OPINION8 June 2015

Longitudinal logic


As part of his series on rethinking market research, Colin Strong calls for a greater focus on longitudinal research as part of a bigger discussion on the need to update our consumer model.


We like to think that as humans we have a consistent outlook on the world. If you ask us our attitudes on an issue today then we think these are not likely to be very different to the attitudes we held last week, last month or even last year. Yes, the argument goes, these may shift around a little but fundamentally they remain stable. The idea that we are consistent is important to us as human beings.

This, however seems to be at odds with much of the psychology literature which suggests that we are highly malleable.

When, for example, our estimates of an item’s value can be influenced by the recollection of the last two digits of our credit card number, it’s hard to say that our attitudes are not easily swayed. The behavioural economics literature has demonstrated pretty well that the context in which we find ourselves is very important in shaping the way we see the world. 

And in Byron Sharp’s book, How Brands Grow he provides some empirical data about the way in which attitudes to brands change – showing that we are much more open to switching our brand loyalty than many of us had thought.

So how does this stack up with survey tracking data that shows very little difference in attitudes over time?  Attitudes are typically represented by long flat lines which, outside of calamitous events for a brand, only move up or down very slowly. As market researchers we are therefore fond of saying that our attitudes are very stable, will take a long time to shift and to do so requires a lot of investment in advertising, customer service etc. 

Perhaps the answer lies in the tools we have available. Typically in market research we undertake surveys using independent samples for each wave. The samples are carefully monitored to ensure they are consistent so that any changes between waves are not due to the design of the study – we are all aware of the dangers of how changes to the sample design can influence the data. But, in effect, we are taking a series of ‘snap shots’ which, when put together, will show how attitudes are changing over time. 

However, in the process of doing this, are we not falling into the trap of what psychologists call the fundamental attribution error? This is when what we see in someone else at a particular point in time, however fleeting it is in reality for that person, is considered by the observer to be a consistent and stable trait.

So I may experience a person being cold and rude to me and assume that is just how they are. But maybe the vast majority of the time they are friendly and warm, it’s just that the context led to that behaviour on that occasion. Market researchers use a series of waves each with an independent sample, so are we perhaps falling into exactly that trap? 

Of course, at a macro-level attitudes may well be remarkably stable.  So while individually our attitudes may be moving around, when we look at the aggregate level we simply do not see this.  When we were in an era of mass marketing this worked well enough. It didn’t matter whether any one individual happened to be less disposed to a brand at a particular point because it was compensated by another individual that had become more disposed. So as a group we are a broadly stable lot. 

However, as brands are increasingly using personalised marketing, there is a challenge. If our attitudes and predispositions are moving about, then personalised marketing is going to be much more difficult to pull off effectively.

So we need a better understanding of the way we move about as individuals. Understanding how context shapes us will become a more important part of market researcher.  Which means that longitudinal research will increasingly shed its Cinderella status.

Once the preserve of social research studies, we are starting to see its re-emergence as technology changes the game; Community Panels and in-the-moment tracking are quickly gaining momentum.  The longitudinal element of these is often used to record, and then estimate, the impact of different influences in consumer decision making. This will increasingly be supplemented by measuring how individual consumer attitudes move about as a function of their environment.

This is all inevitably part of a bigger theme where we rethink our model of the consumer (or indeed the ‘self’). Our model of the consumer needs updating: we are much more fluid, connected and influenced by context than previously imagined.  And it is not just market research that needs to be reconsidered but the wider marketing and big data agendas.

Colin Strong a board member of Verve.