Extending Two Cognitive Processing Scalesbneed For Cognition and Need For Evaluationbfor Use in a Health Intervention

ABSTRACT - The present study builds on the works of Cacioppo and Petty (1982) and Jarvis and Petty (1996). They posit that individuals have different cognitive processing styles and that these individual differences can be reliably measured with the Need for Cognition Scale (18 items) and the Need for Evaluation Scale (16 items). This paper presents research that reduces the two scales from a total of 34 to 18 items. Preliminary results support single factor structures for both reduced scales. Suggestions for future scale development research and implications for a health promotion intervention study are presented.

Citation:

Michael Sherrard and Ronald Czaja (1999) ,"Extending Two Cognitive Processing Scalesbneed For Cognition and Need For Evaluationbfor Use in a Health Intervention", in E - European Advances in Consumer Research Volume 4, eds. Bernard Dubois, Tina M. Lowrey, and L. J. Shrum, Marc Vanhuele, Provo, UT : Association for Consumer Research, Pages: 135-142.

European Advances in Consumer Research Volume 4, 1999      Pages 135-142

EXTENDING TWO COGNITIVE PROCESSING SCALESBNEED FOR COGNITION AND NEED FOR EVALUATIONBFOR USE IN A HEALTH INTERVENTION

Michael Sherrard, University of Auckland, New Zealand

Ronald Czaja, North Carolina State University, U.S.A.

ABSTRACT -

The present study builds on the works of Cacioppo and Petty (1982) and Jarvis and Petty (1996). They posit that individuals have different cognitive processing styles and that these individual differences can be reliably measured with the Need for Cognition Scale (18 items) and the Need for Evaluation Scale (16 items). This paper presents research that reduces the two scales from a total of 34 to 18 items. Preliminary results support single factor structures for both reduced scales. Suggestions for future scale development research and implications for a health promotion intervention study are presented.

INTRODUCTION

This study, an early part of a larger research effort, set out to reduce the number of items on the Need for Cognition (Cacioppo and Petty 1982) and the Need for Evaluation (Jarvis and Petty 1996) scales. The overall health promotion intervention research that this study precedes involves administering the two scales in a medical clinic environment. In the clinics it is uncertain who will administer the scales and how much time will be available for this task. It is reasoned that the shorter the combined Need for Cognition (NFC) and Need for Evaluation (NFE) questionnaire, the greater the ease of administering the total instrument in the health clinic environment.

NEED FOR COGNITION CONSTRUCT

Need for cognition (NFC) is defined as the "tendency to engage in and enjoy thinking" (Cacioppo and Petty 1982, p116) and has been well established in the communication and persuasion literature. A sizeable empirical base developed over the past 16 years supports the need for cognition scale. The construct can be reliably measured, is stable over time and situations, and can account for significant variance in information processing beyond that explained by intellectual ability (for a summary see, Cacioppo, Petty, Feinstein, and Jarvis 1996)

A review of the NFC literature shows that researchers have concentrated on either developing a short form of the 34-item scale developed in the original study (Cacioppo and Petty 1982) or investigating the number of required sub-scales (Cacioppo, Petty, Feinstein and Jarvis 1996).

Factor Structure and Reliability of the NFC Scale

Originally, Cacioppo and Petty (1982) developed a 34-item scale to measure the need for cognition, refined from a possible total of 45 items. This was later shortened to a more efficient 18-item scale (Cacioppo, Petty and Kao 1984) that was highly correlated and of equal reliability to the 34-item scale (r=0.95, p<0.001, alpha=0.90 for the 18-item scale and alpha=0.91 for the 34-item scale). The factor analysis to date demonstrates that inter-individual difference in the engagement and enjoyment of effortful cognitive activities is represented by one dominant factor. The dominant one-factor structure and reliability of the 18-item scale has been replicated (Sadowski and Gulgoz 1992), translated into Dutch (for example, Verplanken, Hazenberg and Palenewen 1992), and administered to different types of populations (Cacioppo, Petty, Feinstein and Jarvis 1996).

Three studies by Chaiken (1987) investigated if high and low NFC loaded onto separate factors. The results supported the single factor structure of the NFC and showed that individuals high in NFC prefer effortful information processing while low NFC individuals prefer heuristic information processing (Cacioppo, Petty, Feinstein and Jarvis 1996).

Replication studies of the 34-item and the 18-item NFC scale have produced similar levels of reliability to the original scales (majority of results, alpha>0.85). The reliability of the scale is demonstrated by the correlations in two test-retest studies of 0.88 (p<0.001; Sadowski and Gulgoz 1992) and 0.66 (p<0.001; Verplanken 1991). Studies using the NFC scale items have not always maintained the original items and some researchers have selected a reduced number of items to be used to measure the NFC construct (Cacioppo, Petty, Feinstein and Jarvis 1996). The 34-item scale was reduced to 9 items in a mail survey (Manfredo and Bright 1991) resulting in a reliability score of alpha=0.76 and reduced to 15 items (Wolfe and Grosche 1990) with a reliability score of alpha=0.88. The 18-item NFC scale that was translated into Dutch was reduced to 11 items (Verplanken 1991) with a reliability score of alpha=0.74 and 6 items were used twice (Verplanken et al. 1992) producing reliability scores of alpha=0.82 and 0.85.

Relationship of the NFC to Other Individual Differences

Numerous studies have shown positive relationships with the NFC scale, although, other than intelligence (Cacioppo and Petty 1982), none have shown more than a moderate correlation. For example, scales measuring devotion to ongoing cognitive tasks had a r=0.37, p<0.01 (Cacioppo, Petty, Feinstein and Jarvis 1996); perception of personal relevance or involvement had a r=23, p<.05 (Thompson and Zanna 1995); and values had a r=0.29, p<0.05 (Berzonsky and Sullivan 1992).

A number of studies have shown that the NFC scale is not related to other individual difference variables. These studies show that the NFC is negatively related to dogmatism (r=-0.23 to -0.24, p<0.05; Cacioppo and Petty), close-mindedness (meta-analysis r_ave=-0.34, p<0.01, where r_ave is the average correlation index; Petty and Jarvis 1996), and a need for closure (meta-analysis r_ave=-0.26, p<0.01; Petty and Jarvis 1996). These studies offer support that the NFC scale is a unique, unidimensional measure (Cacioppo, Petty, Feinstein and Jarvis 1996, Petty and Jarvis 1996, Thompson and Zanna 1995).

NFC Characteristics

Many studies have also investigated characteristics or behaviour of individuals who are high compared to low on NFC. The three main areas of study are (1) tendency to engage in effortful cognitive activity, (2) perception of level of cognitive activity, and (3) potential bias in information processing (Cacioppo, Petty, Feinstein and Jarvis 1996).

Tendency to engage in effortful cognitive activity. Individuals who are classified as high in their need for cognition compared to those classified as low, have been shown to: recall more information with which they have been supplied, search for more information to process, be more responsive to argument quality and less responsive to heuristic-systematic cues, generate more task relevant thoughts, have a higher correlation between their judgement and thoughts and beliefs, possess more knowledge, perform better on general cognitive tasks, have a stronger link between thoughts and attitudes over time, and manifest extreme attitudes (Cacioppo, Petty, Feinstein and Jarvis 1996, Petty and Jarvis 1996, Verplanken, Hazenberg and Palenewen, 1992, Manfredo and Bright 1991).

Perception of the level of cognitive effort. High NFC individuals perceive that they put forth more cognitive effort than low NFC individuals on specific cognitive tasks.

Potential bias in information processing. Three types of bias have been found to influence high and low NFC individuals in their information processing. First, mood has an effect on a high NFC individual’s cognition while it may affect both high and low NFC individuals’ attitudes. Second, priming of cognitive material may affect the response of high NFC individuals, however, if two messages are presented to a high NFC individual, then primacy effects may influence his or her choice based on the perceived stronger message argument. Last, low NFC individuals are more likely to be influenced by recency effects in information processing (for a full summary see, Cacioppo, Petty, Feinstein and Jarvis 1996, Petty and Jarvis 1996).

NEED FOR EVALUATION CONSTRUCT

The Need for Evaluation (NFE) is defined as "the extent to which people naturally tend to engage in evaluation, that is, consider what is good versus bad about various people, objects, and issues" (Petty and Jarvis 1996, p222). Jarvis and Petty (1996) have shown that, similar to the NFC variable, the NFE variable can be reliably measured, is stable over both time and situations, and is significantly different from oher individual difference variables. This last finding contradicts common beliefs that evaluative responding is pervasive and closely tied to cognition. Although the NFE construct is relatively new and untested compared to the NFC, Jarvis and Petty (1996) have suggested that it is an influential and a complementary individual difference variable to the NFC.

Factor Structure and Reliability of the NFE Scale

Five studies to date have been undertaken on scaling individual differences in the NFE (Jarvis and Petty 1996). The factor analysis in the scale development process shows that inter-individual differences in evaluative activity are represented by one dominant factor. In addition, the 16-item NFE scale, refined from an original pool of 46 items, is shown to be highly reliable (alpha=0.83 to 0.87).

Relationship of the NFE to Other Individual Differences

The NFE was predicted to be related to existing individual difference scales. For example, theorists have commented that individuals who are high in NFE make more global evaluations than individuals low in NFE and this gives them greater control over their environment (Jarvis and Petty 1996). One hypothesized relationship is between the NFE and the Desire for Control Scales (Burger and Cooper 1979). The correlation was significant but relatively small (r=0.22, p<.05), suggesting that the motivation for control may only be one dimension of the NFE. Jarvis and Petty (1996) also demonstrated low or no significant correlations between the NFE and four other individual differences: motivation to attain structure and knowledge, self-monitoring, affect intensity, and social desirability. One scale that the NFE was expected to be significantly related to is the NFC. Jarvis and Petty (1996) found a moderate and positive relationship (r=0.35, p<.001). The authors suggest that the NFE is partly based on differences in people’s motivation to think and that evaluation does not require continuous effortful thought (Petty and Jarvis 1996, Jarvis and Petty 1996). These studies have suggested two characteristics of individuals that distinguish between those who are high versus low in NFE: (1) the amount and time of attitude retrieval from memory, and (2) the amount of opinions offered by an individual (Petty and Jarvis 1996).

The aims of the current research are threefold: 1) to provide another test of support for the unidimensionality of the NFC and NFE constructs, while 2) reducing the number of scale items to make them easier to administer in a health clinic environment, and 3) maintaining the reliability and validity of the reduced scales.

METHOD

Study Participants

A self-administered questionnaire containing both the NFC and the NFE scales was administered to 111 primarily undergraduate university and technical institute students majoring in arts, business, or social science subjects. Four subjects did not complete the instrument and were excluded from the analysis, thus, the results consist of 107 respondents.

As in the original study of the 16-item NFE scale (Jarvis and Petty 1996) and the short form 18-item NFC scale (Cacioppo, Petty, Feinstein and Jarvis 1996), subjects’ responses in our study were recorded on a 5 point Likert scale ranging from 1=extremely uncharacteristic to 5=extremely characteristic. To control for response acquiescence, 9 items in the NFC scale and 6 items in the NFE scale were reverse-keyed. Participants were classified as high or low in NFC and NFE based on whether their summed score was above or below the median for that scale.

The first stage goal was to replicate the original 18-item NFC (Cacioppo, Petty and Kao 1984) and the 16-iem NFE scale studies (Jarvis and Petty 1996). The analysis of the scale responses included principal components factor analysis, Cronbach’s alpha, item-total correlations, item mean and standard deviation comparisons across studies, and an assessment of validity (Jarvis and Petty 1996, Churchill 1979). A confirmatory factor analysis was also undertaken on the reduced NFC and NFE scales using AmosBVersion 3.61 (Arbuckle 1997). This procedure complements the traditional methods of scale development and provides a more explicit test of unidimensionality (Gerbing and Anderson 1988). We discuss first the process of reducing the number of items used to measure the two constructs.

Assessing Content Validity

Content validity was assessed to ensure measurement of the two individual difference constructs as defined by Churchill (1995) and Zaichkowsky (1985). Two judges who are well versed with the NFC and NFE literature independently grouped the items into categories. The categories for the NFC items were: (1) need for puzzle solving, (2) thought avoidance tasks, (3) enjoy thinking, (4) don’t enjoy thinking, and (5) enjoy important thinking tasks. The NFE items were grouped into: (1) strong opinions, (2) having opinions, (3) evaluate as good or bad, and (4) indifference.

To further assess content validity, two reviews of the items were undertaken. First, the two full scales were administered to 24 graduate students who are not part of the data analyzed in this paper. The students were asked to write comments on the meaning of each item and the appropriateness of the items to measure the two constructs. Concurrently, a total of five judges, including the principal investigators, commented on the items’ meaning and appropriateness of the wording. Comments from both reviews were used to choose the items that were retained (retained and deleted items are shown in Tables 1 and 3). Additionally, a number of wording modifications were suggested. The reviewers noted that some NFE items and reverse NFC items could not be answered accurately because they felt the items were too extreme. For example, the reviewers commented that the NFE item, "I enjoy strongly liking and disliking new things", and the NFC reverse item, "I only think as hard as I have to", may be perceived as too negative and socially undesirable.

TABLE 1

9 ITEM SCALE AND DELETED ITEMS FROM 18-ITEM NFC SCALE

RESULTS

Need for Cognition Measure

The retained and deleted items, corresponding factor loadings, and item to total scale correlations of the 18-item NFC scale are presented in Table 1.

Factor analysis. The principal components analysis revealed a single dominant factor for both the 18-item and 9-item NFC scales. The first factor of the 18-item NFC scale accounted for 25.5% of the variance while the second factor accounted for 10.5% of the variance. A total of 15 items with factor loadings 0.35 or greater loaded on the first factor of the 18-item NFC scale. The first factor of the 9-item NFC scale accounted for 39.5% of the variance while the second factor accounted for 9.5% of the variance. The percentage of variance attributable to the first factor in the reduced scale is greater than for the first factor in the 18-item scale. Factor loadings for the first dominant factor of the 9-item NFC scale were of moderate to high strength: absolute values range from 0.55 to 0.74. The Scree test shown in Figure 1 supports the retention of one factor for the 9-item scale. Hence, the longer 18-item scale does not appear more efficient with our sample.

Scale reliability. The 18-item and the reduced 9-item NFC scales were both subjected to reliability analysis. The 18-item scale had a Cronbach’s alpha of 0.81 while the 9-item scale had a Cronbach’s alpha of 0.80. This result suggsts that in using the reduced scale little is sacrificed in terms of the reliability.

Scale correlation and convergent validity. If the first factor extracted from the data in both the 18-item and 9-item NFC scales were the same construct, then there should be a high degree of correspondence in the factor loadings between scales. Table 2 shows the correlation between the scales is 0.70 (p<0.001), suggesting a conceptual redundancy of the 18-item scale with the 9-item scale. This result also establishes convergent validity of the reduced scale.

FIGURE 1

9-ITEM NFC SCREE PLOT

TABLE 2

CORRELATION BETWEEN 18-ITEM NFC AND 9-ITEM NFC

Need Evaluation Measure

The retained and deleted items, corresponding factor loadings and item to total scale correlations of the 16-item NFE scale are presented in Table 3.

Factor analysis. The principal components analysis revealed a single dominant factor for both the 16-item and 9-item NFE scales. The first factor of the 16-item NFE scale accounted for 25.1% of the variance while the second factor accounted for 12% of the variance. Factor loadings for the first dominant factor were of low-moderate to high strength for the 16-item NFE scale: absolute values range from 0.32 to 0.74. The first factor of the 9-item NFE scale accounted for 35% of the variance while the second factor accounted for 13% of the variance. The variance attributable to the first factor in the reduced scale is greater than the variance associated with the first factor in the 16-item scale. Factor loadings for the first dominant factor of the 9-item NFE scale were of at least moderate strength: absolute values range from 0.43 to 0.74. The Scree test shown in Figure 2 supports the retention of one factor for the 9-item scale. This result suggests that the 9-item NFE scale may be more efficient than the longer scale.

Scale reliability. The 16-item scale had a Cronbach’s alpha of 0.79 while the 9-item scale had a Cronbach’s alpha of 0.76. Similar to the NFC reduced scale, this result demonstrates that little is sacrificed in terms of the reliability with the 9-item scale.

Scale correlation and convergent validity. The 9-item NFE first factor scores correlated highly and significantly to the original 16-item NFE first factor scores. Table 4 indicates a correlation of 0.74 (p<0.001), again suggesting a conceptual redundancy between the original 16-item scale and the reduced 9-item scale. This result also shows convergent validity of the reduced scale.

NFE scale predictive validity. The NFE is a newer construct than the NFC and therefore it is important to test for predictive validity. Past research by Jarvis and Petty (see Jarvis and Petty 1996, Petty and Jarvis 1996) has shown three related measures of a need for evaluation are (1) level of message retrieval from memory, (2) level of agreement with authority spokespersons on incorrect statements, and (3) level of opinions on fictitious or unknown statements. Individuals who have high need for evaluation retrieve more information from memory. This suggests that they rely on information retrieved from memory and are not as persuaded by fictitious statements by authority spokespersons. Below is the statement that incorporated these measures of need for evaluation in the present study. Participants were asked to state if they agreed with this statement, disagreed, or did not know.

Recently, Dr Hatherwaite from the British College of Physicians, stated that, "the importance of having a diet high in fiber and low in fat is exaggerated."

Dr Hatherwaite believes that studies that link fiber and fat levels in diet to colon cancer and heart disease, respectively, are inconclusive.

Table 5 illustrates that a significant majority of participants who are high in need for evaluation disagreed with this fictitious health message, while participants low in need for evaluation either agreed or offered no opinion. This result supports the predictive validity of the original 16-item NFE scaleand the reduced 9-item scale.

TABLE 3

9 ITEM SCALE AND DELETED ITEMS FROM 16-ITEM NFE SCALE

FIGURE 2

9-ITEM NFE SCREE PLOT

TABLE 4

CORRELATION BETWEEN 16-ITEM NFE AND 9-ITEM NFE

TABLE 5

RESPONSES TO FICTITIOUS HEALTH STATEMENT

TABLE 6

CORRELATION BETWEEN THE NFC AND NFE SCALES

TABLE 7

9-ITEM NFC AND 9-ITEM NFE CHI-SQUARE (X2), DEGREES OF FREDOM (df), PROBABILITY LEVEL (p), GOODNESS-OF-FIT INDEX (GFI), ADJUSTED GOODNESS-OF-FIT INDEX (AGFI), AND THE COMPARATIVE FIT INDEX (SPI).

Additional Tests

NFC and NFE reduced scale discriminant validity. As noted by Jarvis and Petty (1996) "evaluation by no means requires effortful thought" (p182). This assertion indicates that the two constructs, NFC and NFE, should not correlate highly. Table 6 shows that there is no significant correlation between the reduced NFC and the reduced NFE scales, thus offering evidence of discriminant validity.

Preliminary results of a confirmatory factor analysis. The exploratory factor analysis and reliability analysis supports the position that a single dominant factor is responsible for a large percentage of the variance in each scale. To more rigorously test the hypothesis that the reduced NFC and NFE scales are unidimensional, a confirmatory factor analysis was undertaken using AmosBVersion 3.61 (Arbuckle 1997, Baumgartner and Homburg 1996). Table 7 shows that our goodness of fit measures indicate fair to good results that are consistent with the exploratory analysis. The 9-item NFC scale model has the best fit with a Chi-square value of 29.10 (df=27, p=0.35). Other fit indices support the one factor solution of the reduced NFC scale: Goodness-of-Fit Index (GFI)=0.94, the Adjusted-Goodness-of-Fit Index (AGFI)=0.90, Comparative Fit Index (CFI)=0.99). The Chi-square value for the 9-item NFE scale indicates that our model is less than perfect (x²=44.49, df=27, p=0.02,) but the GFI=0.912, the AGFI=0.854, and the CFI=0.89. These latter measures of fit contrast with the Chi-square result and indicate a reasonable model.

CONCLUSION AND SUGGESTIONS FOR FUTURE RESEARCH

This study set out to reduce the number of items in the NFC and the NFE for use in a health promotion intervention. The two scales were reliably reduced to 9 items each and remained valid and representative of the original constructs. The new reduced scales will be more manageable than the original 34-item combined scales and this will be beneficial in the planned health clinic application. Further, the positive predictive validity test result of the 9-item NFE scale, using a health based question, is a preliminary confirmation of the use of the reduced scales in health interventions.

The confirmatory factor analysis generally supports the single factor structure of the two new reduced scales, but the results suggest a need for further refinement and testing to assure unidimensionality. One direction for future work would be to reexamine the scale item wording using cognitive pretesting methods that have been developed for and applied to survey questions. The present study suggests that some items were too extreme or ambiguous to truly measure latent individual characteristics.

The present study supports Jarvis and Petty’s (1996) exploratory factor analysis findings of a weak but positive correlation between the NFC and the NFE constructs. Future research needs to examine if these two individual difference constructs are subdimensions of a higher-order cognitive processing construct.

REFERENCES

Arbuckle, James L. (1997), Amos Users’ Guide Version 3.6. Chicago: SmallWaters Corporation.

Baumgartner, Hans and Christian Homburg (1996), "Applications of Structural Equation Modelling in Marketing and Consumer Research: A Review," International Journal of Research in Mareting, 13, 139-161.

Berzonsky, Michael D. and C. Sullivan (1992), "Social-cognitive Aspects of Identity Style: Need for Cognition, Experiential Openness, and Introspection," Journal of Adolescent Research, 7, 140-155.

Burger, J.M. and H.M. Cooper (1979), "The Desirability for Control," Motivation and Emotion, 3, 381-393.

Cacioppo, John T. and Richard E. Petty (1982), "The Need for Cognition," Journal of Personality and Social Psychology, 42 (1), 116-131.

Cacioppo, John T., Richard E Petty, J.A. Feinstein, and W.Blair G. Jarvis (1996), "Dispositional Differences in Cognitive Motivation: The Life and Times of Individuals Varying in Need for Cognition," Psychological Bulletin, 119 (2), 197-253.

Cacioppo, John T., Richard E. Petty, and Chuan F. Kao (1984), "The Efficient Assessment of Need for Cognition," Journal of Personality Assessment, 48, 306-307.

Celsi, Richard L. and Jerry C. Olson (1988), "The Role of Involvement in Attention and Comprehension Processes," Journal of Consumer Research, 15, 210-224.

Chaiken, Shelly (1987), "The Heuristic Model of Persuasion," in M.P. Zanna, J.M. Olson, and C.P. Herman (Eds.), Social Influence: The Ontario Symposium, 5, 3-39, Hillsdale, NJ: Erlbaum.

Churchill, Gilbert A. (1995), Marketing Research: Methodological Foundations (6th Edition). Fort Worth: The Dryden Press.

Dawes, John, Margaret Faulkner, and Byron Sharp (1998), "Business Orientation Scales: Development and Psychometric Assessment," Proceedings of the 27th European Marketing Academy Conference (EMAC), 5. 461-478.

Gerbing, David H. and James Anderson (1988), "An Updated Paradigm for Scale Development Incorporating Unidimensionality and Its Assessment," Journal of Marketing Research, 25, 186-192.

Heeler, Richard M. and M.L. Ray (1972), "Measure Validation in Marketing," Journal of Marketing Research, 9, 361-370.

Jarvis, W.Blair G. and Richard E. Petty (1996), "The Need to Evaluate," Journal of Personality and Social Psychology, 70 (1), 172-194.

Manfredo, Michael J. and Alan D. Bright (1991), "A Model for Assessing the Effects of Communications on Recreationists," Journal of Leisure Research, 23, 1-20.

Markus, H. and Robert B. Zajonc (1985), "The Cognitive Perspectives in Social Pychology," in Lindzey, G. and Aronson, E. (Eds.), Handbook of Social Psychology, 1, Erlbaum, Hillsdale, New Jersey, 137-230.

Moorman, Christine and Erika Matulich (1993), "A Model of Consumers’ Preventative Health Behaviors: The Role of Health Motivation and Health Ability, " Journal of Consumer Research, 20 (September), 208-228.

Petty, Richard E. and John.T. Cacioppo (1979), "Effects of Forewarning of Persuasive Intent and Involvement on Cognitive Responses and Persuasion," Personality and Social Psychology Bulletin, 5(2): 173-176.

Petty, Richard E. and W.Blair.G. Jarvis (1996), "An Individual Differences Perspective on Assessing Cognitive Processes," in Schwarz, N. and Sudman, S. (Eds.), Answering Questions: Methodology for Determining Cognitive and Communicative Processes in Survey Research, Jossey-Bass Inc., San Fransisco, California, 221-257.

Rothman, Alexander J. and Peter Salovey (1997), "Shaping Perceptions to Motivate Healthy Behavior: The Role of Message Framing," Psychological Bulletin, 121 (1), 3-19.

Sadowski, Cyril J. and S. Gulgoz, (1992), "Internal Consistency and Test-Retest Reliability of the Need for Cognition Scale," Perceptual and Motor Skills, 74, 498.

Skelton, John A. and Robert T. Croyle (1991), Mental Representations in Health and Illness. Springer-Verlag, New York, New York.

Thompson, Michael M. and Mark P. Zanna (1995), "The Conflicted Individual: Personalty-based and Domain-specific Antecedents of Ambivalent Social Attitudes," Journal of Personality, 63, 259-288.

Verplanken, Bas (1991), "Persuasive Communication of Risk Information: A Test of Cue Versus Message Processing Effects in a Field Experiment," Personality and Social Psychology Bulletin, 17, 188-193.

Verplanken, Bas, Pieter T. Hazenberg, and Grace R. Palenewen (1992), "Need for Cognition and External Information Search Effort," Journal of Research in Personality, 26, 128-136.

Wolfe, Raymond .N. and James W. Grosch (1990), "Personality Correlates of Confidence in One’s Decisions," Journal of Personality, 58, 515-534.

Zaichkowsky, Judith L. (1985), "Measuring the Involvement Construct," Journal of Consumer Research, 12: 341-352.

----------------------------------------