Choices in education by their very nature are dependent upon social interactions. These social interactions are complex and cognitively demanding due the number of variables involved, and problems of incomplete and asymmetric information. Consequently, ‘‘the ability to sort people (or objects) spontaneously and with minimum effort and awareness into meaningful categories is a universal facet of human perception essential for efficient functioning’’ (Bodenhausen, Todd and Becker 2006). A person’s identity defines who they are with regards to their social category, the ‘in-group’ (Akerlof and Kranton 2010). Having a common ‘identity’ in social interactions significantly reduces the amount of information asymmetry present with regards to individuals within the group, thereby decreasing the complexity of decision making.
The same heuristics that are valuable in reducing complexity and cognitive load can also lead to bias-confirming assessments of inter-group relations giving rise to stereotyping. The perception of an individual’s identity status via social cues can reinforce confirmation biases associated with maintaining a state of identity threat (Darley & Gross, 1983). Identity threat is one of the mechanisms that lie behind persistent achievement gaps in education outcomes (females: Spencer et al., (1999); African-Americans: Steele and Aronson (1995); students from low socio-economic backgrounds: Croizet and Claire (1998)). However, being a socially context dependent behaviour, identity is localised and does not persist beyond its context frame. For example, low achieving boys when changing grades experience large gains when leaving behind old identity norms and expectations (Dweck et al., 1978). For an explanation of the decision processes that underlie poor academic achievement due to identity threat (see Cohen & Garcia, 2008).
One of the clearest examples of the critical nature of context framing and the malleability of academic performance due to social identity is an experiment by Shih et al. (1999). In their study a group of Asian-American women were randomly split into two groups where either the individual’s gender or their ethnicity was made salient using semantic conditioning. Results were compared with a separate, randomly composed control group without any semantic conditioning. For the gender salient group individuals were asked to indicate their gender and answer gender related questions but excluding any reference to ethnicity. Questions for the ethnicity salient group were constructed in a similar manner while the control group answered questions without reference to either gender or ethnicity. Individuals in all groups then completed the same mathematics test. The researchers found that simply switching identity salience produced diametrically opposite levels of performance in the test. When identity was aligned with Asian ethnicity individuals achieved a higher level of accuracy than the control group (54% versus 49%). However, when identity was aligned with female gender individuals performed worse than the control group for exactly the same test (43% versus 49%). The important implication of this study is that individuals maintain multiple identities which can be triggered by social context leading to divergent performance in an academic environment.
Social identity has also been shown to affect the willingness of individuals to compete and thereby participate in educational choices. The gender gap in mathematics has been shown by Gneezy et al. (2003), and more recently Niederle and Vesterlund (2010), to be influenced by a screening effect where girls self-select out of mathematics subjects due the perceived competitiveness of the environment. A similar gender gap has been shown for competitive entrance exams in university choice (Jurajda & Munich, 2011; Pekkarinen, 2014). In studies of girls attending co-educational and single-sex schools, the social context in which students make choices has been shown to change their risk preferences (Booth, Cardona-Sosa & Nolen, 2014). However, negative consequences of identity on academic performance can be remedied by either reducing the salience of a particular identity threat (Cohen & Garcia, 2005) or replacing conflicting identities with a new shared identity (West et al., 2009).