Conducting experiments present the advantage of approaching the demonstration of causal relationships. On the other hand, the specific conditions created for the need of the experiments reduce our ability to generalize to real-world settings without some reservations.
In our studies, we attempted to re-create an environment akin to experience of teachers. However, despite the careful descriptions we provided for participants in our studies, we cannot claim that these experimental conditions are directly comparable to spending years in educational institutions with rather selective or educational policies or several months interacting with students. Our experiments were crafted to resemble actual tracking scenario, but in Switzerland, tracking decisions are not always entirely dependent on a teacher’s sole opinion and procedures vary substantially across cantons. We nevertheless concluded that our results remain relevant for the following reasons. First, the material for the scenario were created so as to remain as close as possible to actual settings for instance by 1) providing student information that is accessible to teachers in Switzerland (and adding a general description of the student’s classroom behavior to avoid a possible confound with social class expectations), 2) by reminding participants of what are the standard procedures and expectations in the canton of Vaud for tracking, and finally 3) by using actual excerpts from a Swiss teachers’ conference to manipulate the selective or educational function of institutions. Second, our results may not reflect final institutional tracking decisions, but we believe they are indicative of teachers’ individual decisions, which carry substantial weight in the real-life process.
The second limitation pertains to the population used for our samples. Our access to teachers or pre-service teachers’ population was limited, given the restricted access to teachers in Switzerland. We therefore varied the samples with students playing the role of teachers, pre- service and in-service teachers. We believe that the socialization of the students in the educational institution renders them clearsighted of the institutional functions, norms and practices (see Darnon et al., 2009) and we expected them to enact the practices as teachers would. Replicating these findings with teachers would certainly provide greater ecological validity, but we would expect similar results. Our theoretical approach relies on the assumption that it is the institutional norms and functions which shape the agents’ behaviors. While students are usually on the receiving end of these practices, they are nonetheless socialized in that same sociocultural context, and as such, are well placed to reproduce the observed effects that have been experimentally induced. Indeed, previous research using role-playing paradigms have also evidenced that participants adjust their attitudes to the role they were given (Covington &
Omelich, 1979; Harari & Covington, 1981; Houston & Holmes, 1975). Previous studies showed similar results with students acting as teachers and actual teachers (Simon, Ditrichs, & Grier, 1995; Rattan, Good, & Dweck, 2012). Furthermore, in our own studies, the same tracking paradigm was conducted on a sample of students and in-service teachers which yielded similar results.
4.2. Function of selection: underlying mechanisms
The objective of the first two lines of research is to integrate institutions and practices to predict teacher’s behavior. We therefore focused on uncovering a structural-level mechanism (i.e. the function of selection vs. education) to explain the discriminatory behavior. Although this approach is in line with the sociocultural perspective that has influenced our hypothesis, further investigation could examine the specific cognitions and motivations involved when teachers are aiming to select vs. to educate students. At this point, we cannot conclude on how selection orients individuals’ thinking process. While we do consider that for practical considerations, concentrating on teachers’ behavior rather than cognition is likely to be more informative for targeting solutions for discrimination at school; examining individual-level variables at play would be theoretically instructive for understanding the cognitive and motivational processes which inform the behavior.
Literature studying discrimination in educational settings have often proposed that teachers’ biased behaviors are related to stereotypes (Glock & Krolak-Schwerdt, 2014). Given that teachers’ (behavior or performance) expectations have been shown to vary as a function of students’ social class (Auwarter & Aruguete, 2008; Dunkake & Schuchart, 2015; Mizala & Martinez, 2015), it is reasonable to hypothesize that social class stereotypes underlie lower expectations of these students. Indeed, previous research on social class stereotype threat at university is also based on the assumption that low-SES students are seen as less smart and competent (Croizet & Claire, 1998). This in turn could (un)intentionally affect teachers’ behavior. Our research design does not permit us to conclude whether stereotypes are the mediating factor in the case of our studies. However, the postulate that discrimination in educational settings are driven by teachers’ stereotypes is not in itself incompatible with the fact that structural variables, such as the use of selection practices, shape individual behavior. If selection practices conjure, as we surmise, a hierarchizing mindset; then this process may indeed lead to greater focus (or motivation to look for) information signaling existing hierarchies and activate related stereotypes. It would therefore be an interesting venue for future
research to test whether the use of selection practices facilitate the situational accessibility of social class stereotypes.
Another related hypothesis, derived from the SDT literature, could be that enacting institutional selection practices more generally affect individuals’ tendency to create hierarchies. To quote Pratto (1999, p.195): “stereotypes can lead to discrimination, can stem from social structure, and reflect the social relationships and practices that constitute culture”. As selection practices at their core rely on the hierarchizing of individuals, they might be considered as hierarchy-enhancing practices (or act as situational hierarchy-enhancing environments) which increase individuals’ SDO. This hypothesis would however rely on the assumption that the practice of hierarchizing individuals would lead to a preference to hierarchize groups which, to the best of our knowledge, has not been tested before. Moreover, this hypothesis, as opposed to the previous one which postulated a temporary effect of selection, would predict that repeated exposure to selection practices would enhance individuals’ SDO over time.
It is also possible that selective practices or the selection function do not directly activate stereotypes but trigger the salience of social norms which allow for their expression. Indeed, previous research has suggested that social norms exert strong influence for the expression of individual prejudice (Crandall, Eshleman, & O’Brien, 2002). Future studies could test if it is perceived as more normative to express stereotypes in selective rather than in educational contexts. Or conversely, following the theroretical predictions of the justification-suppression model of prejudice, we could hypothesize that using educational practices conveys the perception of an institutional normative climate endorsing egalitarian social norms, which would lead to the suppression of the expression of prejudice (Crandall & Eshleman, 2003). In this case, in both conditions (selection vs. education) stereotypes would be equally accessible, but the manipulation of the institutional function or the function of the practices to educate would reduce individuals’ tendency to rely on them to guide their behavior.
In another domain, prior research examining hiring decision found that perceived objectivity induces a mindset that validates individual beliefs and facilitates expressions of prejudice. An experiment priming a sense of objectivity found that it leads to more gender- biased decisions (Uhlmann & Cohen, 2007). Similarly, research on moral credentials proposed that thinking of oneself as unbiased could also increase the expression of prejudice (Monin & Miller, 2001). Another experiment, set this time in an organization, showed that emphasizing meritocracy also led to more gender-biased pay rewards (Castilla & Bernard, 2010). Based on their results, the authors speculated that when individuals hold the impression that they are
conducting fair and objective evaluations -which could in itself act as a moral credential - they could be more inclined to trust their subjective impressions and biases, causing them to make more biased decisions. As we think that meritocratic principles underlie selection practices, we suspect that emphasizing the selection function could activate teacher’s subjective impressions of objectivity and favor unequal decisions.
Another concept closely related to selection is competition which has been linked to rigid thinking (Carnevale & Probst, 1998), lower perception of similarity with a target (Toma, Yzerbyt, & Corneille, 2010) and higher level of prejudice (Sassenberg, Moskowitz, Jacoby, & Hansen, 2007). The function of selection may also trigger similar processes that could contribute to the discriminatory behavior in evaluation by affecting perception of competition between groups. An interesting research venue to explore is the perception of intergroup dynamics in selective vs. non-selective mindset. Perception of scarcity and zero-sum thinking beliefs in previous research has been linked to stereotyped perceptual differences in black and whites faces and differential resources allocation (Krosch, Tyler, & Amodio, 2017). It might be the case that in a selective mindset, individuals perceive more competition and less resources available which could elicit motivations to perceive more differences among individuals or to allocate more resources to members of favored groups.
Finally, a possible extension of our research could focus on teachers’ identity-processes inside the institution depending on its function. We believe that teacher’s professional identity and identification may shift with institutional goals, as proposed by the literature in institutional logic (Ocasio, 1997). This suggestion would be consistent with Social Identity Theory and Self- Categorization theory, which propose that making identities salient will structure the behavior of the individual to reflect the norms and beliefs of the group. Furthermore, Haslam (2017) suggests that learning in educational institutions is inherently a mutual process of influence which relies on a commonly shared identity membership between students and teachers. Cultivating a shared identity between students and teachers has been shown to increase learning outcomes and student engagement but is also linked to teachers’ well-being (Christ, van Dick, Wagner, & Stellmacher, 2003). It may be that an institutional focus on selection vs. education could lead teachers to rather identify with either the institution or with students. Indeed, engaging in selection practices could cause teachers to identify with the institution and its hierarchizing goals and therefore see the students as an outgroup. Instead, the education function- which is focused on increasing students’ learning outcomes - might encourage teachers to see themselves as closer to students and be encouraged to individuate them instead of perceiving them in a stereotyped manner.
4.3. Measuring systemic threat
The last limitation relates to the indirect measure of threat of Study 31. & 3.2. The results were consistent with our predictions: manipulation of social disorder (in the form of high- achieving low-SES students) prompted participants to either misremember the low-SES students’ good grades as being mediocre or deliver a lower grade for the same performance. We therefore consider the results as supportive of the hypothesis that disruption of the social class hierarchy is threatening. Nevertheless, it is important to acknowledge that by indirectly measuring threat, we can only presume that threat is the underlying cause of the reaction.
Another possibility could have been to explicitly ask participants to report their feelings of threat as a manipulation check. We believe however that this operationalization of threat comprises some limitations which would render the measure unreliable. First, using explicit measures of threat relies on the assumption that participants are engaging altogether consciously with the information provided and carefully calibrating their reaction. Yet, behavior can be affected by threat without participants being aware of it (Blascovich, Mendes, Hunter, & Lickel, 2000). For these reasons, some researchers advise against using manipulation checks altogether (Fayant, Sigall, Lemonnier, Retsin, & Alexopoulos, 2017). Second, given the subject matter (i.e. hierarchies in school), we believe that, especially for pre-service teachers, responses would be subject to a strong social desirability bias. This could possibly lead to a floor effect which would render the measure unusable. Third, we think it possible that participants experiencing the threat the most be particularly motivated to deny it as a defense mechanism; either to dissipate the negative feelings engendered by threat or to rationalize their negative reactions/behavior in the study (Branscombe et al., 1999).
Finally, although our studies are the first to hypothesize that disruption of the social class hierarchy is threatening, other studies have relied on similar assumptions with regards to race relations to operationalize threat. For instance, Wilkins and Kaiser (2014) hypothesized that making racial progress salient elicits a threat for white individuals who believe the hierarchy is legitimate. Their results, which the researchers interpret as emerging from threat, show that participants are more likely to claim their group is victim of anti-white bias. Another study manipulated the perception of a majority-minority in racial relations shift in the U.S. to induce threat and observe that white participants express more conservative views (Craig & Richeson, 2014).
One aim of future research in this area should be to use other dependent variables consistent with manifestations of threat to decisively confirm its presence. For instance, perceived system-threat and societal instability have been associated with higher endorsement
of conservative policies and support for the status quo (Jost et al., 2003; Bonanno & Jost, 2006). One interesting venue could be to test the effects of social-disorder exposure on measures of support for the social system to see if it leads people to defend the system. An alternative could be to use physiological measures to infer participants’ psychological state after exposure to social disorder. Cortisol levels have been associated in prior research with negative psychological states including social evaluative stress (Dickerson & Kemeny, 2004). Measuring them could test whether participants are experiencing more stress in the social disorder condition comparatively to social order.
Finally, to additionally test if disruption of the hierarchy is threatening, a future study could attempt to mitigate the threat before measuring the dependent variable. Self-affirmation techniques have previously been used to this effect (Sherman & Cohen, 2006). To buffer the effects of threat, half of participants describe a previous behavior that exemplified their most important value. Affirming one’s self-integrity should alleviate feelings of threat and reduce the incentives to act on it afterwards comparatively to the control condition.