left-sided electrodermal activity in response to neutral and happy faces and greater right-sided activity in response to disgust and fear faces (Banks et al. 2012). Consequently, EDA laterality is proposed here as an indicator of affective openness to the other.

To summarize, openness to the other is an important philosophical concept well suited to understanding dialogic processes. In an effort to leverage the concept for quantitative research purposes, this study offers an innovative operationalization through the use of a trio of existing measures. However, because the concept is also relational in nature, the ‘other’ to whom one opens must also be given due consideration. In the context of interracial dialogue, this can be done by employing the construct of system threat.

System Threat

According to system justification theory (Jost and Hunyady 2002), people are often psychologically motivated to defend the status quo, even or especially when the system is “broken.” They tend to hold more favorable attitudes toward the system than otherwise warranted simply because it already exists. Thus, individual motivation is akin to the idiom “better the devil you know than the devil you don’t.” Variation in support of the system can be predicted by both dispositional and situational attributes. People intolerant of uncertainty more frequently adopt system-justifying, conservative ideologies (Jost et al. 2007) and conditions of system threat engender increased defense of the status quo (Ullrich and Cohrs 2007). In other words, because some people are attached to and are therefore motivated to uphold the status quo (i.e., system justification motive), they are likely to respond defensively (and perhaps irrationally) in situations that are perceived to be an attack on the system (i.e., system threat).

System threats are defined as events or activities that are “potentially threatening to the legitimacy or stability of the social system” (Jost 2011) and they can take either direct or indirect form. Some threats stem from direct attacks on the system, such as terroristic or revolutionary events. Other activities pose more indirect challenges through criticism or other revelations of the system’s shortcomings that call into question its effectiveness or legitimacy. Although previously not investigated in the context of interracial dialogue, the construct of system threat provides useful terms for understanding resistance. Because interracial dialogue in the USA is most often constituted by arguments and testimonies centered around extant racism, it is replete with messages that pose indirect threats to the system.

Qualitative research has revealed that among the many patterned behaviors of interracial dialogue, one of the most common includes people of color offering testimony of their own oppressive experiences (e.g., Drew 2012; Srivastava and Francis 2006; Simpson 2008). Such testimonies can be classified as accounts of “social suffering”—that is, stories of distress engendered by cultural, social, and political forces (Graubard 1996) which “become embodied as individual experience” (Farmer 2003). As many dialogue participants recognize, these testimonies function as “social utterances which intervene in a present social context, rather than [as] simple representations of a past event” (Kennedy 2004). They are thus frequently employed as discursive moves in an attempt to interrupt the dominant group’s “master” narratives and unmask oppressive structures (Anderson et al. 2007). Provided that testimonies of race-based suffering commonly appear in interracial dialogue, speakers who provide such testimony are one class of important “others” to whom participant openness is relevant and should be assessed.

Of course, not all speakers who offer testimony of race-based suffering are same. There are a wide variety of characteristics that could ostensibly affect the openness of dialogue participants, including the speaker’s accent, pitch, rate, and ascribed gender (see Bradac et al. 2001). However, chief among all variables worth considering is the testimony itself. Does participant openness to the other vary as a function of the testimony offered by the speaker? According to system justification theory, indirect threat messages may be deflected through victim blaming which places responsibility for a negative event on the individual rather than on the system (Jost et al. 2005). Consequently, it is likely that testimony representing greater system threat will engender less trust, behavioral engagement, and prosocial affect in relation to the speaker than testimony representing less system threat. Despite this, the impact of system-threatening testimony may also depend on individual differences in listener prejudice. Perhaps only more prejudice listeners will be sensitive to testimony content and thus respond to decreasing openness to the other as the testimony increases in degrees of system threat.

Prejudice

While some participants may exhibit openness in the face of all testimony, others may not demonstrate much openness at all, regardless of the testimony. As a result, it is important to begin exploring the potential role of individual difference variables in shaping responses to testimonies of non-dominant cultural experiences that challenge the status quo. Although any number of variables may ultimately predict such individual difference (e.g., mindfulness or uncertainty intolerance), the two most important constructs of prejudice within the individual difference literature serve as an appropriate starting point for the present investigation: right-wing authoritarianism (RWA) and social dominance orientation (SDO). These two constructs have been the most widely investigated and, together, most capably explain the generality of prejudice across different groups and circumstances (Duckitt and Sibley 2007).

Both RWA and SDO are ideological constructs (Duckitt 2001) that can be understood as forms of attachment to the status quo. Individuals high in RWA are described as submissive to authority and protective of traditional societal norms (Altemeyer 1981). Those high in SDO are also characterized by authoritarianism (Altemeyer 1998), but an aggressive (not submissive) strain which reflects a desire “that one’s in-group dominate and be superior to outgroups” (Pratto et al. 1994). In either case, both variables indicate the extent to which individuals endorse (or even desire to elevate) the status of the dominant group. Conceived thusly, it is unsurprising to find that both RWA and SDO have been found negatively related to “openness to experience” (e.g., Lee et al. 2010, however cf., Sibley and Duckitt 2008), as well as other forms of multicultural competence (e.g., Dru 2007). Together, these constructs suggest that although some individual participants in interracial dialogue may open themselves up to other (racialized) standpoints, others (high in RWA or SDO) may simply not want to hear it because dialogue about racism may represent threat to a system in which they are ideologically invested. In a racist system, racist ideologies and system ideologies are confounded as the yoking of racist attitudes with both RWA and SDO demonstrates (e.g., Hiel and Mervielde 2005; Duriez and Soenens 2009).

Though these constructs are similar in many respects, RWA and SDO are nevertheless well differentiated (e.g., Thomsen et al. 2008; Duckitt and Sibley 2010). As Crawford et al. explain, “RWA and SDO originate from divergent social and psychological bases. Specifically, RWA originates from perceptions of the world as a dangerous place, whereas SDO originates from a view of the world as a competitive jungle” (2013). Individuals high in RWA emphasize conservatism (i.e., maintaining the status quo), authoritarianism (i.e., coercive control), and traditionalism (i.e., old-fashioned values) in an effort to generate a sense of self-protective security (Duckitt 2001; Duckitt et al. 2010). Individuals high in SDO, on the other hand, emphasize the status quo but with the particular aim of maintaining the unequal distribution of resources. As Sidanius and Pratto describe, group-based hierarchies help ensure that “members of dominant groups secure a disproportionate share of the good things in life (powerful roles, good housing, good health), and members of subordinate groups receive a disproportionate share of the bad things in life (relatively poor housing and poor health)” (2012). Individuals high in SDO thus seek to enhance the hierarchy and often accomplish this through the endorsement of “legitimizing myths” (i.e., system-justifying beliefs).

Because of their divergent social and psychological bases, individuals high in SDO are expected to be more sensitive to increasing degrees of system threat than those high in RWA. SDO is primarily concerned with intergroup power which, as just mentioned, is often reinforced via legitimizing myths (Sidanius and Pratto 2012). As a result, among individuals high in SDO, speakers who offer testimonies representing greater system threat may be rebuffed more than speakers with less threatening testimony. RWA, on the other hand, is concerned with maintaining the status quo in an effort to psychologically shield oneself from a dangerous world. Thus, any testimony of social suffering may be experienced as alarming and therefore engender a defensive interpersonal posture, regardless of the degree of system threat posed. Based on the above arguments, this study was designed to test the following three hypotheses.

H1

A testimony representing greater system threat will engender less trust, behavioral engagement, and prosocial affect than a testimony representing less system threat.

H2

Speaker testimony will interact with participant levels of SDO such that individuals high in SDO will exhibit less trust, behavioral engagement, and prosocial affect in response to a testimony of greater system threat than in response to one of less system threat.

H3

RWA will have a main effect across both types of testimonies such that individuals high in RWA will exhibit less trust, behavioral engagement, and prosocial affect than those low in RWA.

Materials and Methods

Participants and Procedures

Participants were 47 undergraduate students at a large urban university in Southern California, recruited in class to take part on a voluntary basis. Study protocol was approved by the university’s institutional review board, and informed consent was collected at the outset. Because participants were asked to respond to testimonies of race-based suffering that are best understood only with experience of the US system of racial hierarchy (e.g., Bonilla-Silva and Lewis 1999), non-citizens were not included in the analysis. Past experience working with this student population indicated that most non-citizens would be international student sojourners without full, embodied knowledge of the system; thus, they were excluded a priori. In addition, given the obvious relevance of African American group membership to the testimonies used here, African American participants (n = 2) were also excluded in order to limit this study to outgroup responses only. Lastly, in order to ensure that results from this small sample were not unduly affected by outliers, an influential case analysis was conducted. Inspection of both leverage and influence statistics indicated one such case, a Caucasian female; she was subsequently excluded from the analyses reported below. This left a sample of 36, including 12 males and 24 females. They were on average 22.19 (SD = 2.58) years old and reported a variety of racial/ethnic backgrounds (13 Caucasian, 12 Hispanic, 4 Asian, 6 other, and 1 declined to state).

After consent was collected, participants completed a questionnaire including RWA and SDO measures and were then bilaterally fit with two Q-sensors (a small, commercially available device that measures skin conductance). Next, participants were presented with an eight-step problem of basic math operations designed to encourage cognitive activation as part of the manufacturer’s recommended protocol to calibrate the devices. This was followed by a five-minute rest period, after which participants donned a microphone headset and then finally listened and responded to two testimonies of social suffering via computer.

Measures and Materials

RWA and SDO

Right-wing authoritarianism was assessed using the short (18-item) form of the ACT Scales measure (Duckitt et al. 2010). Sample items included “Our country will be great if we show respect for authority and obey our leaders” and “The ‘old-fashioned ways’ still show the best way to live.” Although the ACT Scales allow for multidimensional investigations of the traditionally unidimensional RWA construct, only a single, combined score was used here given the preliminary nature of this investigation (α = .80). SDO was measured using the short (8-item) form of the original scale (Pratto et al. 1994), which demonstrated acceptable reliability (α = .75). Sample items included “To get ahead in life, it is sometimes necessary to step on other groups” and “Some groups of people are simply inferior to other groups.” Both RWA and SDO were measured using a 9-point response scales (1 = very strongly disagree, 9 = very strongly agree). Means, standard deviations, and ranges for these, as well as all other measures, can be found in Table 1. As a note, the mean RWA score for this sample is comparable to other college samples, though the mean SDO score is less than typical (cf., Crawford and Pilanski 2014).

Table 1 Correlations among and descriptive statistics for study variables
Full size table

Testimonies of Race-Based Suffering

Although openness to the other should ideally be assessed in the context of spontaneous or even confederate-enacted interracial dialogue, experimental demands require both control and replication. With spontaneous dialogue, the context would invariably change with each participant and with confederate-enacted dialogue, the emotional quality of the testimony would be impossible to replicate consistently. For these reasons, recorded testimonies of genuine social suffering were employed in order to provide a suitable and ecologically valid environment for the assessment of openness to the other.

Because ecologically valid stimulus materials were of paramount importance, two true stories of racism were used here. The stories were audio recordings (only) taken from video clips in a series of oral histories (Glide Racial Justice n.d.)—one told by a man named Terry describing a time when he was treated differently than white customers when purchasing a pair of shoes and another told by a woman named Diane describing her experiences in New Orleans during hurricane Katrina (see “Appendix” section). Terry’s story was selected to represent testimony of “individual racism” (i.e., events experienced on a personal level, Dovidio and Gaertner 1986). As an ostensibly isolated event involving few people, it was intended to provide considerable, but not maximum system threat. On the other hand, Diane’s story was selected to represent testimony of “collective racism” (i.e., organized racial others seeking to restrict the rights of African Americans, Utsey and Ponterotto 1996). More specifically, because her account of racial discrimination in the wake of hurricane Katrina dramatically highlights widespread failings that call into question the legitimacy of system, it was intended provide a high degree of system threat (Napier et al. 2006). A manipulation check conducted using a separate sample of participants from the same population indicated that the woman’s testimony (N = 18, M = 5.278, SD = .752) was indeed perceived as more system-threatening than the man’s testimony (N = 18, M = 4.722, SD = 1.127), t(17) = 2.557, p = .010, 95 % CI [−1.013, −.972], d = 1.205. Both audio clips were approximately 1:45 min in length and were presented in random order. Order effects were assessed for all of the dependent measures and were not found to be significant.

Openness to the Other

As described earlier, openness to the other was operationalized as a multidimensional measure of trust, verbal responsiveness, and autonomic nervous activity. Trust of the speaker offering testimony of social suffering was measured using four adjective items from the Individualized Trust Scale (trustworthy, candid, honest, reliable; Wheeless and Grotz 1977). A 7-point response scale was used, and with all items averaged, the measure was found to demonstrate acceptable reliability for reactions to both the male (α = .69) and female speakers (α = .86).

After listening to the first, randomly selected speaker’s testimony, but before completing the self-report measure of trust, participants (who were wearing a microphone headset) were asked to “imagine that this person told this story to you” and then to respond orally (i.e., “what would you say, if anything, to this individual?”). Their responses were audio recorded, transcribed, and submitted to a word count as a behavioral measure of verbal responsiveness. Participants then listened to the second speaker’s testimony and completed the same procedures.

Simultaneous to their oral responses, participant bilateral EDA values (skin conductance level) were recorded using the Q-sensor. A six-second sample (48 data points) was registered nine seconds following the response instructions, allowing participants time to read the instructions (i.e., six second average in pretesting) as well as providing a three-second latency window. Because EDA levels can vary widely between participants (Crider 1993), this study employed a within-subject design to test the effect of system threat (i.e., responses to different testimonies of social suffering), on the laterality of EDA activity. Instead of simply gauging mean-level activity, skin conductance levels were log-transformed (Venables and Christie 1980) and then used to calculate a mean laterality index [(left value − right value)/(left value + right value)] as an indicator of affective openness to the other. Higher scores on this index indicate more left-sided, “approach” affect.

Results

In order to address the first research question, a doubly multivariate, repeated measure ANOVA was conducted in order to determine whether a variate comprised of participant trust, word count, and EDA laterality differed in response to the two different testimonies, as well as in response to between-subject levels of SDO and RWA. Because the word count data exhibited a positive skew, it was log-transformed (along with the EDA data, as noted above) before submitting to parametric testing. Results indicated that speaker testimony had a marginal effect on openness to the other (i.e., the variate), Wilks’ lambda = .773, F(3,27) = 2.636, p = .070, partial η 2 = .227, interacted with SDO scores, Wilks’ lambda = .632, F(3,27) = 5.236, p = .006, partial η 2 = .368, but did not interact with RWA scores, Wilks’ lambda = .875, F(3,27) = 1.290, p = .298, partial η 2 = .125. Although not interactive in their impact, RWA scores were found to have a main effect on openness to the other, Wilks’ lambda = .532, F(3,27) = 7.926, p = .001, partial η 2 = .468.

Subsequent univariate analysis indicated speaker testimony impacted trust ratings, partial η 2 = .222, F(1, 29) = 8.271, p = .007, but neither word counts, partial η 2 = .005, F(1, 29) = .155, p = .696, nor EDA laterality, partial η 2 = .006, F(1, 29) = .183, p = .672. The woman’s testimony about hurricane Katrina engendered less trust among participants than the man’s testimony of individual racism (see Table 1). In addition, the interaction between speaker testimony and SDO scores explained differences in both trust ratings, partial η 2 = .286, F(1, 29) = 11.639, p = .002, and word counts, partial η 2 = .153, F(1, 29) = 5.227, p = .030, but not EDA laterality, partial η 2 = .007, F(1, 29) = .194, p = .663. Scatterplots illustrating the significant interactions are presented in Figs. 1 and 2. As suggested by the regression line slopes, participants who reported higher levels of SDO demonstrated more openness to the other (i.e., higher trust and greater word counts) when reacting to the man’s less system-threatening testimony of individual racism than the woman’s testimony of collective racism. Univariate analysis also indicated a significant effect of RWA on trust, partial η 2 = .195, F(1, 29) = 7.017, p = .013, EDA laterality, partial η 2 = .165, F(1, 29) = 5.727, p = .023, and marginally on word count, partial η 2 = .102, F(1, 29) = 3.310, p = .079. Negative correlation coefficients indicate that participants reporting higher levels of RWA spoke less, reported less trust, and displayed less left-sided EDA activity when responding to either the man’s or the woman’s testimony of racism (word count, r = −.356, p = .004; trust, r = −.423, p = .000; EDA laterality, r = −.373, p = .002).

Fig. 1
figure 1

Scatterplot of the relationship between participant social dominance orientation (SDO) and speaker trust ratings (Trust) across reactions to both the male and female testimony of individual and collective racism, respectively

Full size image
Fig. 2
figure 2

Scatterplot of the relationship between participant social dominance orientation (SDO) and raw word counts (WC) across reactions to both the male and female testimonies of individual and collective racism, respectively

Full size image

In order to assess whether these findings from a racially heterogeneous outgroup sample were representative of Anglo Americans, the analyses were rerun including only this subsample of participants (n = 13). Unsurprisingly, none of the effects were statistically significant (given the extremely small sample size), but estimates of the direction and magnitude of the effects (partial η 2) were nearly identical (i.e., ±.10) in all cases except three. These included the univariate interaction effects of speaker testimony and SDO on word count (sample partial η 2 = .153; subsample partial η 2 = .303) and trust (sample partial η 2 = .286; subsample partial η 2 = .125), as well as the univariate effects of RWA on trust (sample partial η 2 = .195; subsample partial η 2 = .004). Thus, although reported trust may be less impacted by both RWA and the testimony by SDO interaction, all of the other effects observed in this heterogeneous sample are likely to be the same, if not greater, among a homogenous sample of white participants.

Discussion

As Neisser describes, controlled studies “usually use stimulus material that is abstract, discontinuous, and only marginally real” (1976). In contrast, the present study used a multidimensional measure of openness to the other in the context of real-life testimonies of social suffering and succeeded in providing preliminary answers to three hypotheses concerned with the circumstances of resistance to interracial dialogue. Hypothesis 1 was partially supported as testimony representing greater system threat engendered less trust, but not fewer words or decreased EDA laterality. Hypothesis 2 was more robustly, but not entirely, supported as speaker testimony interacted with SDO to predict less trust and behavioral engagement, but not less prosocial affect among participants high in SDO when listening to the woman’s testimony regarding hurricane Katrina. Lastly, hypothesis 3 was fully supported as participants with higher levels of RWA were found to exhibit less trust, behavioral engagement, and prosocial affect for both speakers compared with those with low levels in RWA.

As predicted by hypothesis 1, the woman’s testimony was found to engender less trust, but was not associated with significantly fewer words or less left-sided EDA. Despite this, the direction of the word count difference was in the predicted direction (see Table 1). Thus, the low power of this analysis (produced by the small sample size) is a likely explanation for these nonsignificant results. Explaining the results with respect to EDA laterality is, however, more problematic.

As suggested by the high correlation of EDA laterality across conditions of speaker testimony (r = .871, p = .000), this affective measure did not turn out to be sensitive to differences in system threat. In addition, the measure was also not sensitive to interactions between speaker testimony and participant SDO (as predicted in hypothesis 2), nor was it found to correlate significantly with either measure of trust or word count (see Table 1). Even so, EDA laterality did vary (along with trust and word count) in relation to participant RWA (as indicated by full support of hypothesis 3). How, then, might this performance of EDA laterality as a measure of affective openness to the other be interpreted?

As previously argued, left- and right-sided EDA responses are likely to be indicators of approach and withdrawal emotions, respectively. RWA is strongly and negatively related to positive emotions (Van Hiela and Kossowskab 2006); thus, the significant negative correlation that RWA demonstrated here in relation to left-sided EDA helps confirm that this measure is indeed an indicator of some valenced affective state. However, a problem seems to lie in reductively describing a variety of affective states merely in terms of their valence or orientation (i.e., as simply “positive” or “approach” vs. “negative” or “withdrawl”), as well as in conflating valence with orientation (Tellegen and Waller 2008).

Reasoning that a “positive” or “approach” emotion would be a component part of openness to the other, hypotheses 1 and 2 predicted that increasing (left-sided) EDA laterality would accompany greater levels of reported trust and word counts. However, not all “positive” emotions are the same; a host of “positive” and “negative” affective systems have been found to be neurally distinct (e.g., “play,” “care,” and “seeking” vs. “fear,” “rage,” and “panic”; Panksepp 1998). In addition, we should not expect one measure to indicate all “positive” and “negative” affective states with equal validity. For example, increased blood flow to the middle frontal gyrus is believed to indicate caring affective states in the context of maternal bonding (Wan et al. 2014). However, the lack of blood flow to this neural region is not necessarily a good indicator of fear; researchers have instead looked to amygdala activity (Olsson et al. 2007), though increased activity in the middle frontal gyrus is also implicated in fear responses (LaBar et al. 1998). As this example suggests, the physiological measurement of affective states is complex. Thus, to argue left-sided EDA is an indicator of affective openness to the other may have been too simplistic; the proposition was not supported here. However, as already stated, the data do suggest that lateralized EDA is indicating some affective state—one that might still be relevant to interracial contexts.

The present results demonstrate a connection between right-sided EDA levels (i.e., negative laterality values) and greater RWA. Thus, it may be the case that lateralized EDA, although not a good indicator of general affective openness, may still gauge a specific negative affective state—namely, fear. Indeed, researchers are beginning to coalesce around the idea of a fundamental connection between political conservatism (i.e., RWA) and feelings of personal (though not system) threat (see Lilienfeld and Latzman 2014). If this connection is borne out, we might consider right-sided EDA as a suitable affective indicator of other constructs relevant to interracial dialogue, particularly interracial anxiety (Plant and Devine 2003). As a measure of fear/anxiety, EDA laterality could then be considered conceptually orthogonal to openness to the other. With the right resources, listeners may follow a path of stress-related growth to manage their anxieties and open up to both trust and talk to outgroup others. Alternatively, they may be overwhelmed by their anxieties and shut down in response to outgroup testimony. As already noted, greater right-sided EDA levels are correlated with a variety of negative emotions, including fear (Banks et al. 2012). Thus, further investigation of this measure has the potential to broaden our study and improve our understanding of the dynamics involved in interracial interactions.

Although EDA laterality did not perform as predicted by the first two hypotheses, hypotheses 2 and 3 did successfully predict the differential impact of SDO and RWA on participant openness observed here. SDO interacted with speaker testimony to negatively influence the cognitive and behavioral measures of openness only in the context of the woman’s increased system-threatening testimony of collective racism. RWA, on the other hand, was negatively related to all three measures of openness across both speakers. Participant levels of RWA and SDO thus explained a significant amount of variance in openness to the other—and did so in distinct yet theoretically consistent manners.

As already discussed, RWA reflects an ideology used to psychologically shield oneself from a dangerous world; thus, it is unsurprising to find that both testimonies engendered less trust, fewer words, and higher right-sided EDA levels among higher-scoring participants. In Asbrock et al. (2012) words, “RWA represents a threat-driven motivation for collective security and ingroup cohesion… [thus] prejudice can be understood partially as an avoidance-oriented response to threat, predicted by RWA.” Previous research has demonstrated that this avoidance response can be invoked by the mere presence of outgroup members (e.g., Renfro et al. 2006), and these results extend this finding.

In contrast to RWA-based prejudice, reactions explained by SDO are those that enhance and maintain the social hierarchy. Outgroup members do not, by themselves, threaten the hierarchy—after all, its existence depends on the presence of socially subordinate outgroup members. Instead, SDO-based reactions are sensitive to what outgroup members say or do in relation to the hierarchy (e.g., Crawford et al. 2013). In the case of testimonies of social suffering, greater defensive reactions can be engendered by more system-threatening accounts. Thus, the present results—in which SDO scores were found to predict between-subject variation in both trust scores and word count but only in reaction to the testimony of collective racism—extend findings that SDO relates to system threat sensitivity (Quist and Resendez 2002; Crowson 2009), although not in all cases (e.g., Crawford and Pilanski 2014). Together, these results emphasize the potential interactive nature of resistance: Certain kinds of prejudice seem to engender certain kinds of resistance in certain contexts.

This study began with the aim of quantitatively investigating the origins of resistance to interracial dialogue by employing three indicators of participant openness to the other in reaction to two speakers’ testimonies of race-based suffering. Although the findings offer insight and raise additional, important questions, they should nevertheless be interpreted very provisionally due to several limitations that are worth (re)emphasizing. First, while the use ecologically valid stimulus material is undoubtedly a strength of this quantitative study, it necessitated the sacrifice of some control. Specifically, although the two testimonies used here represent different degrees of system threat, they also confound a host of variables—most notably gender. Perhaps something other than system threat (e.g., pitch, prosody, gender stereotypes) can account for the response differences observed here (e.g., individuals high in SDO typically have less favorable attitudes toward women; Bates and Heaven 2001); thus, these results must be interpreted with caution. Indeed, the impact of these potential confounds should be investigated using other study designs before any firm conclusions can be reached.

Second, although many studies of prejudice employ racially homogeneous samples (e.g., Inzlicht et al. 2012; Peck et al. 2013; Levin et al. 2012), this study used a racially heterogeneous sample comprised largely of Caucasian and Hispanic participants. This is a limitation in that such a small, mixed sample confounds potential racial/ethnic response differences between participants; system-threatening testimonies are likely to be heard differently based on one’s position in the system of racial stratification (e.g., Dover et al. 2014). Even so, the fact that significant differences were found here despite using a racially heterogeneous sample suggests that the effects of prejudices such as SDO and RWA are not limited to the dominant group, but instead operate across all racial and ethnic groups. For example, although studies have shown that the effects of SDO are often more pronounced among members of the dominant group (i.e., ideological asymmetry; Mitchell and Sidanius 1993), members of both dominant and subordinate groups who are high in SDO employ biases that favor the high-status group (Sidanius 1993; Dover et al. 2014), especially when subordinate group members perceive the system to be legitimate (Levin et al. 2002). Together, these findings suggest that researchers should not limit investigation of bias to dominant group members, but should instead employ racially heterogeneous samples to begin unpacking the ways in which bias plays out within and across various racial and ethnic groups.

Third, although reactions of the volunteer sample of students used here likely represent those of similar student bodies engaged in interracial dialogue, the current results reflect only a small sample and certainly cannot be generalized to other populations, nor to other environments of cultural contact. Even so, it is worth noting that the impact of prejudice on openness to other is likely to be greater in non-student populations as community samples have been found to score higher on both RWA and SDO measures than student samples (e.g., Lee et al. 2010).

Fourth, it must be remembered that these findings describe responses only in relation to these two audio recordings. Undoubtedly, other testimonies—and other modalities of presentation (e.g., video recordings, written transcripts)—will engender patterned responses different from those observed here. Lastly, the affective measure of openness to the other employed here (i.e., EDA laterality) did not perform as anticipated; thus, alternative measures should be investigated (e.g., respiratory sinus arrhythmia; Stellar, Cohen, Oveis, and Keltner 2015). However, the trust and word count measures served as promising cognitive and behavioral indicators of openness to the other and should continue to be employed in this capacity and assessed further.

As Fowers and Davidov note, “the richest form of dialogue is not merely an exchange of interesting information but a process in which the interlocutors actively question their own perspectives” (2006). The results of this study quantitatively confirm that testimonies of social suffering—those social utterances that have perhaps the greatest potential to interrupt master narratives and invite this deepest kind of learning—are likely to met with some degree of resistance, particularly and differentially among individuals high in RWA or SDO. Is that to say then that such testimonies should not be offered? Absolutely not. It does suggest, however, that there may be strategically appropriate times and forms for such testimony that perhaps allow it to be received with less defensive posturing (e.g., Danso et al. 2007). As one example, a recent study demonstrated that an emotionally engaging video intervention augmented empathic concern for a person offering testimony of individual racism (Cargile 2015). Thus, testimonies of social suffering might be more impactful if offered after such an intervention. Regardless, such speculation underscores that although understanding resistance to interracial dialogue is critical, there is much that remains to be learned. This study identified how limited forms of this resistance may function in relation to two testimonies of social suffering. Future research should continue to elucidate the nature, form, and function of such resistance so that we may be better able to maintain our openness to others.