The Effect of Imagery Processing and Imagery Content on Behavioral Intentions

Paula Fitzgerald Bone, West Virginia University
Pam Scholder Ellen, Georgia State University
ABSTRACT - Imagery processing has been recognized as superior to discursive (or verbal) processing and thus has become an important tool in designing advertising. Evidence suggests that imagery-producing ads result in superior recall and more positive attitudes toward the product. However. there has been limited examination of the relationship between imagery and behavioral intentions. We investigate this latter relationship going beyond other research in the area by examining imagery content. Specifically, we report the results of an experiment which tests the effects of three message-relevant factors on behavioral intentions: (1) self-relatedness, (2) plausibility, and (3) distinctiveness.
[ to cite ]:
Paula Fitzgerald Bone and Pam Scholder Ellen (1990) ,"The Effect of Imagery Processing and Imagery Content on Behavioral Intentions", in NA - Advances in Consumer Research Volume 17, eds. Marvin E. Goldberg, Gerald Gorn, and Richard W. Pollay, Provo, UT : Association for Consumer Research, Pages: 449-454.

Advances in Consumer Research Volume 17, 1990      Pages 449-454


Paula Fitzgerald Bone, West Virginia University

Pam Scholder Ellen, Georgia State University


Imagery processing has been recognized as superior to discursive (or verbal) processing and thus has become an important tool in designing advertising. Evidence suggests that imagery-producing ads result in superior recall and more positive attitudes toward the product. However. there has been limited examination of the relationship between imagery and behavioral intentions. We investigate this latter relationship going beyond other research in the area by examining imagery content. Specifically, we report the results of an experiment which tests the effects of three message-relevant factors on behavioral intentions: (1) self-relatedness, (2) plausibility, and (3) distinctiveness.


Consumer researchers have long recognized the advantage of imagery processing over verbal processing. While imagery has traditionally been linked to visual stimuli, i.e., "thinking in pictures" (Fodor 1981; Lutz and Lutz 1978), MacInnis and Price (1987) point out that imagery processing can entail all of the senses. One can imagine the song of a bird, the crunch of a Dorito, the feel of sandpaper, and the taste of a lemon just as vividly as one can mentally picture these objects. Therefore, consistent with MacInnis and Price (1987), we define imagery as an information processing mode which uses one or more of the five senses.

Several studies have shown that imagery processing can improve recall and enhance attitude toward the brand (see Lutz and Lutz 1978 and MacInnis and Price 1987 for a review). There is even some evidence of a positive relationship between imagery processing and behavioral intentions (BI) (cf., Anderson 1983; Gregory, Cialdini, and Carpenter 1982). Yet, only Gregory, Cialdini, and Carpenter (1982) have shown this relationship to exist in a marketing context. In a personal selling situation, Gregory, et al. (1982) found that individuals who were asked to imagine themselves enjoying the benefits of cable TV had greater intentions to subscribe to the cable service than did individuals who were simply told about these benefits.

In this investigation, we empirically examine the relationship between imagery and BI in an advertising context. In addition, we identify and test the effects of several message factors which may differentially affect imagery's relation to BI.

Why should imagined sensory experiences result in more positive BI? Two prominent theories may explain these effects: (1) the availability/ valence hypothesis (Anderson 1983; Gregory, Cialdini, and Carpenter 1982; MacInnis and Price 1987; Mowen 1980; Tversky and Kahneman 1973) and (2) the semantic network theory (Mitchell 1983). Availability/valence hypothesis states that the information which is most available to an individual when making a judgment has the greatest impact on that individual's judgment. Many researchers (Anderson 1983; Gregory, Cialdini and Carpenter 1982; Mowen 1980) have hypothesized that imagery results in more concrete scenarios and greater information elaboration than does discursive processing (i.e., verbal processing). This, in turn, makes information "spring to mind" more easily when judgments are made. These judgments will be positively (negatively) influenced if the available information is favorable (unfavorable).

Semantic network theory (Mitchell 1983) augments availability/valence theory by considering specific elements of the imagined scene, rather than only the valence of the scene. According to this theory, an individual's long-term memory (LTM) is made up of links and nodes. Links represent relationships (positive or negative; strong or weak) and nodes represent concepts (e.g., products, persons, emotions, tastes, sounds). Imagery content may determine which concepts are linked together in LTM. For instance, an ad which asks a consumer to imagine his/her spouse driving a Volkswagen could create links between the consumer's spouse and the Volkswagen in LTM. This direct link may cause the consumer to perceive a greater chance that her/his spouse would purchase the Volkswagen.

It appears that combining availability/ valence theory and semantic network theory can aid us when examining the imagery/BI relationship. These theories are not presented as competing explanations of-imagery effects, but rather as theories that work hand-in-hand to help explain this potentially important relationship. Based on these two theories, we make the following proposition:

P1: Imagery processing will create greater Bl than discursive processing.

Additionally, it is likely that the imagery/BI relationship is affected by many message factors. We discuss a few of these variables in the following section.


Extant literature suggests several variables which may affect the imagery/BI relationship. We have classified these variables into two groups: (1) message elements that facilitate imagery processing, and (2) message elements that differentially affect the relationship between imagery and BI. The first group consists of message elements which researchers have used to "manipulate" imagery processing. These include the use of (1) pictures, (2) concrete words, and (3) instructions to imagine (Lutz and Lutz 1978; MacInnis and Price 1987).

The second variable set is expected to differentially affect BI when imagery processing occurs. In this study we examine three such variables: self-relatedness, plausibility, and distinctiveness. We chose to investigate these variables because several researchers have alluded to their importance (cf., Anderson 1983; Carroll 1978; Lutz and Lutz 1978; MacInnis and Price 1987) and because these variables have direct implications for designing marketing communications.


Self-relatedness deals with the focal character in the imagined scenario (Anderson 1983; MacInnis and Price 1987). A scenario is self-related when the main character is the consumer exposed to the ad. A scenario is other-related if the main character is not the consumer him/herself. Semantic network theory argues that direct links will be made between the self and the product advertised when the consumer is the subject of the imagery processing. This link should make the prospect of future interaction with the product seem more likely. If another person (e.g, friend, relative, celebrity) is the focal character of the imagery experience, there will be no direct link between the consumer and the product. Consequently, the relationship between imagery processing and BI may not be as strong.

Anderson (1983) found that self- vs. other-relatedness does indeed affect the relationship between imagery and BI. In a clinical context, he demonstrated that subjects asked to imagine and draw themselves in various scenarios formed BI consistent with the scenarios. However, no changes in BI were found when subjects were asked to draw scenarios involving other people. This leads us to the following proposition:

P2: Self-related imagery will create more positive Bl than other-related imagery.


The plausibility of the imagined scenario should affect imagery's relationship to BI (Anderson 1983; Carroll 1978). A plausible situation is one that is consistent with a consumer's lifestyle--a situation in which a consumer could reasonably expect to participate. BI should be greater for plausible situations since links are formed in LTM between the product and a realistic personal situation. Thus, a plausible context should enhance availability of the brand name and increase the consumer's perception that s/he is likely to purchase the product.

For example, an advertisement for a new line of elegant paper plates could ask an individual to imagine him/herself in a multimillion dollar home, using the paper plates while entertaining international dignitaries and movie stars--an implausible situation. Alternatively, it could ask him/her to imagine using the plates at home while entertaining a few close friends. The latter, more plausible situation, should create direct links in the consumer's LTM which relate the product to a personal situation, thus making future interaction with the product seem more likely. Therefore, we propose that:

P3: Imagery which contains plausible scenarios will create more positive Bl than imagery which contains implausible scenarios.


A final factor expected to affect the imagery BI relationship is situational distinctiveness (Lorayne and Lucas 1974; Lutz and Lutz 1978). Distinctiveness takes into account the imagined scenes' uniqueness. It is bounded at the lower end by mundane and at the upper end by bizarre.

Presenting a product in a mundane situation is unlikely to enhance the relationship between imagery and BI. When a consumer has numerous associations with a situation (which is highly possible with a common situation), the association between the brand and the situation may be drowned out and unavailable (Nisbett and Ross 1980). For example, a consumer asked to imagine him/herself waking up, getting dressed, and making a cup of Folgers coffee is likely to have many different images associated with this mundane event. It is possible that the image of drinking Folger's coffee may be drowned out by other images associated with getting ready for the day.

Evidence suggests that bizarre situations (ones with very high distinctiveness) positively influence advertisement and brand recall (Lutz and Lutz 1978); Yet, semantic network theory suggests that marketers should create links between the product and a situation which could actually occur in the consumer's life. This theory implies that a bizarre situation is unlikely to enhance the relationship between imagery and BI.

Note that plausibility and distinctiveness are inherently confounded since a bizarre product-usage scenario is almost certainly implausible whereas a mundane event would be quite plausible. Thus, these two factors, plausibility and distinctiveness will be considered together and lead to:

P4: Imagery which is plausible and moderately distinctive will create greater Bl than imagery which is plausible and indistinctive or implausible and highly distinctive.

In the next section of this paper we discuss an empirical test of the following hypotheses. The specific proposition(s) upon which each hypothesis is based is noted in parentheses.

H1. High-imagery ads will create greater BI than the low- imagery control ad. (Proposition 1 )

H2. High-imagery self-related ads will create greater BI than high-imagery other-related ads. (Proposition 2)

H3. High-imagery, moderately distinctive, and plausible ads will create greater BI than either high-imagery, indistinctive, and plausible ads or high-imagery, highly distinctive, and implausible ads. (Propositions 4)


Seven radio advertisements were created in order to investigate the effects of imagery, self-relatedness, plausibility, and distinctiveness on BI. Radio was selected as the experimental medium because self-generated imagery should have a greater effect than other-generated imagery (Lorayne and Lucas 1974; Lutz and Lutz 1978). If a consumer is forced to create his/her own images, the mental processing is at a deeper level than if the images are created for him/her.

Focus interviews indicated that popcorn could serve as the experimental product. Its consumption abounds with high sensory experiences (e.g., aroma, texture, taste, sound), and it is popular with the subject population (college students). A fictional brand name was used to control for familiarity with particular brands.

A 3 (plausibility/distinctiveness) by 2 (self/other-relatedness) design with a low-imagery control group was used in this experiment. The first factor incorporates several combinations of plausibility and distinctiveness to create three different situational scenarios. The independent manipulation of these two factors is impossible because they are inextricably confounded.

The first scenario was plausible and indistinctive. Subjects assigned to this treatment listened to an ad which asked them to imagine someone (self or other) lying on a couch, watching TV, and getting a craving for popcorn. The second scenario was still plausible and moderately distinctive. This ad asked subjects to imagine someone (self or other) cooking out on the beach while on vacation and getting a craving for popcorn. The final scenario was implausible and highly distinctive. In this scenario, subjects were asked to imagine someone (self or other) walking through a dark, dusty hall in a haunted house with Scooby-do and Shaggy, the cartoon characters. Pretesting showed that all scenarios were perceived as intended and that subjects held positive attitudes toward all three scenarios.

The self/other factor was manipulated by having students imagine themselves or an eccentric chemistry professor as the focal character in one of the three scenes described above. We chose someone very different from the student subjects so that they would not identify with the "other." Pretesting showed that subjects were imagining the intended character.

Fully crossing the two main characters with the three situational scenarios resulted in six high-imagery cells. All six ads used concrete words, actionable sentences, present tense, and instructions to imagine to facilitate imagery. Importantly, each ad described the focal person interacting with the product (Lutz and Lutz 1978). The following is a segment of the self-related, plausible, and indistinctive ad:

Imagine this. It's late Sunday night and you're lying on the couch watching TV. Your stomach is growling--you get a craving that only light, fluffy popcorn will satisfy. You go to the kitchen, turn on the stove and grab a shiny silver package of Pop-quick popcorn from the cabinet. You swirl the pan over the red-hot burner. Suddenly, you hear one kernel pop, then another and another until there is a thunderous drum-roll of bursting kernels. Imagine ripping open. the package--the steam dampens you face. You smell the unmistakable buttery aroma.

The other two situational scenarios used the same popcorn imagery but the first few sentences were changed to incorporate the different scene.

The introduction to the other-related, plausible, indistinctive ad shows how the self/other manipulation was incorporated into the ads:

Imagine this. It's late Sunday night and an eccentric chemistry professor is lying on his couch watching TV. His stomach is growling--he gets a craving that only light, fluffy popcorn will satisfy.

The seventh cell, the low-imagery control ad, differed significantly from the six high-imagery ads. This ad used passive tense, more complex sentence structure, more abstract terminology, and no instructions to imagine. In addition, there was no situational context and no focal person in the ad. Care was taken to ensure that the information given about the product was equivalent in all seven ads and that the brand name was used the same number of times. Pretesting showed that the control ad created less imagery than did the six experimental ads.


College students received extra credit in an introductory marketing course for participating in the experiment. Each of the 25 experimental sessions was randomly assigned to one of the seven advertisements. One hundred and forty-two students participated with cell sizes ranging from 17 to 23.

After completing consent forms, subjects were told that they were to evaluate three different radio announcers. This guise was used to account for the three repetitions of the same ad needed for sufficient exposure to the ad. Subjects then listened to one of the seven experimental ads recorded by three different professional male radio announcers (i.e., three repetitions of the same ad). A questionnaire booklet was distributed and subjects completed questions about the three radio announcers to lend credibility to the guise. Subjects completed imagery, distinctiveness, plausibility, and self-relatedness manipulation checks, then completed BI items. Attitude toward the ad and attitude toward purchasing the brand measures were also taken. Approximately one week later, subjects were orally debriefed.


Our study differs significantly from previous marketing investigations of imagery in that imagery processing was measured, rather than assumed to occur. Eight items were used to measure imagery processing (three items based on those used by Cartwright, Markes, and Durrett 1978; five items developed for this study). Prior to responding to these eight items, subjects were provided a clear definition of imagery. The first imagery item was a -simple yes/no response to, "Did you experience any imagery?" If the subject did, s/he then completed a seven-item imagery scale. The first item stated, "The imagery from the ad was aroused" and was anchored with "great difficulty" and "with great ease." The second stated, "The imagery from the advertisement was" and was anchored with "not at all vivid, a very vague image almost like nothing at all," and "very vivid, image was almost like a real experience." The final five items were seven-point Likert-type items such as, "I had no difficulty imagining the scene in my head," and "All sorts of pictures, sounds and smells came to my mind while I listened to the ad." Coefficient alpha for the seven items was 0.87.

Several items were used to check the message manipulations. The self-relatedness measure was an open-ended question asking, "Who was the main character(s) in the scene that YOU imagined?" Plausibility was measured on two correlated (r=0.82) seven-point items which asked subjects whether the situation in the ad could really happen and how likely it was that a typical person might find him/herself in that situation. Distinctiveness was measured as the sum of two moderately correlated (r=0.44) items. The first item asked whether the situation in the ad was an ordinary event, and the second asked whether the situation was a special event.

Attitude toward the ad was measured using the sum of seven four-point items (alpha=0.90). Subjects stated the degree to which the ad made them feel good, happy, cheerful, pleased, amused, soothed, and warm. Attitude toward purchasing the brand was measured as the sum of six semantic differential items similar to those used by Ajzen and Fishbein (1980; alpha 0.82).

Three items were standardized and summed to measure BI (alpha=0.8( ). First, an eleven-point item asked subjects to estimate the probability that they would purchase the advertised brand. Scale responses ranged from "virtually certain, (99 in 100)" to "no chance, (0-in 100)." Subjects indicated the likelihood of their purchasing the advertised brand the next time they bought popcorn using a five-point item anchored with "definitely will buy" to "definitely will not buy." The final item was a seven-point Likert-type item stating, 'The next time I purchase popcorn, I will buy the brand in the advertisement."

Manipulation Checks

The three different scenario manipulations were effective. One-way ANOVAs and individual contrasts indicated significant differences in the predicted direction for both the plausibility and distinctiveness manipulations.

Only subjects who experienced some imagery processing responded to the self-relatedness manipulation check. On the surface, the self-relatedness manipulation appears effective (chi-square=12.25, p<0.01); however, over half (29 out of 49) the individuals in the self-related treatment did not imagine themselves during the ad. The manipulation was not as strong as we would have liked.

The six high-imagery groups did not experience more imagery than the low-imagery control group. The yes/no item showed no significant differences (chi-square=9.60, p=0.14) nor did the ANOVA for the seven-item scale (F=l.89, p=0.09). Individual contrasts showed no differences between the low-imagery control group and any of the six high-imagery groups. Possible explanations for this result are provided in the discussion.

While the imagery and self-related manipulation checks indicated that these treatments did not create the desired effects in each cell, differences across individuals did occur. Therefore, rather than conducting the hypothesis tests using the experimental cells as planned, we used an internal analysis to explore these hypotheses (Carlsmith, Ellsworth, and Aronson 1976). In an internal analysis, individuals are grouped by their actual responses to the treatment as indicated by manipulation checks rather than by their cell assignment. While this procedure reduces the ability to make causal statements, it does allow researchers to explore some basic relationships.

Hypotheses Tests Using Internal Analysis

H1 was based on the proposition that imagery processing will create greater BI than discursive processing (P1). To test this relationship, subjects were divided into three imagery processing groups based on their responses to the imagery manipulation checks. Subjects reporting no imagery evoked by the ad were assigned to the no-imagery group. Those experiencing some imagery, but scoring below the mean of the summed imagery scale (M=35.24, SD=7.61) were assigned to the low-imagery group, and subjects scoring above the mean were assigned to the high-imagery group.

A one-way ANOVA was used to analyze the results, with imagery group serving as the independent variable and BI serving as the dependent variable. The ANOVA showed a significant difference between the three groups (F=6.78, P<0-01. Mhigh imagery 0.76 Mlow-imagery= - 0-88, Mno-imagery=-l 05, omega-squared=o.lo). These results support H1; however, it appears that the degree of imagery processing (e.g., high vs. low) facilitates the effect rather than the simple presence or absence of imagery.

H2 was based on Proposition 2 which states that self-related imagery creates more positive BI than other-related imagery. Only subjects experiencing some imagery were included in this analysis. Subjects who indicated that they had imagined themselves were placed in the self-related group; all remaining subjects were assigned to the other-related group. A one-way ANOVA in which BI served as the dependent variable and self/other-relatedness served as the independent variable showed significant differences between the two groups (F=9.84, p<0.01, Mself-related=1 539 Mother-related=-0-34. omega-squared=0.08). This finding clearly supports H2.

H3 was based on Proposition 4 which states that the plausibility and distinctiveness of the imagined scenario should affect the imagery/BI relationship. We explored this hypothesis using the three scenario manipulations since these manipulations were effective; however, we only included subjects who experienced imagery in the analysis.

A one-way ANOVA in which BI served as the dependent variable and the three scenario manipulations served as the independent variable showed no statistically significant differences (F=0.16, p=0.85, Mtv scenario=033, Mbeach scenario= 0.05, Mhaunted house scenario=-0.04). Thus, H3 is not supported.

After finding that imagery and self/other-relatedness did affect BI, we investigated these two factors further. First, we examined whether the effect of imagery on BI was direct or indirect, i.e., a result of imagery's effect on attitudinal variables. It could be that imagery creates more positive attitudes toward the ad or attitudes toward the product, which, in turn, creates BI effects (cf., Ajzen and Fishbein 1980). A MANOVA model which used attitude toward the ad, attitude toward the product, and BI as the dependent variables and the three imagery groups as the independent variable tested the issue at hand. The overall MANOVA model was significant (F=6.98, p<0.01), and the univariate F-tests showed significant effects for each dependent variable. A Roy-Bargman stepdown F test was used to assess whether the observed BI effects were direct or indirect. The stepdown F takes into account the causal relationship between the dependent variables and tests for an incremental effect on each dependent variable by controlling for the causal effects of previous dependent variables. This test detected differences in attitude toward the ad, but not attitude toward the brand or BI. Therefore it appears that the observed imagery effect on BI is indirect, a result of imagery's impact on attitude toward the ad.

Could this also be true for self/other relatedness? Could it be that consumers like ads about themselves better than ads about others and that this change in attitude toward the ad creates differences in BI? We explored this question with a second MANOVA model in which attitude toward the ad, attitude toward the brand, and BI were the independent variables and self/ other-relatedness served as the dependent variable. Again, the overall MANOVA model was significant (F=3.47, p=0.02), but the only univariate test which showed significant differences was the BI test. The stepdown F test also detected a difference in BI, but not in the attitude variables. This suggests that a direct relationship exists between the imagery's focal character and BI, as predicted by semantic network theory.


This study provides evidence that imagery processing in and of itself can affect attitudes toward the advertisement which, in turn, affect attitude toward the brand and B1. Furthermore, the significant effect of the self/other factor suggests that the content of the imagined experience is important and can directly influence a consumer's BI. While these results are correlational due to the use of internal analysis, they are encouraging and indicate that research in the area should continue.

Why didn't subjects in the six high-imagery cells experience more imagery than subjects in the low-imagery control cell? Recall that we did get significant differences in pretesting, so the experimental situation itself may have reduced imagery processing. Subjects in the experiment (but not the pretest) were asked to listen to the ads in order to evaluate the radio announcers. MacInnis and Price (1987) point out that imagery processing is enhanced when subjects are told to relax and concentrate on imaging. Our subjects processing set was on evaluating the announcers which may have interfered with imagery generation.

There are many directions in which this research could proceed. First, richer marketing-oriented measures of ad-evoked imagery need to be developed. Recall that this study is the first to specifically measure ad-evoked imagery, rather than the propensity to use imagery (e.g., Style of Processing Questionnaire; Childers, Houston and Heckler 1985) or the ability to imagine (e.g., Betts Questionnaire Upon Mental Imagery; Betts 1909; Sheehan 1967). A reliable and valid measure is needed. Another measurement issue is whether imagery is a unidimensional or multidimensional construct. We are currently developing items to tap three potential imagery components--ease, vividness, and quantity. Finally, given the encouraging results from the self-relatedness factor, it appears that consumer researchers may wish to investigate other message characteristics which could directly or indirectly affect recall, attitudes, and/or BI when imagery processing occurs.


Ajzen, Icek and Martin Fishbein (1980), Understanding Attitudes and Predicting Social Behavior, Englewood Cliffs, NJ: Prentice-Hall, Inc.

Anderson, Craig A. (1983), "Imagination and Expectation: The Effect of Imagining Behavioral Scripts on Personal Intentions," Journal of Personality and Social Psychology, 45, 2, 293305.

Betts, G. H. (1901), The Distribution and Functions of Mental Imagery, Teachers College, Columbia University, N. Y., NY.

Carlsmith, J. Merrill, Phoebe C. Ellsworth and Elliot Aronson (1976), Methods of Research in Social Psychology, Addison-Wesley Publishing Company, Reading, Mass.

Carroll, John (1978), "The Effect of Imagining an Event on Expectations for the Event: An Interpretation in Terms of the Availability Heuristic," Journal of Experimental Social Psychology, 14, 88-96.

Cartwright, Desmond S., Mary E. Marks, and John H. Durrett, Jr. (1978), ''Definition and Measurement of Three Processes of Imagery Representation: Exploratory Studies of Verbally Stimulated Imagery," Multivariate Behavioral Research, 13, (October) 449-473.

Childers, T., M. Houston and S. Heckler (1985), "Measurement of Individual Differences in Visual vs. Verbal Information Processing," Journal of Consumer Research, 12 (September), 125-134.

Fodor, Jerry A. (1981), "Imagistic Representation," in Imagery, ed. Ned Block, Cambridge, Mass: MIT Press, 63-86.

Greenwald, Anthony A. and Clark Leavitt (1984), "Audience Involvement in Advertising: Four Levels," Journal of Consumer Research, 12 - (June), 581 -592.

Gregory, W. Larry, Robert D. Cialdini, and Kathleen M. Carpenter (1982), "Self-Relevant Scenarios as Mediators of Likelihood Estimates and Compliance: Does Imagining Make It So," Journal of Personality and Social Psychology, 43, 1, 89-99.

Lorayne, Harry and Jerry Lucas (1974), The Memory Book, New York: Stein and Day Publishers.

Lutz, Kathy and Richard J. Lutz (1978), "Imagery-Eliciting Strategies: Review and Implications of Research," Advances in Consumer Research, 9, 611 -620.

MacInnis, Deborah J. and Linda L. Price (1987), "The Role of Imagery in Information Processing: Review and Extensions," Journal of Consumer Research, 13 (March), 473-91.

Mitchell, Anderson A. (1983), "Cognitive Processes Initiated by Advertising," in R.J. Harris (ed.) Information Processing Research in Advertising, Hillsdale, NJ: Lawrence Erlbaum Associates, 13-42.

Mowen, John C. (1980), 'The Availability Heuristic: The Effect of Imagining the Use of a Product on Product Perceptions," 1980 Educators' Conference Proceedings, eds. Richard Bagozzi, Kenneth Bernhardt, Paul Busch, David Cravens, Joseph Hair, Jr., Carol Scott, Chicago, IL: American Marketing Association 140-142.

Nisbett, Richard and Lee Ross (1980), Human Inference: Strategies and Shortcomings of Social Judgment, Prentice-Hall, Inc., Englewood Cliffs, New Jersey.

Sheehan, Peter (1967), "A Shortened Form of Betts' Questionnaire Upon Mental Imagery," Journal of Clinical Psychology, 23, 386-389.

Sherman, Steven, Kim S. Zehmer, James Johnson, and Edward R. Hirt (1983), "Social Explanation: The Role of Timing, Set and Recall on Subjective Likelihood Estimates," Journal of Personality and Social Psychology, 44, 1127-43.

Tversky, Amos and Daniel Kahneman (1973), "Availability: A Heuristic for Judging Frequency and Probability," Cognitive Psychology, 5(2), 207 -32.