[Tirta Susilo is a PhD student in psychology, and a co-author of mine on a recent study, published in (appropriately enough) the Journal of Economic Psychology. Tirta has written a guest-post on some fascinating new research about skin colour and politics.]
Earlier this year Andrew Leigh and I observed that skin colour can predict vote share: in a Northern Territory election, darker-skinned candidates won more votes in predominantly darker-skinned electorates, while lighter-skinned candidates fared better in predominantly lighter-skinned electorates. Our finding argues that, in the absence of complete political information, voters might use skin colour to help them cast the ballot.
But how do voters actually perceive skin colour? Is voters’ perception of skin colour veridical? Or do political leaning biased voters’ perception of skin colour in a systematic way? This question was recently examined in a study by psychologists Eugene Caruso, Nicole Mead, and Emily Balcetis.
Two lines of research form the backdrop of the study. First is research showing how people implicitly associate white with good, and black with bad. Second is research showing how people’s social and visual judgement are often biased according to the object’s perceived group membership. For example, a previous study by psychologists Daniel Levin and Mahzarin Banaji revealed how people could perceive a racially ambiguous face as lighter or darker depending on their expectations of the group to which the face belongs.
Caruso and colleagues predicted that since political partisanship is a form of group membership, voters would associate preferred candidates with lighter skin colour, and disliked candidates with darker skinned colour. They also expected that skin colour perception would correlate with intention to vote.
To test their hypotheses, the researchers collected photographs of a hypothetical biracial candidate and Barack Obama, both in several different poses. (Subjects – mostly White students – were previously led to believe that this candidate did or did not share their political beliefs.) Three versions of each pose were created: the original one, a lightened one, and a darkened one. Subjects were given a random pick of three different poses (one for each version) and were asked to rate how representative of the candidate each photograph was. Each subject saw only one version of each pose, to ensure representativeness rating would not be confounded with photo characteristics.
As predicted, subjects were more likely to rate the lightened version as most representative when they thought the hypothetical candidate agreed with their political views, whereas those who believed the opposite were more likely to pick the darkened version as most representative. Similarly with Obama, those who identified themselves as liberal rated the lightened version of him as most representative, whereas those who identified themselves as conservative chose his darkened version as most representative.
Since it is known that some media deliberately attempted to shape public perceptions of Obama, one might wonder whether perceived skin colour has something to do with subjects’ media consumption. To check this possibility, Caruso and colleagues measured subjects’ consumption of media in general, as well as liberal (e.g. New York Times) and conservative (e.g. Fox News) media in particular. Controlling for political orientation and general media consumption, they failed to find any correlation between light-advantage score for Obama and liberal or conservative media diet, which rules out the media-bias explanation.
Caruso and colleagues then computed a light-advantage score (lightened version rating minus darkened version rating, over original version rating). This score was found to predict intention to vote for the hypothetical candidate; it also predicted both intention to and reported vote for Obama (this research was done pre- and post- 2008 US presidential election). Crucially, in the Obama case, the correlation held even when political orientation, implicit prejudice, and explicit prejudice were controlled. This allowed the researchers to conclude that the degree to which subjects perceive a photograph as most representative is systematically related to their stated and reported voting behaviour.
It remains to be shown in what direction the causal arrow points (e.g. does political leaning alter perception of skin colour, or does skin colour perception sway political leaning?). In any case, Caruso and colleagues have shown that the link between skin colour and voting behaviour is two-way: as skin colour could affect how voters cast their ballot, political partisanship could also affect how skin colour is perceived.
(xposted @ Core Economics)