The last few years have seen an increasing interest in the use of so called ‘implicit measures’ to assess consumer opinion of products, packs and marketing materials. So what are Implicit Measures and how do they differ from our more traditional, explicit, or self-report, methods?
When we use explicit, or ‘self-report’ methods, we ask the consumer what they think of something. For example, “how much do you like this product?” They give us their answer after they have thought about it and rationalised their opinion. They may tell us the truth, or they may tell us what they think we want to hear, or perhaps they find the question too hard to answer and so they just make something up. This slow, logical, rationalised thinking is known as system 2 thinking (Kahneman 2011)1.
When we use implicit methods, we do not ask the consumer what they think of something. Instead, we ask them to do a task, which taps into their fast, instinctive, or system 1, thinking. System 1 thinking is based on pre-existing associations in the mind and forms instant reactions before the response is rationalised by system 2.
As such, system 1 responses are less likely to be tinged by bias based on public opinion, the latest knowledge, and the myriad of other factors that can modify our preferences and purchase decisions. It is argued therefore that implicit measures give a more reliable and accurate picture of the real influences on a person’s behaviour than more traditional explicit, questionnaire based approaches. Of particular interest is the apparent ability of implicit measures to explain behaviours or opinions that explicit testing fails to do. In a review, Claudia Dimofte2, cites a number of examples including that of evaluation of an advertising slogan for the ski resort, Aspen. Explicit responses to the slogan ‘Going down fast in Aspen’, focused on the speed of the slopes and the positive associations with skiing. However, implicit testing revealed a negative side to the slogan: the quality of the resort was deteriorating!
The most well-known method is the Implicit Association Test (IAT)3. The IAT measures the speed at which words are associated with one of two pairs of concepts. The response time is expected to be shorter when respondents are asked to assign words to strongly associated pairings and longer when they are asked to assign words to incongruent pairings.
Responses are collected by hitting target letter or number keys on a keyboard. The respondent practises first to become familiar with what keys to hit to register a given response.
To compare the opinion of the taste of Coke and Pepsi, there could be four categories: Coke; Pepsi; Pleasant; Unpleasant. Categories are associated with keys on a computer and respondents practise to learn these key-category associations. For example, respondents might be trained to hit the Z key when a Coke image is displayed and M when a Pepsi image appears. They might then be trained to press Z for Pleasant words (e.g. thirst quenching, refreshing) and M for Unpleasant words (e.g. sickly, bitter). In the next stage, the two are then combined so that the Z key would be pressed for Coke and pleasant words and the M key for Pepsi and unpleasant words.
Examples of both drinks plus pleasant and unpleasant words would appear on the screen and the respondent must categorise all of them. The theory is that respondents will categorise words faster, if the pairing is perceived as ‘congruent’. So if the consumer thinks Coke is more pleasant, then both Coke photos, and ‘pleasant’ words, will be categorised faster, since both words are going to the same side. However if the consumer thinks Pepsi is more pleasant, then ‘pleasant’ words would be categorised more slowly since they are paired with Coke.
Respondents repeat with two of the categories having swapped sides. In our example, unpleasant would now be on the Z key with Coke and pleasant on the M key with Pepsi.
The IAT score is the difference in response times between the two combined tasks. So here a consumer who thinks Coke is more pleasant would be expected to assign pleasant words more quickly when paired with Coke on the Z key than when the pleasant words were paired with Pepsi on the M key.
Whilst there have been a few studies that look at consumer response to food brands using these techniques, none to date have measured implicit attitudes to the sensory properties of foods although some have run sensory tests in parallel to implicit tests.
We are undertaking a research project to evaluate how implicit measures compare to explicit reports of Preference and Purchase Intent for impulse and non-impulse foods. The results of a study on chocolate cup-cakes were presented at the SSG Conference ‘A Slice of Sensory’.
Results showed good correlation with implicit methods for preference and purchase intent but indicated differences between samples on the basis of emotional reaction.
1.Thinking Fast and Slow. Daniel Kahneman, Farrar, Straus and Giroux, 2011
2.Implicit Measures of Consumer Cognition : a review. Claudia V. Dimofte, Psychology & Marketing, Vol 27 (10): 921 – 937 (October 2010)
3.Measuring Individual Differences in Implicit Cognition: the Implicit Association Test. A,G Greenwald, D.E McGhee and J.L.K. Schwartz. Journal of Personality and Social Psychology, Vol 74 1464 – 1480 (1998).