If you asked me how many hours I exercise in a week, I would probably give an answer that would make me feel and appear good (even if the actual number was lower). For a new local gym trying to assess how many visitors it can expect in an average week, surveying people such as myself, may lead it to make gross miscalculations.
In Say what you mean, mean what you say we briefly introduced the Implicit Association Test, developed by psychologists from Harvard, University of Virginia, and the University of Washington.
The IAT was developed as a technique to explore unconscious roots of thinking and feeling1. Many of the initial applications of the test involved understanding social prejudices in the community around race, gender, and sexual orientation, among others.
Over the years, a number of organisations have adapted the IAT to understand behaviours which may be limited by self-reporting—areas like diversity training in human resources2, brand and advertising attitude measurement, alcohol and tobacco use and even political preferences of voters in elections3.
IAT in brand and marketing
Today, the marketplace has become exceedingly cluttered with a wide variety of brands making big promises vying for a share of the consumer’s wallet. In some categories, there is very little differentiation among brands on functional attributes; they all do the job well. So, brands emphasize on the emotional appeal and the brand’s values with their audience. The challenge lies in communicating this effectively through various consumer touchpoints and assessing whether consumer take-aways are in line with what is intended.
Sounds familiar and easy? A marketer may turn to conduct some focus groups or a quantitative research assessing brand-attribute associations pre and post a campaign. Explicitly stated answers coming out of such research has its pros and cons. While they have a high degree of validity and statistical confidence, they may not be the most suitable when you want to understand the strength of association. For e.g., Cola brand 1 vs. Cola brand 2 – both are seen to be great tasting, but one gives more joy than the other. Explicit research may not be able to disassociate the brands well enough here. Incorporating IAT may enable the marketer to forego some of biases that can creep in with self-reporting and provide a degree of differentiation between the brands on more emotional aspects.
Several research firms have successfully experimented with IAT in predicting voter behaviour pre and post elections. Traditionally run voter polls are often unable to predict the outcome on election day with accuracy; the reasons for this are many—some refuse to participate in polls, and some fail to express a clear preference during the poll or are undecided.
Researchers have used IAT to assess political attitudes among reticent and undecided voters. These indirect methods may be more reliable than directly asking this segment of the population who they would vote for4.
It is very important to note that the IAT is not built to predict an individual’s behaviour based on a single test—analysis is always done to show how aggregates or groups respond5.
How is an IAT conducted?
The IAT measures the strength of association between concepts or brands (e.g. Single vs. Married, Coke vs. Pepsi), and the attribute dimensions or evaluations (e.g., happy, sad or modern, traditional). Respondents are primed with the concepts and the dimensions and then asked to make associations on the screen as quickly as possible. The speed of associations is assessed among different combinations shown and this helps determine an individual’s implicit attitudes6.
To conduct an implicit association test, respondents need to complete five blocks of questions7.
Block 1 Respondents sort images of different concepts into polar categories (e.g. Good vs. Bad or pleasant vs. unpleasant). The images of the concepts appear in the centre of the screen and the categories appear to the left and right of the screen.
Block 2 A set of attributes are chosen to describe each of the categories used in block 1 (such as ‘inspiring’, ‘for me’, ‘disgusting’, ‘boring’ etc) and respondents are asked to associate these attributes with the same two categories as quickly as they can.
Block 3 This is a combined task which involves sorting the concepts and the attributes which appear alternately on the screen into one of the two categories above (i.e., good or bad) as quickly as possible. Respondents will typically perform an initial block of 20 trials as practice and then repeat this for 40 trials which is termed the critical block.
Block 4 The same concepts used in block 1 are shown to the respondents; here the main distinction is that the key to assign them to their respective categories is reversed.
Block 5 A last combined task is administered which asks respondents to sort the concepts and attributes to one of the two categories. The key difference from block 3 is that the keys used to assign them to these categories is reversed.
After completing the five blocks of enquiry, the IAT score is calculated which is a function of the difference in average response speed between block 3 and block 5.
Market research is not a perfect science, and good researchers agree that tools like IAT do not replace existing methods. However, combining implicit methods wherever possible with traditional techniques can help to add another layer of insights and minimize the risks of making wrong decisions.