Predicting food purchasing behaviour: view on different methods

Dutch and Finnish researchers have been studying the different tools used to predict consumer choice of food in the supermarket. While scores on traditional instruments such as surveys are thought to be the best predictors, a recently published study co-authored by a scientist at NIZO suggests that neurological measurements may also be useful for predicting purchasing frequency.

Understanding consumer preferences

The marketing efforts of food manufacturers are guided by the specific needs and preferences of potential customers. To improve the chances of their products being successful, food professionals therefore employ consumer-oriented approaches to improve the match with consumer needs. Understanding the factors that influence consumers’ food purchasing behaviour is a science in itself. This field makes use of a wide range of surveys and other instruments that measure consumer responses to products in order to predict purchasing behaviour.

Measuring conscious drivers

To determine consumer preference and predict purchasing behaviour, firms frequently test the market potential of new foods in consumer studies, also known as product evaluations. These evaluations often make use of specific scores that include the test subjects’ degree of ‘liking’ or ‘pleasure’ in response to the packaging and labelling, as well as to the look, feel and taste of the product itself. But studies have shown that purchasing behaviour is influenced by both conscious and subconscious drivers. It is therefore thought that merely asking people to give their opinion about a product does not always reflect their true subconscious emotions towards that product. If this is indeed the case, verbal responses on questionnaires may not always accurately predict the person’s likelihood of actually buying that product. Improving the accuracy of these predictions may therefore require other methods that can also measure such subconscious factors.

Measuring subconscious drivers

For an instrument to measure a subconscious response, the test subject must be unaware of the behaviour that is being measured. One way of doing this is to make use of a method that measures subconscious approach and avoidance tendencies towards visual stimuli. This so-called approach-avoidance task (AAT) has been applied in different areas of psychology to measure involuntary responses to a wide variety of stimuli such as spiders and alcohol, for example. It has also been proven useful in measuring responses to wanted and unwanted foods: the food images on a screen cause an involuntary response while the test subject performs a behavioural task. While subconscious responses of test subjects can also be measured accurately using electroencephalography (EEG), collecting data using EEG is more complex than using AAT and therefore less practical for use during product evaluations.

Testing the instruments used in product evaluations

So can food choice be predicted by tapping into the brain processes that control choice behaviour? Are food professionals using the right kinds of tests and surveys during product evaluations? To answer both of these questions, the authors of the study in Food Quality and Preference determined how consumer purchases of a choice of three different products were related to their responses to these products. They measured the responses using both a traditional survey and neurological behavioural tests. The researchers compared these measures to determine the best way of predicting which product – in this case three different blueberry-flavoured quarks – the subjects were likely to buy most frequently in the month following the evaluation. Both before and after a product tasting session, the 136 subjects filled in an online survey that also collected pictorial emoji scores. After the tasting session, neurological responses were collected from a subset of 50 subjects using EEG, AAT, and pupil size responses, all while subjects were shown images of the products. Afterwards, all subjects were asked to keep track of their daily quark purchases for one month.

Subconscious responses also influence purchase frequency

Overall, the study demonstrated that the most accurate measures for predicting purchasing behaviour were the more traditional scores derived from the survey – both the verbal scores and the pictorial emoji scores. In terms of the neurological behavioural responses, while correlations with purchasing behaviour were higher for EEG responses than for AAT responses, in first instance the data indicated that these scores were not very good predictors. However, when the researchers further analysed the data to look for relationships between purchasing behaviour and combined scores from the different instruments, multiple linear regression modelling identified the AAT response as a key factor: it predicted aspects of the test subjects’ purchasing behaviour that could not be explained by other predictors.

Different products may require different instruments

When discussing their results, the researchers speculate that AAT measurement was likely not a predictor in the less complex analyses because the three products were rather similar in terms of availability, composition, familiarity, and health-promoting claims. This meant there was likely little influence of social desirability when subjects filled in the surveys. In other words, there was no apparent factor that could have biased subjects towards withholding their true impressions in favour of more socially desirable alternatives. Such impressions are particularly suited to detection with instruments that measure subconscious neurological responses. The researchers suggest, therefore, that their results need further study using products more susceptible to social desirability, or products that are totally novel and unfamiliar to the respondents. Nevertheless, the ease of use of the AAT instrument certainly makes it a promising addition to the current range of methods available to food professionals looking to measure subconscious drivers of consumer behaviour.