Special Session Summary Implicit Measures As Effective Market Research Tools

Nancy M. Puccinelli, Emerson College
[ to cite ]:
Nancy M. Puccinelli (2002) ,"Special Session Summary Implicit Measures As Effective Market Research Tools", in NA - Advances in Consumer Research Volume 29, eds. Susan M. Broniarczyk and Kent Nakamoto, Valdosta, GA : Association for Consumer Research, Pages: 498-500.

Advances in Consumer Research Volume 29, 2002     Pages 498-500



Nancy M. Puccinelli, Emerson College

While it is possible that explicit and implicit knowledge correspond, the exciting opportunity for marketers is that often there is a discrepancy; that is, what a consumer believes explicitly may have no bearing on their actual behavior. Further, evidence suggests that contextual cues, such as the type of program an ad appears in, can have a significant impact on consumer behavior by directly affecting one’s attitudes at an implicit level and/or activating pre-existing attitudes. Thus, knowing how implicit attitudes work can better inform our understanding of consumer behavior.

Given the important roles of implicit attitudes in understanding and predicting consumer behavior the proposed session explored how implicit measures may be used more effectively in marketing. The first paper entitled "When what consumers say isn’t what they do: The case of ethnocentrism" extended work on implicit measures as a measure of ethnocentric bias to branding. The author emphasizes the implications for this work for a global versus multinational approach to international branding.

Entitled "How to influence the implicit mind? The role of contextual information via mood manipulation", the second paper examined how context may moderate implicit associations with brands. The findings presented illustrate a way in which a contextual factor could impact implicit associations with brands and influence consumer judgment.

The third paper, "I don’t know why but I don’t like it: The impact of context on brand evaluation" examines how context can impact evaluation of a brand. This work extends the work on implicit associations by shedding light on how a given context might both directly and indirectly affect brand evaluation and purchase intention.

The goal of this session was to demonstrate the range of applications of implicit measures to branding and consumers’ associations with brands. In addition, this session sought to shed a unique light on our understanding of implicit attitudes by drawing on the strong associations consumers have with brands. First, demonstrating an ethnocentric bias among consumers for local brands, adds greater ecological validity to laboratory research showing a similar effect. Second, the investigations examining contextual factors impacting implicit associations extends te current body of knowledge on implicit associations by identifying moderators of these effects. This assembly of papers moves us toward identifying ways in which we might best employ an integrated market research approach that better predicts consumer behavior. Implicit measures may provide a richer understanding of evaluation of a product and the meaning conveyed by a brand. In the future, market research that combines explicit, implicit and behavioral variables will be necessary to further reveal the interrelation between conscious and unconscious information and its impact on consumer behavior.



Kathryn Braun, Marketing Memories

Gerald Zaltman, Harvard Business School

Ethnocentrism is the human tendency to view one’s own culture as superior to others. The trait has evolutionary roots in the practice of territoriality: our ancestors needed a quick, automatic mechanism to detect members of their tribe from members of the enemy. The result of group categorization, coupled with the motivation to uphold a positive self concept, leads group members to evaluate their own group more favorably than other groups.

In marketing, ethnocentrism emerges as the consumer preference for domestic products or prejudice against imports (Sharma, Shimp & Shin, 1987). Consumer ethnocentrism is supported through slogans like "Buy American." Being able to identify consumers that hold ethnocentric beliefs can be a means for both positioning the home country advantage, or, in the case of exporting, knowing when to diminish country-of-origin information. Traditionally, consumer ethnocentrism has been measured using the CETSCALE, a paper and pencil measure that asks consumers to report on their beliefs toward buying American products.

The difficulty in assessing consumer beliefs arises when consumers may feel that their beliefs or attitudes may not be socially acceptable. In addition, there are some beliefs that consumers may not be aware they hold. Social psychologists have attempted to circumvent self-reporting biases by using indirect measurement techniques, such as computer assisted response-time software. It is typical to find that what a person reports to be their attitude (for instance, racial) may differ dramatically from how they behave in tasks that put that attitude to test (Greenwald & Banaji, 1995). One such response-time measure, the Implicit Association Test (IAT), has been used to measure implicit consumer attitudes, but thus far has only been applied in situations where it was not expected that the explicit and implicit would differ (Brunel, Collins, Greenwald & Teitje, 1999).

Admitting an own-country bias, like admitting a racial stereotype, holds negative connotations, and presents an opportunity where explicit reporting and implicit behavior may differ. We measured the attitudes and behavior of 71 Harvard MBA students (34 male, 37 female; 53 US citizens, 18 foreign students) using both an explicit self-report measure (CETSCALE) and an implicit behavioral measure (the IAT).

The CETSCALE includes statements like: "American people should always buy American products instead of imports" and respondents indicate the degree to which they agree with the statements. The CETSCALE has been widely used in marketing and has purportedly been validated across cultures to capture people’s belief regarding their interest in own-country brand preference. The question is whether or not this scale is really capturing the ethnocentric bias? Might people be afraid to admit their bias in such an explicit way? From our survey that appeared to be the case: the average CET score was very low, 1.7 on a scale of 1-7, where higher values indicate more ethnocentrism. There was little variation (SD=.86) with the highest value being 3, lowest 1. Most troubling, though, there was no difference between US and foreign students on this measure: US mean=1.7, foreign mean=1.9, t<1.

The IAT test was designed to revealown-country bias (in the case of US citizens) for brands. The issue was whether or not US brands would be more associated to favorable words than foreign brands as revealed by faster reaction times. We found the US students were faster (and more accurate) when pairing a US brand like "Hallmark" with a positive attribute like "lucky" than pairing a foreign brand like "Kawasaki" with a positive trait. An IAT score was formed for each student where higher values indicate more bias. The average IAT for US students was 247.2; -29.33 for foreign students, significantly different at t(69) p<.0001. 98% of the US students showed an IAT effect; while only 27% of the foreign students showed an effect.

The CETSCALE is an explicit measure of ethnocentrism; the IAT an implicit one. A marketer could argue that the CETSCALE may not capture all the bias but at least it should be related to it. However, as found in other literatures involving explicit and implicit measures, we found a disassociation: the overall correlation between the CETSCALE and IAT was -.18. For US students the correlation was -.09; for foreign students the correlation was actually quite high at .66. It could be that because US ethnocentrism is not considered a stigma for foreigners there is greater consistency between what they say and what they do.

In conclusion, there is a strong underlying preference for own-brands, as demonstrated through the IAT. This effect is not captured by traditional marketing measures of ethnocentrism (CETSCALE). US students were faster, and more accurate when associating US brands with favorable terms, foreign brands with negative terms. This trend did not hold for foreign students, very few showed any bias and as a group they exhibited a negative IAT effect.

This ACR presentation occurred several weeks after the September 11th terrorist attacks. The data had been gathered more than a year earlier. A question arises: could ethnocentrism be seen as more favorable in times of patriotism? Would there be less of a disassociation under such conditions? Another question arising from this study is which scale would best predict choice behavior in a real supermarket setting. Future research should address these issues.


Brunel, F.F., C.M. Collins, A.G. Greenwald & B.C. Teitje (1999), "Making the Private Public, Accessing the Inaccessible: Marketing Applications of the Implicit Association Test". Working paper.

Greenwald, A. G. & M.R. Banaji (1995), "Implicit Social Cognition: Attitudes, Self-Esteem and Stereotypes," Psychological Review, 102, 4-27.

Shimp, T.A. & S. Sharma (1987), "Consumer Ethnocentrism: Construction and Validation of the CETSCALE," Journal of Marketing Research, 24, 280-289.

Sharma, S., T.A. Shimp & J. Shin (1995), "Consumer Ethnocentrism: A Test of Antecedents and Moderators," Journal of the Academy of Marketing Science, 23(1), 26-37.



Fred W. Mast, Harvard University

Nancy M. Puccinelli, Emerson College

An important distinction has been drawn between explicit and implicit measures in assessing the associations a consumer holds about a product. Explicit measures are typically quite familiar to marketers. They are direct, relatively fast and easy to apply. For example, they miht take the form of paper and pencil surveys in which consumers might be asked how much they like a specific brand, how positively they think about it, or whether they remember seeing an ad for that brand. Implicit measures are much less direct and they seek to measure automatic evaluations that underlie our attitudes and beliefs. For the most part, implicit measures rely on actual observable behavior such as response times in a computerized task. A particular method for diagnosing the associative structures is provided by the implicit association test (IAT).

A computerized IAT was used in this study to measure differential association of two brands (Ben & Jerry’s and Marlboro) with the attributes "good" and "bad". Pictures of the two brands appear in a two-choice task that is paired with a second task, in which words have to be evaluated with respect to the two attributes. When participants are instructed to use the same response key for highly associated categories (e.g., Marlboro + bad), performance is faster than when less associated categories (e.g., Ben & Jerry’s + bad) share the same response key. This performance difference represents an implicit measure of differential association of the concepts with the attribute.

Though the IAT has been used in marketing research questions, it is still relatively unknown what factors influence the formation of consumers’ implicit attitudes. Is it possible to influence consumers’ implicit attitudes? For example, knowledge about the role of immediate context information might help to identify important factors leading to more effective product placement and its impact on purchase behavior. Specifically, in this study, we were examining the role of contextual information in the form of consumers’ mood. Prior to the IAT procedure, the participants (N=20) were exposed to a 20-minute video sequence which either induced a happy (from "Liar Liar" and "Bill Cosby himself") or sad mood (from "Steel Magnolias"). After they finished the IAT with Ben & Jerry’s and Marlboro as brands and good and bad words as evaluative attributes, they filled out a questionnaire about their explicit brand attitudes. All participants were tested individually in a quiet testing room equipped with a desktop computer and VCR.

As expected this research overall found positive associations with Ben & Jerry’s and negative associations with Marlboro. The difference between the two parings, noncompatible (e.g., Ben & Jerry’s + bad) minus compatible (e.g., Ben & Jerry’s + good) defines the IAT effect, 242 ms, t=4.007, p=.0003). Considering how consumers mood might impact the implicit associations, the research found that differential associations were stronger for consumers in the sad mood condition, effect size: r=.21. In more detail, this IAT effect was produced by the slower responses to noncompatible pairings in the sad mood condition, whereas the responses to compatible pairings were not influenced by the mood manipulation. Furthermore, while the implicit measure turned out to be highly predictive of explicit attitudes for happy participants (r=.66, p=0.034), an implicit measure actually predicted a tendency in the opposite among sad consumers (r=-.63, p=.067).

The key finding of this study was that we were able to capture the effect of mood on implicit associations. In particular, the slower response to noncompatible pairings for sad consumers suggests evidence for a selective sad mood effect when product information is ambiguous. Interestingly, a general slowdown in the sad mood condition cannot account for our results because the responses to compatible pairings did not differ between mood conditions. In a real-world marketing context, ambiguous product information can arise from various factors such as an inappropriate price level (e.g., when an expensive brand product goes on sale) or from unsuitably positioning the product within the immediate store environment. Future research including neutral mood inductions will be conducted to further investigate the role of sad mood on the processing of ambiguous information. Our approach is to combine explicit and implicit methods with behavioral choice designs to further reveal the interrelation between conscious and unconscious information and its impact on consumer behavior.



Nancy M. Puccinelli, Emerson College

Fred W. Mast, Harvard University

Given the resources allocated by firms to build brand equity, it would seem crucial to know the impact of brand placement on consumer evaluation. Imagine, for a moment, that you are watching a serious film and during a particularly sad scene, in walks a character drinking a can of Coke. While you are watching the movie, you are hardly aware that the character was even drinking anything, yet, you find later that day you prefer juice to your usual Coke. The present research sought to examine this phenomenon, asking: How does context affect stimulus evaluation?


Considerable research suggests that context serves to implicitly prime individuals and lead them to evaluate a stimulus differently (Forgas 1995). Recent research suggests these primes also impact behavior, finding that individuals primed with a college professor stereotype perform better at Trivial Pursuit (Dijksterhuis et al. 1998). Further, research suggests that the valence of a prime may interact with the valence of a target. For example, individuals in a negative mood are less likely to prefer a positive story (Erber, Wegner, and Therriault 1996) or watch a positive spokesperson (Puccinelli and Moon 2001). The current research examined the impact of a positive or negative context on attitudes toward a positive or negative brand.

Study 1

Study 1 examined the impact of positive and negative context on brand evaluation, hypothesizing, 1) positive and negative context would have an affect-congruent effect on brand evaluation and, 2) the effect of negative context would be greater for a positive than a negative brand.

To test these hypotheses, 119 MBAs viewed videos that presented positive and negative brands in a positive or negative context. Using 2.5-minute clips from Bill Cosby Himself and Steel Magnolias, two experimental videos were created in which the clip appeared in the center of the screen while images of a positive brand (i.e. Ben and Jerry’s) and a negative brand (i.e. Marlboro) appeared around the perimeter. That is, the clip in the center of the screen served as the context while images of the brands served as stimuli, appearing above, below, to the left or right of the clip. After viewing the experimental video, participants indicated their evaluation of each brand on seven dimensions (e.g. good-bad).

To examine the effect of the positive and negative context on brand evaluation, a 2 (brand: positive or negative) within x 2 (context: positive or negative) between-participant analysis of variance was calculated on a composite of the attitude scale. The main effect observed for brand served as a manipulation heck and as expected Ben and Jerry’s was rated more positively than Marlboro (F (1, 117)=231.49, p<.001). Consistent with the first hypothesis, a main effect for context found that both brands were rated more negatively when embedded in the negative rather than the positive context (F (1, 117)=8.61, p<.01). Follow-up t-tests show some support for the second hypothesis as well, finding that the effect of context on the positive brand (t (117)=2.32; p<.05) is greater than its effect on the negative brand (t (117)=1.67; p=.09).

Study 2

In Study 2, a similar investigation examined the impact of context on purchase intention for positive and negative brands, hypothesizing, 1) negative and positive context would have an affect-congruent effect on purchase intention and, 2) the effect of negative context would be greater for a positive than a negative brand

To test these hypotheses, 20 participants viewed the same experimental video used in Study 1. After viewing the video, they indicated their purchase intention for each brand on a six-item scale. For example, "How much are you willing to pay for a Ben and Jerry’s T-shirt?"

To examine the effect of the positive and negative context on purchase intention, a 2 (brand: positive or negative) within x 2 (context: positive or negative) between-participant analysis of variance was calculated on a composite of the six-item scale. As hypothesized, a main effect of context found that purchase intention was lower when brands appeared in the negative context opposed to the positive one (F (1, 18)=6.10; p<.05). Consistent with the second hypothesis, follow-up t-tests found that the effect of context on purchase intention applied to the positive brand (t (18)=2.43; p<.05) but not the negative brand (t (18)=.32; p=.76).


It is clear from the studies above that context impacts brand evaluation and purchase intention. Despite the use of strongly valenced brands we found a robust effect of negative versus positive context. Moreover, these results suggest that the impact of negative context may be greater for positive brands. Practically, this result suggests that a brand manager for a positive brand may want to carefully consider placement of his/her brand in contexts with negative content (e.g. films, newspapers, and magazines). Theoretically, future research is needed to identify the mechanism underlying the impact of context on consumer perception and behavior.


Dijksterhuis, A.; Spears, R.; Postmes, T.; Stapel, D.; Koomen, W.; van Knippenberg, A.; Scheepers, D. (1998). Seeing one thing and doing another: Contrast effects in automatic behavior. Journal of Personality & Social Psychology, 75, 862-871.

Erber, R., Wegner, D. M., & Therriault, N. (1996). On being cool and collected: Mood regulation in anticipation of social interaction. Journal of Personality and Social Psychology, 70, 757-766.

Forgas, Joseph P. (1995) Mood and judgment: The affect infusion model (AIM). Psychological Bulletin. 117, 39-66

Puccinelli, N. M. and Moon, Y. (2001). Putting Your Best Face Forward: The Effects of Mood on Agent Preference. Working paper.