Guidelines for Choosing Attributes in Implicit (and Explicit) Association Research

These days, leading-edge consumer researchers are conducting studies that measure implicit processing, which is popularly known as System 1. Importantly, this type of research reveals consumers’ automatic (a.k.a., implicit) associations with “stimuli of interest” (SOIs), which can be brands, products, ads, package designs, prices, taglines, people, or any number of other marketing elements. Choosing attributes for implicit association studies can be challenging.

Implicit associations represent how consumers first react to SOIs before they consciously deliberate on them in purchase decisions (if they deliberate at all). Conscious deliberative processing is also known as explicit or System 2. For example, this woman’s implicit reaction about her neighbor’s “darling” little girl might be one of great distress. However, upon deliberation, and especially in a social situation, she might express great delight.

All purchase decisions involve implicit processing, but not all involve deliberative, explicit, System 2 processing. Stated another way, we don’t always consciously think about what we decide to buy before we buy it, but we always have a System 1, implicit reaction. Therefore, implicit reactions are critical purchase decision drivers.

Simply put, implicit associations are unintentional, uncontrollable, and in some way nonconscious “thoughts and feelings” we have about things (SOIs). For quantitative studies, thoughts and feelings (both implicit and explicit) are often represented as attribute words. Emotion attributes include words such as happy, surprised, proud, confident, bored, frustrated, worried, and disappointed. Trait attributes — which can relate to personality, physical features, or functional benefits — include words such as friendly, smart, strong, easy, funny, attractive, hostile, dumb, weak, difficult, boring, and ugly.

Studies are often conducted to measure the degree to which SOIs are associated with emotional and trait attributes, both implicitly (automatically, without conscious deliberation) and explicitly (upon conscious deliberation). Our IE Pro YOU® platform allows clients to easily conduct implicit and explicit association studies for their SOIs. These studies are important because companies base their product development and positioning strategies on attribute associations that lead to purchase. (Think of Coca-Cola’s association with ‘happy’ or Michelin’s association with ‘safe’.)

Again, choosing attributes for implicit association studies can be challenging. In designing such studies, clients often grapple with what attributes they should select. Having conducted many of these studies, we’ve developed some guidelines for selecting attributes for implicit (and explicit) association studies.

  1. This may seem obvious, but surprisingly, it isn’t always considered: the attributes should relate to the SOI’s positioning strategy. If you want your brand, product, or service to be positioned as playful in your consumer target’s mind, then attributes related to playful (and its opposite) should be chosen.
  2. Furthermore, along this “strategic” criterion, choose attributes that represent your competitors’ strategies to see how far or near you are to your competitors.
  3. If strategy has not yet been developed, choose attributes that cover the spectrum of possible emotions or traits. This is no easy task, because the spectrum can be very wide and attribute selection is limited in studies. However, emotional taxonomies exist from which a reasonable number of emotion attributes can be selected. (See Russell’s Circumplex or Plutchik’s Emotion Wheel as examples.) Trait attribute taxonomies are less prevalent (at least to my knowledge), but common sense and Google searches can help.
  4. When choosing attributes for implicit associations, refrain from choosing opposites by putting the word “not” in front of the attribute. For example, if proud is a selected attribute, don’t choose ‘not proud’ as its opposite. Implicit, System 1 processing is associative, not propositional, in nature. This means that, since the word ‘not’ requires propositional, logical thinking to understand that it’s the opposite of the word that follows it, implicit processing can interpret the attribute as the word without the word not — the exact opposite of what you want. For example, the implicit mind is likely to interpret ‘proud’ and ‘not proud’ as the same.
  5. Keep the attributes simple. It’s best to assume that many respondents may not fully understand the meaning of higher-order emotional or trait attributes like irreverent, copious, malevolent, blissful, buoyant, and appeased. If they don’t understand what the words mean, they will likely not choose them, even if they feel or think the SOI conveys them. Not only does simplicity make it easier for respondents to know what you’re talking about, it mitigates System 2, propositional thinking.
  6. Be careful not to mix emotion attributes with trait attributes. This can confuse respondents and interpretation of results. For example, if a respondent is asked to choose which of the following attributes — happy, confident, convenient, bored, difficult, or angry — best conveys how one feels when they consider using a certain brand of blender, the trait attributes convenient and difficult can be inconsistent with the emotion attributes happy, confident, bored, and angry. Respondents might find it difficult to choose between happy and convenient or angry and difficult. Furthermore, if more respondents choose convenient than happy, concluding that the blender evokes more convenience than happiness doesn’t make sense. It’s true that some attributes are versatile enough to fit both categories (e.g., a product can be safe and one can feel safe from it), but for the most part mixing emotion and trait attributes in the same response task creates an apples-to-oranges situation.
  7. With emotional attributes, sometimes researchers and clients want to be exact in choosing the word for similar emotions because, as the thinking goes, emotions are distinct “natural” phenomena. For example, choosing between the attributes ‘frustrated’ and ‘annoyed’ can be excruciating to those who think the two are naturally distinct emotions (and even that they can be distinguished neurologically or psychophysiologically). Concerns like this are often exaggerated, and final decisions on which attribute to choose are subjective. Feldman Barrett (2017) makes the strong case that emotions are cognitively constructed rather than neurologically or psychophysiologically constant, which makes whatever people call them conceptually correct. The point here is not to agonize over choosing attributes that may be close in meaning. Discuss choices with your team and, if in doubt, and if it’s practical, choose both just to see what happens!
  8. Finally, we recommend using words not faces to represent emotional attributes, especially when discrete secondary feelings (vs. primary, basic emotions) are involved. Following the belief that emotions are more conceptual, cognitive constructions than distinctly and consistently represented by specific facial muscle movements, finding facial expressions that universally represent feelings like surprised (vs. afraid), happy (vs. excited), embarrassed (vs. sad), or relaxed (vs. satisfied) is quite challenging. Why not just use the words? Yes, words can mean different things to different people, too; but we think not as much as facial expressions can.

Implicit attribute association research can be extremely valuable. And choosing attributes for implicit association studies can be quite challenging. We hope these guidelines help in this work, if you decide to take it on.

We’re happy to help, so please contact us if you have specific questions.

Until next time.

Leave a Reply

Your email address will not be published. Required fields are marked *