Although wealth is perhaps best conceptualized as an objective continuum, the wealth spectrum is often used to divide people into distinct socioeconomic groups. These groupings can have significant effects on how a person behaves (Johnson, Richeson & Finkel, 2011; Stephens, Markus & Fryberg, 2012; Sinha & Mishra, 2015), and how a person is responded to across a variety of situations (Gilmore & Harris, 2008; John-Henderson et al. 2013; Haider et al, 2011). Furthermore, socioeconomic status can have dramatic effects on important life outcomes, including health outcomes and involvement in the criminal justice system (Freeman, 2006; Lott, 2002; Adler & Newman, 2002).
Within an educational context, children with a low socioeconomic background show severely reduced academic attainment. Indeed, it has been convincingly argued that socioeconomic status is the most important factor in predicting academic success or failure (Strand, 2014; Kingdon & Cassen, 2007). For example, longitudinal studies find that such pupils underperform in literacy skills compared to their middle class peers (Hartas, 2011). Similar weaknesses are found across low SES pupils in relations to mathematical skills (Ritchie & Bates, 2013). Additionally, the dropout rate of low SES pupils from university in the UK is considerably higher than average (Chowdry et al. 2013; Johnes & McNabb, 2004). These poor educational outcomes largely result from factors beyond the school environment such as lower parental involvement or reduced access to resources that promote academic achievement.
However, it has also been suggested by numerous studies that a major problem faced by low SES pupils is lowered teacher expectations (Speybroeck et al., 2012; Hinnant et al., 2009; Benner & Mistry, 2007; Glock & Krolak-Schwerd, 2014; Diamond, 2004; McCombs & Gay, 2007). These expectations are linked to attitudinal biases (e.g. Auwarter, 2008; De Boer et al 2010). In particular, studies have emphasised the Pygmalion effect (Merton, 1948; Rosenthal & Jacobson, 1968). This refers to “the effects of interpersonal expectancies, that is, the finding that what one person expects of another can come to serve as a self-fulfilling prophecy” (Rosenthal, 2010, p. 1398). Various studies support the view that lowered teacher expectations have many disadvantageous implications for students (for meta-analyses and reviews see Jussim, Harber, 2005, Rosenthal, Rubin, 1978 and Tenenbaum, Ruck, 2007). For example, teachers typically spend less time with and ask fewer questions of students perceived as having less potential (Friedrich et al., 2015). Teachers have also been found to display more negative emotions in relation to low expectation children and
Indeed, lowered teacher expectations have also been shown to correlate with lower pupil motivation (Nugent, 2009) and self-esteem (Alvidrez & Weinstein, 1999; Rubie-Davies, 2006). These and other negative outcomes related to teacher expectations illustrate the importance of teacher attitudinal biases in student outcomes. Furthermore, studies suggest that low SES pupils are among the most vulnerable to the Pygmalion effect (For a recent review, see Li, 2016). Additionally, individual teacher biases have been shown to mediate Pygmalion effects (Babad, 2009; McKown & Weinstein, 2008). More biased teachers have been shown to be more likely to generate negative Pygmalion effects (Babad, 2009), and that these teachers tend to hold more stable and rigid expectations for students (Kuklinski & Weinstein, 2000 & Weinstein, 2002). Furthermore, the psychologically complex environment of the classroom suggests many channels by which teacher bias effects may be exponentially increased. For instance, studies have found that pupil perception of teacher bias mediates teacher expectation effects (Brattesani et al., 1984 and Kuklinski and Weinstein, 2001). This corroborates the need for further study of teacher attitudinal biases.
Biases can be divided into two categories, those that are measured explicitly, and those that are measured implicitly. (For a review, see Fazio & Olson, 2003). In attitude research, self-report measures are dominant. These measures typically tap explicit attitude constructs since such attitudes are consciously expressed and controlled (Gawronski & Bodenhausen, 2006; Nosek et al. 2007). Explicit attitudes often form in response to new information and are known to change over time (DeCoster et al, 2006). In contrast, implicit attitudes are automatically activated responses, often without conscious awareness (Nosek et al 2007). Implicit attitude constructs form early in life and their effects often persist into adulthood, despite the presence of divergent explicit attitudes (Rudman, Phelan, & Heppen, 2007). These measures have been suggested to be more effective than explicit measures in predicting non-verbal behaviors such as eye contact and smiling (Dovidio, Kawakami and Gaertner, 2002). Whereas explicitly measured attitudes are thought to predict deliberative actions, such as the content of conversation (Dovidio et al, 2002). As such, both these constructs hold high relevance to teacher-pupil interactions.
The most widely used measure of implicit attitudes is the Implicit Association Test (IAT). This type of measure employs a response latency paradigm (Fazio & Olson, 2003). The IAT tests the association between various concepts by asking participants to repeatedly pair two concepts (e.g., poor and good, rich and good). The more strongly the participant associates two concepts, the faster the participant will complete the pairing. Meta-analyses of the IAT have suggested it to be a highly reliable measure of biases (Greenwald, Banaji & Nosek, 2015). This measure has been shown to be largely free from limitations imposed by social desirability effects (Steffens, 2004) or lack of introspective awareness (Hofman, 2005). The ecological validity of the IAT has been demonstrated for various domains of human functioning, and is capable of predicting behaviours such as (Greenwald et al. 2015)
However, the relevance of implicit attitudes to discriminative behavior has been challenged by others (Oswald, Mitchell, Blanton, Jaccard, & Tetlock, 2015; Oswald, Mitchell, Blanton, Jaccard & Tetlock, 2013). These authors argue that such measures predict only small variance in discriminatory behavior. Such findings have been contested on numerous methodological grounds (Greenwald et al. 2015). In light of the present contentiousness, a call for greater conceptual and methodological vigor relating to implicit attitude studies has emerged (e.g. Blanton & Jaccard, 2015). In resolving such issues, the importance of further, high quality studies of implicit attitudes seems paramount.
Within a teaching population, implicit measures have been used in relation to a number of socially sensitive issues. In studies of implicit racial biases, teachers typically show preference to racial majority students over minority ones (Glock et al. 2013; Van der bergh et al, 2010). Similarly, studies of implicit teacher attitudes relating to children with special educational needs suggest that teachers show negative bias against pupils with special needs (Hornstra et al. 2010; Levins et al. 2005).
To the author’s knowledge, the present study is the first exploring implicit attitudes towards SES within a teaching population. As such, the study represents a method of consolidating understanding of implicit attitudes within an educational context, as well as having the benefit of contributing to discussions regarding implicit measures in general. Furthermore, some researchers have called for further studies of implicit biases in education, in order to aid teachers in overcoming these biases (Glock & Kovacs, 2013). The present research can be seen as a partial response to this call. In accordance with previous research, the following hypotheses were tested.
H1. Participants will show implicit preference towards higher Socioeconomic status.
H2. Implicit SES biases will be stronger than explicit SES biases.
These hypotheses are one-tailed in accordance with much of the previous research undergone relating to SES bias and the IAT (e.g. Horwitz & Dovidio, 2015; John-Henderson et al. 2013; Haider et al. 2011).