The IAT – A Guide for Marketing Researchers
by Audrey Guinn, Ph.D.
What is the IAT?
The Implicit Association Test (IAT) created by Greenwald, McGee, & Schwartz (1998) is believed to measure implicit associations with regards to topics such as race, sexuality, weight, gender, nationality, age, skin tone, religion, and disability (among others).
Implicit associations (also known as implicit bias) are unconscious attitudes that are formed via learned associations, also known as stereotypes. Stereotypes are positive or negative associations that we make about a group of people. For example, some prominent American stereotypes are that men are better at math and that African Americans are superior athletes. These associations are shaped by our experience with the world. We learn them through the media, societal assumptions, interactions with others, relationships with others, and our own personal experiences. Whether we agree with the stereotypes or not, they can influence our thoughts and behaviors. Because this bias is unconscious, we may not even be aware of the biases we hold, hence the wording “implicit.” It is believed that everyone has different types of implicit bias that affect them.
Explicit attitudes, on the other hand, are conscious preferences that people can directly articulate. Prejudice and racism are explicit forms of bias. Prejudice is a conscious thought, feeling, or belief that is negative towards another group of people. Racism takes prejudice one step further. Racism is a set of beliefs in racial superiority, which justify and are-the-basis-for discriminatory and racist actions. Not everyone is prejudiced or racist. These are conscious beliefs which are learned, agreed with, supported, and practiced over time.
Many explicit attitudes, racism included, are considered socially undesirable and therefore, people who subscribe to these ideologies are taught to suppress them. Even if they believe in the ideologies wholeheartedly and even if they are given the opportunity to report their beliefs anonymously, they are unlikely to openly admit their beliefs. Because of this, researchers have had a difficult time investigating these socially undesirable forms of explicit bias. And this is exactly the reason the IAT was created.
The IAT is a video game-like activity where participants make quick decisions sorting concepts (e.g., black people and white people) and evaluations (e.g., good or bad). The idea behind the IAT is that those who associate black people with negative thoughts, beliefs, or feelings will have an easier time sorting the words when white people and good words are paired together as compared to when black people and good words are paired. Similarly, people who have no negative associations or no preference between black or white people will sort the words roughly equally regardless of how the words are paired together.
In theory, the more implicit bias you exhibit, the greater likelihood of discriminatory behavior. Therefore, someone who is outwardly racist should in theory have high associations between the categories “white” and “good” and at the same time have high associations between the categories “black” and “bad.” Therefore, they should show high implicit racial bias on the IAT. Also in theory, people who are not racist may still exhibit slight associations between the categories “white” and “good” as well as “black” and “bad.” This is due to the stereotypes we are exposed to and how they impact us unconsciously. This idea extends beyond race categories and can be used to assess implicit associations for many other areas.
State of the Current Research
While the IAT has been highly regarded by researchers for roughly two decades in the field of Psychology, it has come under scrutiny as of late. Some researchers question the reliability and validity of the IAT. Some researchers believe the IAT is measuring implicit associations while others are not so sure. Others wonder what exactly is being measured. While some researchers point out that, at the very least, the IAT is impacted by explicit preferences. Some researchers suggest that the IAT is an instrument that measures more noise than anything else, and that the results are no more useful than asking respondents their opinions on surveys. Others critique the scoring method which has an arbitrary cutoff signifying the strength of implicit bias. Many scientists in this field agree that, while the IAT could potentially be very important, it needs much more research to flesh out its credibility, and therefore, should not be released to the public or used to inform laws.
Despite the discontinuity between researchers investigating the IAT, marketing researchers have already begun to use it in their research. The next sections are geared towards helping marketing researchers draw their own conclusions regarding whether to use the IAT in their research.
Pros to Using the IAT in Marketing Research
- Respondents often have a difficult time articulating their preferences. If the original creators of the IAT are correct, that it measures implicit associations, then this tool allows researchers to uncover unconscious experiences and hard-to-articulate preferences.
- Respondents often show acquiescence bias. Respondents often want to present themselves as agreeable and that tends to show up in research as heavily selecting the top-2 box (somewhat agree/strongly agree). Because the IAT does not explicitly ask for preferences, there is no potential for acquiescence bias. Instead, preferences are revealed through comparison of reaction times and calculated implicit association scores. Even negative preferences for a brand, category, or product could be discovered.
- Respondents often grow bored with typical survey methods. Fatigue and boredom are two major problems in survey research today. The IAT provides respondents with a fun, new, engaging experience.
- The IAT could provide an additional way to investigate negative or positive preferences for brands/categories/products.
Cons to Using the IAT in Marketing Research
- It is unclear as to whether the IAT is measuring implicit associations. While some scientists argue that it is, others are not so sure. If the scientists in the IAT field are not agreeing on what is being measured, it begs the question, how can marketing researchers base results on this methodology which could have a massive impact on real client dollars?
- Results are likely to be uninformative unless the brands/products/categories tested are associated with stereotypes. Let’s assume that the original creators of the IAT were correct – that the IAT measures implicit associations and is therefore built on the associations learned from exposure to stereotypes. That means, to work correctly, the brands, categories, and products would need to be associated with stereotypes, and the respondent would need to have been exposed to those stereotypes in order to elicit an implicit response. Brands, products, or categories that do not have stereotypes enmeshed with them would likely yield uninformative results – showing no implicit bias or minimal differences for those tested. Furthermore, brands/products/categories that do have strong stereotypes associated with them would likely elicit strong explicit preferences, which could be assessed using a simple survey question.
- Results are likely to be uninformative unless the consumers reviewing the brands/products/categories tested have strong opinions about them. Some IAT researchers believe that the IAT may be influenced by the respondent’s explicit preferences. If we take this one step further and assume that the IAT is another measure of explicit preference, the respondent would need to be, at minimum, aware of the brand being tested in the IAT. Even still, if respondents do not care about the brand, category, or product, the IAT is unlikely to be helpful. For example, testing two grass seed brands among the general population would likely not yield anything useful. Testing these brands among landscaping professionals probably would reveal helpful results.
- The IAT does not reveal reasons for the preference or aversion. When asking a respondent if they prefer a brand in a traditional survey, marketing researchers can include a follow-up question for those who answer in the negative to probe the reason why. This is not available to marketing researchers who are using an IAT to uncover preferences – unless – they calculate the scores in real time and assign respondents to answer the probing question based on the IAT score. But, if the IAT does indeed measure implicit associations, respondents may be surprised by their preference and may not be able to articulate the reasons for it.
- The strength of the implicit associations is determined by an arbitrary cutoff. Because there is no scientific or statistical reason for the cutoff, how can researchers be sure that they are truly understanding the respondents’ implicit associations or preferences?
- The IAT may be more time and cost prohibitive. First, the IAT requires a full experimental method to ensure a balanced design and to rule out extraneous effects. This requires someone with a background in experimental methods to ensure correct setup. Second, the IAT is similar to a video game, which would require someone with programming knowledge who could implement the task in a survey. Third, it requires calculations on the backend to determine implicit associations. All of these may incur more time and costs, compared to traditional methods.
- The IAT may not yield any additional information beyond what is available through traditional research methods. Asking respondents their preferences directly yields information as to whether the brand/category/product is perceived as negative or positive on the whole. Including a choice model in the survey not only reveals brand/category/product preferences but also can reveal purchasing behavior. Examining client databases can reveal actual consumer purchasing behavior and preferences. Thus, why spend money and time programming the IAT into a survey and analyzing the data if traditional survey methods reveal the same information?
Considerations for Marketing Researchers Who Want to Use the IAT in Marketing Research
With the information above, the question then boils down to: can the IAT provide additional information that traditional survey methods cannot? Before this question can even be answered, marketing researchers who are already using or are considering using the IAT should perform their own research on the IAT. The analyses listed below will help ensure that the IAT is performing in the way expected and that relaying results to clients based on the IAT will not cause potential harm.
- Use the IAT scores to predict explicit preference. Include both explicit measures and the IAT on the same subject (brand/category/product etc.) within the survey. Strong IAT scores should be related to explicit preference. Slight and neutral IAT scores will likely not be related to explicit preference.
- Use the IAT scores to predict past purchasing behavior. Retrieve consumer purchasing behavior from a client database and have the same consumer complete an IAT on the same products. The IAT scores should be related to purchasing behavior.
- Use the IAT scores to predict choice model preference. Include both a discrete choice model and IAT on the same subject within the survey. The IAT scores should be positively related to choice model preference.
- Use the IAT scores to predict behavior over and above the traditional survey measures using a hierarchical regression. After controlling for explicit preference via traditional survey measures, the IAT scores should contribute unique variance to predicting behavior. This last analysis will answer the question of whether the IAT can provide additional information above what traditional survey measures provide.
While the IAT has been considered the gold standard in accessing unconscious associations for many years, new research shows that there are more questions that need to be answered before applying the stamp of approval. At the very least, more research needs to be conducted to maintain confidence in the IAT. Even still, some marketing researchers are currently using this tool in consumer research. There are several benefits to using the IAT, namely that it provides an engaging task that can bypass respondents’ tendencies to rate everything positively, and that can reveal negative preferences for products, categories, or brands. However, the questions remaining about the IAT should serve as barriers to basing million-dollar decisions solely on the IAT results. At a minimum, marketing researchers should conduct their own research to determine whether the IAT measures what it’s supposed to measure, whether it is adding helpful information beyond traditional survey methods, and whether it’s cost effective for your research agency.
- Blanton H., Jaccard J., Klick J. Mellers B., Mitchell G., Tetlock P. (2009) Strong Claims and Weak Evidence: Reassessing the Predictive Validity of the IAT. Journal of Applied Psychology. (94), 567
- Nosek B. A., Banaji M. R., Greenwald A. G. (2002). Harvesting implicit group attitudes and beliefs from a demonstration web site. Group Dynamics: Theory, Research, and Practice. (6), 101–115.
- Greenwald A. G., Banaji M. R. (1995). Implicit social cognition: Attitudes, self-esteem, and stereotypes. Psychological Review, 102(1), 4–27. 10.1037/0033-295X.102.1.4
- Greenwald A. G., McGhee D. E., Schwartz J. L. K. (1998). Measuring individual differences in implicit cognition: The Implicit Association Test. Journal of Personality and Social Psychology, 74, 1464–1480.
- Hahn, A., Judd, C. M., Hirsh, H. K., & Blair, I. V. (2014). Awareness of implicit attitudes. Journal of Experimental Psychology: General, 143(3), 1369–1392. https://doi.org/10.1037/a0035028
- Schimmack U. (2021). The Implicit Association Test: A method in search of a construct. Perspectives on Psychological Science, 16(2), 396–414. 10.1177/1745691619863798
- Schimmack U. (2021). Invalid Claims About the Validity of Implicit Association Tests by Prisoners of the Implicit Social-Cognition Paradigm. Perspectives on Psychological Science, 16(2): 435-442. 10.1177/1745691621991860
About the Author
Audrey Guinn, Ph.D. (firstname.lastname@example.org) is a Statistical Consultant in the Advanced Analytics Group at Decision Analyst. She may be reached at 1-800-262-5974 or 1-817-640-6166.
Copyright © 2023 by Decision Analyst, Inc.
This posting may not be copied, published, or used in any way without written permission of Decision Analyst.