Why Consumer Preferences Don’t Predict Consumer Behavior
Obsessive tracking of product and brand preferences may be a fool’s errand
People’s automatic responses to repetitive exposure and processing fluency may create what appears to marketers to be stable and reliable preferences.
But brain science researchers have found that human preferences are often much more temporary and much more easily manipulated than consumers’ sincere declarations of brand love might lead marketers to believe.
This is an area where traditional marketing theory has been led astray by the classic economic model of rational choice, which predicts behavior based on assumptions that preferences are stable, consistent, known before choices are made, and known with adequate precision to make the process of choosing among alternatives unambiguous.
In contrast to this approach, much of the behavioral economics research has found those assumptions of economic rational choice to be unrealistic as descriptions of how individuals acquire and use preferences in everyday life.
Although it’s true that the rational choice model often works well in predicting the overall behavior of large aggregates of individuals in economic systems, it can provide a very misleading picture of the role played by preferences in individual consumer choice and behavior.¹
And it’s that role that both marketers and consumers care about the most.
Consumer Preferences Are Fragile
When scientists study how preferences actually get incorporated into consumer decision making (and human decision making in general), they consistently find empirical results that violate the assumptions of economic rational choice.
Marketers often fail to recognize these violations, and they tend to overemphasize the importance of managing preferences as a goal of marketing.
Brain science has demonstrated that preferences play a quite different role in consumer decision making than is assumed by traditional persuasive marketing models.
Here are four important findings that explain why consumer preferences aren’t reliable predictors of consumer behavior.²
Preferences Are Constructed
In 2006, social psychologists Sarah Lichtenstein and Paul Slovic published a book titled “The Construction of Preference,” collecting in one volume over 35 years of research on this topic.
The book begins with Lichtenstein and Slovic’s examination of the “preference reversal” phenomenon — the finding that people reverse their preferences when they compare alternatives in different ways, such as jointly vs. sequentially³.
The book also documents a wide variety of contexts in which preferences aren’t independent inputs into choice deliberations, but are rather calculated in real-time as part of the choice process itself, often in ways that are highly dependent on how, where, and when the choice is encountered.
To the extent that consumers have unstable goals, operate under cognitive constraints, or make choices without prior experience or expert knowledge — three common conditions during the shopping stage of the consumer cycle — their preferences are more likely to be constructed and less likely to be reliable indicators of their future choices, preferences, and behavior.
To say preferences are constructed isn’t to deny that human beings have predispositions that are relatively stable over time. There has been some debate among researchers about the relative strength of these inherent preferences versus constructed preferences,⁴ but for the most part, this debate bypasses the point of primary importance for marketers.
The relevance of constructed preferences to consumer choice isn’t dependent on the existence or nonexistence of inherent preferences.
It merely highlights the fact that when consumers cite preferences as causal sources of their decisions, those preferences may actually be constructed as a byproduct of the decision process, even if consumers think otherwise.
Some powerful examples of constructed preferences and preference reversal are found in the “choice blindness” studies of Lars Hall and Petter Johansson.
In a series of experiments focusing on faces, food items, personality traits, political and moral attitudes, and other stimuli, these researchers asked people to identify their preference between two items represented as pictures, physical items, or survey questions.
They then used various tricks to surreptitiously “switch in” the non-preferred item for the just-identified preferred item. Presenting their subjects with the previously non-preferred item, they asked them to explain why they preferred it.
The findings have been consistent across multiple choice categories and presentation methods. In a great majority of cases, people don’t even notice the switch and are quite willing to provide extensive rationales for why they preferred the item they actually didn’t prefer.⁵
Not only are their preferences constructed in real-time, but those “faked” preferences tend to have staying power. When the choice task is repeated without the sleight-of-hand, people are more likely to prefer the item they had falsely been led to believe they preferred in the first task.⁶
The act of describing a non-preferred item as preferred actually reversed the original preference order, at least for a short period of time.
Preferences Are Often Consequences, Not Sources, of Behavior
The counterintuitive implication of constructed preferences is that consumers don’t access pre-existing preferences from memory to solve choice problems — they actually discover what their preferences are by solving choice problems.
This process is highly contextual, and those discovered preferences may emerge differently in a later choice situation presented in a different context.
From the perspective of the rational choice model, this depiction of human choice appears quite irrational. But there’s a different kind of rationality at work here, originally described by James March in a classic 1978 paper, that might be called posterior rationality:
“Ideas of posterior rationality emphasize the discovery of intentions as an interpretation of action rather than as a prior position …. Actions are seen as being exogenous and as producing experiences that are organized into an evaluation after the fact.
The valuation is in terms of preferences generated by the action and its consequences, and choices are justified by virtue of their posterior consistency with goals that have themselves been developed through a critical interpretation of the choice.
Posterior rationality models maintain the idea that action should be consistent with preferences, but they conceive action as being antecedent to goals.”⁷
Consumer choice isn’t normally about applying pre-existing preferences to choice situations. It’s more often about literally creating preferences as a consequence of choosing.⁸
Preferences Are Transient and Elastic
Constructed preferences tend to have short half-lives. They change over time as contexts and goals change.
In two recent studies by psychologist Dan Simon and colleagues, participants were asked to rate their preferences regarding four features that might be associated with a job offer: office space, salary, vacation package, and commute time.
After rating several options for each feature separately, they were shown two hypothetical job offers with different combinations of the four features.
After reviewing the job offers, they were again asked to rate the features. In this new context, they “reconstructed” their preferences for the individual features in a manner that supported the job offer they rated as more attractive overall.
The researchers then waited for various periods of time — 15 minutes, one week, eight weeks — and asked the participants to rate the four features again. After each interval, their preferences returned to approximately the baseline ratings they had given before considering the two job offers.
Their preferences weren’t only constructed, but they were also short-lived and continued to be highly context-dependent. In the second study, Simon found these constructed preferences to be highly elastic.
They changed shape to meet the needs of a particular choice situation, but then they “bounced back” to roughly their original shape when disengaged from that choice context.⁹
Based on these studies and related research, marketers need to think about preferences not only as constructed, but also as both transient and elastic. Simon’s work is important because it shows that expressed preferences not only change in response to different choice situations, but they also tend to revert to a common baseline in the absence of imposed choice constraints.
This reversion to baseline may be a function of inherent preferences hidden beneath preference construction. But as long as those inherent preferences can be so easily overridden in different choice contexts, constructed preferences are more likely to be operative in day-to-day consumer choice and behavior.
Preferences Need No Inferences
Robert Zajonc’s discovery of the mere exposure effect was a great contribution to social psychology. This discovery disproved the seemingly self-evident proposition that our preferences must have logic and reasons behind them — we shouldn’t be able to have a preference for one thing over another unless we have a reason for that preference.
But Zajonc’s famous dictum that “preferences need no inferences”¹⁰ proclaims just the opposite — that liking and preference can be generated by repetitive mere exposure in the complete absence of reasons or even conscious awareness.
If simple repetition can condition our preferences in such a predictable and persistent way, how much should marketers rely on stated preferences as a basis for designing and marketing their products and brands?
Why Expressed Preferences Don’t Reliably Predict Behavior
The elephant in the room for marketers and market researchers is the fact that explicitly stated preferences extracted from surveys and interviews are often unreliable predictors of future consumer choice and behavior.
The problem is not that stated preferences never predict subsequent behavior, but that they fail to do so dependably. This makes it difficult and risky to allocate product development and marketing resources based on consumer preference data alone.
The failure of expressed preferences to reliably predict future behavior leads to an even deeper question regarding the nature of opinion surveys as instruments of knowledge acquisition.
If we take seriously the idea that human beings are highly intuitive thinkers who operate at both conscious and unconscious levels and respond to the world around them in contextually-sensitive ways that might not be available to their conscious minds, we can begin to make sense of that deeper question:
What are people actually doing when they respond to questions on a survey questionnaire?
The standard answer is that when people say they prefer X over Y on a survey questionnaire, they’re simply describing a preexisting state of feeling more favorable toward X than Y.
But, as we have seen, that view doesn’t conform with the large body of accumulated evidence that shows expressed preferences to be more often constructed at the moment than accessed from a preexisting memory. So what else might be going on?
One of the first — and still one of the best — answers to this question was put forward in an influential paper written by political scientists John Zaller and Stanley Feldman in 1992 titled “A simple theory of the survey response: Answering questions versus revealing preferences.”
Zaller and Feldman recognized something that’s still missed by many in the survey research field: Answering questions on a survey form is a unique cognitive experience that has its own rules, risks, and expectations.
Ordinary people don’t usually get asked to state preferences, rate products, or predict their future buying behavior using standardized seven-point scales and other measurement devices researchers learn about in graduate school.
This leads to a conclusion that’s so self-evident it is easy to miss.
When people respond to questions on a survey, they’re not revealing preferences so much as entering into an artificially-imposed social interaction. In this interaction, they’re forced to organize their thoughts and feelings in ways that may be completely foreign to their normal modes of thinking.
So what do ordinary people do when they find themselves in this unique circumstance? Zaller and Feldman’s insight is worth quoting at length:
“Most citizens, we argue, simply do not possess pre-formed attitudes at the level of specificity demanded in surveys. Rather, they carry around in their heads a mix of only partially consistent ideas and considerations.
When questioned, they call to mind a sample of these ideas, including an oversample of ideas made salient by the questionnaire and other recent events, and use them to choose among the options offered.
But their choices do not, in most cases, reflect anything that can be described as true attitudes; rather, they reflect the thoughts that are most accessible in memory at the moment of response.”¹¹
The problem with surveys, according to this perspective, has less to do with how they measure and more to do with what they measure.
Or more correctly, what survey researchers believe they measure. When preferences are seen as inherently ambivalent and answers to survey questions are seen as results of in-the-moment searches for accessible memory cues, several deficiencies in survey results become understandable.
People give inconsistent answers because the attitudes being measured aren’t pre-formed. People are guessing and reconstructing their preferences every time they’re asked, and those processes naturally lead to different answers at different times.
Since the publication of Zaller and Feldman’s article in 1992, additional research has revealed that ambivalence isn’t the only source of variation impacting how individuals express attitudes and preferences. Another potential source of survey answer bias is question difficulty.
As Daniel Kahneman has observed in numerous articles and presentations, when people are confronted with a question they find difficult to answer, they often replace it mentally with an easier question that they answer instead — usually without being aware of the substitution.¹²
This easy-for-hard substitution effect makes survey answers even more suspicious as reliable guides to future behavior.
Not only are survey respondents often constructing their answers on the fly, but they may also be constructing answers to different questions than the ones researchers believe they’re asking.
This effect may be occurring outside the awareness of either the researcher or the respondent.
Expressed preferences are better thought of as stories consumers tell themselves — and market researchers — to make sense of how, what, and why they buy.
Given the fragility of many consumer preferences, marketers are more likely to get generalizable, actionable results by measuring what consumers do, not what they say, when faced with product or brand choices.
Also, marketers need to be aware that consumer behavior can change radically across different contexts and situations. Constructed preferences are deeply influenced by the context within which a choice is made.¹³
This article was originally published on medium.
Stephen Genco is a writer, speaker, researcher, and marketing consultant. He is author of Intuitive Marketing (2019), a study of persuasion and influence in marketing theory and practice, and co-author of Neuromarketing for Dummies (2013), a comprehensive overview of neuromarketing science, applications, methodologies, and ethics. In 2006, he founded one of the first neuromarketing research firms, and from 2009 to 2012 he was Chief Innovation Officer at one of the largest. He is currently Managing Partner at Intuitive Consumer Insights, where he focuses on marketing education and consulting, helping clients develop and execute marketing programs and business strategies that leverage the latest advances in brain science.