In my favorite of his studies (Unconscious Affective Reactions to Masked Versus Angry Faces Influence Consumption Behavior and Judgments of Value, 2005), to regulate people’s moods unbeknownst to them, he (and his colleagues Kent Berridge and Julia Wilbarger) subliminally primed thirsty respondents with pictures of either happy faces or angry faces. He then had them engage in a couple of “consumer” tasks — drinking a beverage and indicating how much they would pay for the beverage. Between the priming and the beverage drinking/rating, respondents engaged in a seemingly unrelated task — explicitly rating how they felt after seeing a series of emotionally neutral faces.
The results were amazing, I think. The thirsty respondents who were subliminally primed with happy faces drank more of the beverage and said they would pay more for it than the thirsty respondents who were subliminally primed with angry faces. In addition, both groups of respondents explicitly rated their feelings the same. Furthermore, the invisibility of the happy vs. sad prime was confirmed — people were not consciously aware of it.
Why is this important? Besides showing that our behavior can be influenced by our emotional state (no big revelation) unconsciously (a bigger revelation), it speaks indirectly to potential misgivings in direct emotional self-report research. Specifically, it warns us not to fully trust what people say when we ask them how they are or were feeling about something or why they would do or did something. Aside from sometimes deliberate attempts to look good to the reseacher or others that might be in a (focus) group discussion, people often don’t know how they are feeling at present or did feel in the past, which speaks to what drives or drove a certain behavior.
If you need more proof, search for and read the works of Dan Ariely, Gerald Clore, or Elizabeth Cowley, just to name a few of many researchers who have found that we are not always in touch later with how we felt (and what caused our behavior) in the past.
In what I think is a classic piece of work (Belief and Feeling: Evidence for an Accessibility Model of Emotional Self-Report, Psychological Bulletin, 2002), Michael Robinson and Gerald Clore point out that self-reports of past emotional experiences can inherently only be estimates of emotional truths. Instead they are one or more of a variety of reconstructions which include episodic memory, situation-specific beliefs, or identity-specific beliefs. The point is that when we ask someone “how did you feel about…,” their answer cannot be an “online emotion” (i.e., the emotion associated with the actual experience). It must be an estimated (and therefore, flawed) reconstruction of it in our memory.
In her article “Remembering an Affective Reaction to a Previous Consumption Experience” (Journal of Consumer Research, 2007), Elizabeth Cowley shows how inaccurate respondents’ emotional memories can be when asked how they felt when engaged in an activity only about 30 minutes prior to being asked.
In work I am currently conducting, I’m seeing that certain implicit feelings (i.e., those measured by indirect non-self report means) are more predictive of past purchases of a certain brand of food product than any of the explicit feelings the respondents said they have toward the brand.
So the next time you think about conducting emotional consumer research that explores how people say they are feeling or did feel, strongly consider supplementing their direct self-report with techniques that, at least to some degree, circumvent changed memories, subconsciously filtered defenses, or, in some cases, consciously deliberate efforts to “put on a good face.” Techniques can include unobtrusive observation (commonly called ethnography), projective techniques, psychodrama, hypnosis-interviewing, misattribution, or psychophysiological techniques to name a few. Yes, it will likely cost a little more. But more effectively surfacing the truth, or truths that you were not aware of, may well be worth a little extra cost.