Which is better the IAT or Affective Priming?
Comparison of two Implicit Association Tests
Affective Priming versus Implicit Association test
One criticism of the IAT is that it may merely tap ‘extrapersonal associations’ – it may be a measure of culturally shared assumptions rather than personal attitudes. For example, this would argue that an IAT that detects my strong association between nurse and female is just reflecting my knowledge that society has historically given the role mainly to women, rather than this being my own personal automated attitude (i.e., that nurses ought to be female).
Another criticism of the IAT is the reliance on the switching of blocks. In the first phase, (and in a hypothetical gender bias test), the word female is paired with gender stereotype attributes, e.g., nurse, and the word male with doctor. After the respondent has learned to do this quickly, the categories are then reversed, so that the word female is now paired with words incongruent with the gender stereotype, which make the test suddenly more difficult. This yields significant reaction time differences in the second block – it is a harder task than the first block (not because of an inherent gender bias attitude, but because the respondent had already learned the correct responses, but in the second block had to unlearn them and then re-learn the new responses). This is worrying because it means that the effect is too prone to changes in procedural issues.
A further problem is that in the IAT, only two dichotomous concepts can be paired (e.g., men vs women, gender stereotypical vs not gender stereotypical), which can be very limiting when one wishes to explore their relationship in more detail. Consequently the IAT produces a single global gender bias score. However, in affective priming one may have more than just a global score and can divide attributes into dimensions and hence provide a more detailed picture of such a relationship. So for example, a gender bias test using affective priming will be based on a large number of ‘attributes’ and these can be categorised (e.g., roles, personal qualities, professions, and so on) and this kind of test produces a score for each dimension. Another statistical advantage of the affective priming approach is that one can conduct a factor analysis on the data to reveal how attributes are grouped (grouped in the minds of the respondents who took the test). Hence it can yield groups of attributes that together are likely to represent an important feature of the concept begin measured (e.g., nurse, carer, ethical, reliable, hardworking, gentle, and female) – of course this example is too obvious and not so informative, but some patterns can emerge from this approach that weren’t predicted. This is much harder to do with the IAT.
Finally, the reasons why affective priming works is because it is based on assumptions that are highly compatible with what is known about how the brain processes information. Neural network models of the brain are based on mental associations – the stronger the association between two concepts (e.g., female and nurse) the quicker one concept will mentally trigger the other. So that’s four reasons why affective priming is the preferred approach, particularly if you are looking to understand the complex processes in the minds of consumers.