Application and Analysis of the Behavioral Differential Inventory For Assessing Situational Effects in Buyer Behavior


Russell W. Belk (1974) ,"Application and Analysis of the Behavioral Differential Inventory For Assessing Situational Effects in Buyer Behavior", in NA - Advances in Consumer Research Volume 01, eds. Scott Ward and Peter Wright, Ann Abor, MI : Association for Consumer Research, Pages: 370-380.

Advances in Consumer Research Volume 1, 1974    Pages 370-380


Russell W. Belk, Temple University

The unresolved debate in psychology over whether personality can be expected to influence behavior as a general main effect or whether personality interacts with situational factors in determining behavior, has produced a methodological legacy for consumer behavior to utilize in the attempt to assess the importance of situational effects on the buyer. In much the same manner as personality theorists, consumer behavior researchers have taken up two opposing views on the nature and significance of situational influences. Engel, Kollat, and Blackwell, for example, conclude that, " is unreasonable to expect a measure of personality alone to give good predictions. In other words, the influence of personality becomes mixed with the effects of the environment" (1966, p. 66). Bass presents the opposing view in arguing that,

Unless improved measurements will substantially reduce noise or unless there are omitted variables with non-random occurrence which have been overlooked in earlier studies...there are serious limits to the amount of unexplained variance reduction which we can realistically expect to achieve in future empirical studies. Perhaps we should begin to focus attention on stochastic theories of consumer behavior. Instead of saying that if a condition holds a consumer will behave in a certain way we make a probability statement about his behavior, we give explicit recognition to the stochastic elements in the process being modeled. (1971, p. 17)

While a resolution of this debate has very substantial implications for the course of consumer research, the lack of comprehensive techniques for investigating situational influence has relegated the argument to sterile discourse with only anecdotal evidence to support or question the importance of buyer behavioral situations. This paper illustrates and analyzes an instrument and methodology for initial assessments of situational effects on consumer behavior. The techniques, derived from psychometrics, yield not only a quantitative statement of the importance of various situational and non-situational influences on consumer choice, but also allow the construction of statements concerning the types of products, brands, attributes, or responses preferred by types of consumers within types of purchase and consumption situations. Thus the potential benefits of the technique to marketers are great.


The Behavioral Differential Inventory

In cross-cultural studies of person perception Triandis (1962) developed a questionnaire format which has since become known as the behavioral differential inventory. Respondents to such an instrument indicate on Likert-like scales the degree to which it is likely that they would make each of several responses to various stimulus configurations. The inventory format, used by Triandis to investigate responses to persons as stimulus configurations, has been since adapted to the use of anxiety-arousing situations (Endler, Hunt, and Rosenstein, 1962), hostility-evoking situations (Endler and Hunt, 1969b), beverage consumption situations (Sandell, 1968), leisure utilization situations (Bishop and Witt, 1970), and food buying situations (Belk, 1974). Each application involves three samples:

1. Situations relevant to the context being examined;

2. Responses of interest within these situations; and

3. Persons to act as respondents.

Each of these samples provides a potential source of variance in response likelihood for which main and interaction effects may be observed.

Results will be presented from two inventories concerning consumer choice of snack products and motion pictures. These inventories were constructed by first collecting lists of situations and choices familiar to consumers for each product. The situations and responses were then categorized and given to two additional samples to test for familiarity and clarity of description. For snack products, the 10 most familiar situations and 10 most familiar products were chosen for inclusion in the final inventory. Since motion pictures are normally attended only once and prior familiarity could substantially alter responses, hypothetical titles and descriptions were constructed to parallel the categories of motion pictures obtained from pretests. This inventory consisted of 8 situations and 12 motion pictures.


In all, 5 groups were tested under the conditions specified in Table 1. Group number 1 consists of a general population sample which was restricted only by age limits of 18 to 60. Greater subject homogeneity was not sought since the relative importance of person variance is one effect under examination. Subject groups 2 through 5 were composed of college juniors from introductory business and psychology courses at a large midwestern university. Group 2 provides a more homogeneous sample for comparison to group 1, and a base sample for comparison to differing inventory compositions in groups 3 and 4. Group 5 was also chosen from student subjects because of the heavy motion picture patronage found among this population.



In an inventory with 10 situations and 10 responses, subjects -replied to 100 stimulus-response pairs. For each such pair responses were obtained on a 5-point scale from "extremely likely" to "not at all likely," to reflect the subject's assessment of his probability of making the choice indicated under the circumstances described. These circumstances related to the nature of the occasion for choice and/or consumption, and were briefly described in a sentence (for snacks) or newspaper advertisement (for motion pictures) preceding each response set. Descriptions of the situations and responses included are shown in the factor analysis tables in Appendices A, B, E, and F.


Instrument Reliability

The nature of the stimulus-response inventories utilized is antithetical to obtaining "test scores" for participants. Since the intent is to examine the variance in reported consumer behavior across situations and responses, any item summation or similar univariate description of subject performance is inappropriate. However, from the multiple administration (4 week interval) of the snack product inventory in group 2, individual item reliabilities are available, where each of the 100 stimulus-response pairs is regarded as an "item."

The Pearson product moment correlations between occasions for these items ranged from .36 to .79 with a mean correlation of .64. The range of mean item correlations for responses averaged within each of the 10 situations was from .59 to .69, and the mean reliabilities for each response averaged over situations ranged from .57 to .68. These levels contrast with an average intercorrelation of only .12 between all pairs of items on the first administration to this group.

For item-retest reliability, it is to be expected that coefficients will be lower than for test-retest reliability where variation over time in responses to particular items may be partially compensated by error variance in the opposite direction on other items combined to derive test scores. Given this consideration it is concluded that this instrument shows sufficient stability over time to warrant further examinations of the preference patterns obtained.

Sources of Behavioral Variance

The general model employed for assessing contributions of variance components is a three-way mixed effects analysis of variance (only persons are considered as a random factor). Analysis of variance procedures, commonly used for F tests of null hypotheses, may also be used to provide estimates of the relative contributions of sources in the model to variance in the dependent measure. This application is particularly germane to the study's interest in the importance of major sources of variance in buyer behavior. An excellent series of discussions and dialogues on the techniques for estimating these effects from expected mean square formulas has been conducted following publication of results from the anxiety and hostility inventories (Medley and Mitzel, 1963; Gleser, Cronbach, and Rajaratnam, 1965; Endler, 1966; Silverstein and Fisher, 1968; and Endler and Hunt, 1968, 1969a).

Table 2 shows results for group 2 with repeated measures. As with each group to which the snack product inventory was applied, the main effect of differences in responses was able to account for less than 10% of the variance. The persons by responses interaction is the largest contributor to variance, which is interpreted to mean that individual reactions to alternative products is the major determinant of product preferences. The second most important effect, however, is the responses by situations interaction. This effect shows that choice of differing snack products is dependent upon the purchase and consumption situations employed. It is inferred from the low persons by situations contribution to variance that reactions to purchase and consumption situations are consistent across snack product consumers. The three-way interaction appears to be a minor effect, contributing less than 5% to the total variance. Since estimation of this effect requires more than one observation on individuals it was not obtained for remaining inventories.



Stability Over Subjects, Products, and Occasions

Table 3 presents a comparison of variance contributions from the motion picture inventory (group 5), the two administrations of the snack product inventory to group 2, and a comparable snack inventory administered to non-students (group 1). The comparison between subject groups in columns 2 and 3 of the data shows a high degree of similarity in results and lends credence to the use of student subjects in remaining groups. If there is a student bias, it is in slightly overstating the estimates of interaction terms and slightly understating main effects.



Endler and Hunt (1968) have suggested that negativism and boredom might arise when subjects are retested and that error variance might increase as a result. The results in columns 3 and 4 of Table 3 show very slight evidence of this effect. There is, however, some tendency for main effect contributions to increase at the expense of interaction terms involving situations. This lesser susceptibility to situational effects upon the second administration supports Hansen's (1972) hypothesis that as familiarity with a situation increases, decision-making becomes more routine and less influenced by situational cues.

Between the two product categories it is apparent that substantial differences exist however. In the motion picture inventory the overwhelming effects are the persons by responses interaction (suggesting individual differences in motion picture preferences), and the main effect of responses (suggesting somewhat less influence from generalized preferences within the group). The responses by situations term (normally thought of as situational influence) drops to a poor third in this inventory.

The primary similarity in variance patterns across all inventories is the small contributions of persons, situations, and persons by situations interaction, as well as the similarity in error component size. These effects lack applied or theoretical importance. The remaining effects all involve responses and therefore are of greater concern to marketing and the study of consumer behavior

Stability Over Inventors Composition

Since application of analysis of sources of variance in buyer behavior requires restricted selections from nearly infinite numbers of situations and products, it is well to examine the influence of resulting sample sizes on variance component estimates. Table 4 compares estimates of component contributions for the three inventory formats given to groups 2 through 4. Only minor variations in contributions of variance components occurs across the three inventory formats. With fewer response alternatives, response main effect contributions are higher and its interactions contribute somewhat less. With fewer situations, the contribution of the responses by situations interaction is higher while the main effects of responses and situations are slightly less important.



Patterns of Variance

In subjecting data from the anxiety inventory of Endler et al. (1962) to a process of three-mode factor analysis, Levin (1963, 1965) has provided a key analytical tool to support applications of the behavioral differential to consumer behavior. Although the mathematics of the technique are somewhat complex, it acts to develop separate factors for each mode of observation (in this caseS persons, situations, and responses), and, most importantly, to show the interrelationships of factors from each mode by means of a "core" matrix. This core matrix, which roughly corresponds to the factor scores for types of persons on combinations of situation and response factors, can potentially be translated into a plan for market segmentation, in that it reflects the types of products (or brands, etc.) preferred by types of buyers in types of situations.

The techniques for performing three-mode factor analysis are well-discussed elsewhere (Levin, 1963, 1965; Tucker, 1963, 1964, 1966a, 1966b; and Vavra, 1972). The method used in the following results is a concise approximation employed by Levin (1963) which utilizes traditional factor analysis programs, supplemented with simple matrix manipulations. For clarity, only the data from Groups 1 and 5 will be presented. Factor names are included in core matrices only as an aid to discussion, and the corresponding submatrices appear in Appendices E and F.

Snack product patterns of variance are summarized in the situation factor matrix in Appendix A, the product factor matrix in Appendix B, and the core matrix in Appendix C. The core matrix for the motion picture inventory is shown in Appendix D.

Examination of the snack product core matrix in Appendix C provides a summary of results from the snack inventory. For each of the three person types derived, a combination of situation and response factors is required to describe response preferences. Person type I generally prefers filling and substantial snacks, except for unplanned purchase situations where he selects light and salty products and avoids the substantial snacks which he otherwise favors. Person type II tends to choose light and salty snacks to serve to others, but does not favor these snacks in individual consumption situations. Person type III is distinguished by his preference for sweet snacks in all situations but those in which he anticipates serving the snack to others. Aside from his craving for sweets this person has only weak preferences among alternative snack products.

Despite the relatively small contributions of situations and situation interactions to variance observed in the motion picture inventory, some rather clear situation and response factors emerged. However, the core matrix for motion pictures, in Appendix D, shows an almost complete absence of situation effects on response preferences. The first person factor shows a movie-goer who prefers drama across all situations with some disdain for pictures of national acclaim. The second type of movie-goer prefers either light entertainment or drama in all three types of situations derived, with a weaker preference for romantic themes, especially in relaxation situations. The third type of person shows the same disdain for nationally acclaimed movies as does person I, but prefers light entertainment pictures across situations. That these preferences should largely ignore situations is consistent with analysis of variance conclusions that situation has little impact on preference patterns for motion pictures.


The primary benefit in applying the behavioral differential inventory to buyer behavior, in addition to allowing a research approach which incorporates situational influences in the study of the consumer, is the complimentary set of analyses which it enables. Through analysis of variance, product categories may be screened for situational influence and the basic nature of behavioral influences may be more fully appreciated. The added benefits of three-mode factor analysis include a nearly operational formula for market segmentation by consumer typologies based on situational as well as product preference factors. Furthermore, the technique is applicable to a broader range of problems than have been considered in this study. It may also be employed to enrich studies of consumer behavior concentrating on attitudes and attitudinal components, preferences for risk and dissonance reduction strategies, and the investigation of situational ethics in business, for example.

Another promising and needed area of research will be the investigation of personalities and other consumer variables which influence individual susceptibility to situational influence. Findings in this area may have implications both for the improvement of marketing strategies and for the education of those consumers found especially susceptible and especially unaffected by situations. The former consumers may be shopping unwisely due to a lack of planning, while the latter consumers may be missing opportunities to save or make more optimal purchases because of their neglect of situational cues.

The primary difficulty in employing the behavioral differential inventory is the necessity of adequately sampling stimulus and response items in order to derive generalizable results. This problem also exists in studies not considering situations, but it is more ponderous in the unresearched task of situation sampling. The results of this study indicate that the inventories are fairly robust across subject groups, occasions, and inventory sizes, but the adequacy of the situation and response samples is indeterminant. This problem requires further basic research to determine relevant situations and responses for various categories of buyer behaviors. Since results show that these relevant stimuli are not consistent across the product categories examined, such research must also be conducted on a product by product basis.


Bass, F. M., Unexplained Variance in Studies of Consumer Behavior, Purdue University, Krannert School of Industrial Administration, Institute for Research in the Behavioral, Economic, and Management Sciences, Paper No. 339, 1971.

Belk, R. W., "An Exploratory Assessment of Situational Effects in Buyer Behavior,t' Journal of Marketing Research, 1974, 11, forthcoming.

Bishop, D. W. and P. A. Witt, "Sources of Behavioral Variance During Leisure Time," Journal of Personality and Social Psychology, 1970, 16, pp. 352-360.

Endler, N.S., "Estimating Variance Components from Mean Squares for Random and Mixed Effects Analysis of Variance Models," Perceptual and Motor Skills, 1966, 22, pp. 559-570.

Endler, N.S. and J. M. Hunt, "Triple Interaction Variance in the S-R Inventory of Anxiousness," Perceptual and Motor Skills, 1968, 27, p. 1098.

Endler, N. S. and J. M. Hunt, "Generalizability of Contributions from Sources of Variance in the S-R Inventories of Anxiousness," Journal of Personality, 1969a, 37, pp. 1-24.

Endler, N. S. and J. M. Hunt, "S-R Inventories of Hostility and Comparisons of the Proportions of Variance from Persons, Responses, and Situations for Hostility and Anxiousness," Journal of Personality and Social Psychology, 1969b, 9, pp. 309-315.

Endler, N. S., J. M. Hunt and A. J. Rosenstein, "An S-R Inventory of Anxiousness," Psychological Monographs, 1962, 76, pp. 1-33.

Engel, J. F., D. T. Kollat and R. D. Blackwell, "Personality Measures and Market Segmentation," Business Horizons, 1969, 12, pp. 61-70.

Gleser, G. C., L. J. Cronbach and N. Rajaratnam, "Generalizability of Scores Influenced by Multiple Sources of Variance," Psychometrika, 1965, 30, pp. 345-418.

Hansen, F., Consumer Choice Behavior (New York: The Free Press, 1972).

Levin, J., Three-Mode Factor Analysis, Ph.D. Dissertation, University of Illinois, 1963.

Levin, J., "Three-Mode Factor Analysis," Psychological Bulletin, 1965, 64, pp. 442-452.

Medley, D. M. and H. E. Mitzel, "Measuring Classroom Behavior by Systematic Observation," in N. L. Gage (Ed.), Handbook of Research on Teaching (Chicago: Rand-McNally, 1963).

Sandell, R. G., "Effects of Attitudinal and Situational Factors on Reported Choice Behavior," Journal of Marketing Research, 1968, 5, pp. 405-408.

Silverstein, A. B. and G. Fisher, "Estimated Variance Components in the S-R Inventory of Anxiousness," Perceptual and Motor Skills, 1968, 27, pp. 740-742.

Triandis, H. C. and L. M. Triandis, "A Cross-Cultural Study of Social Distance," Psychological Monographs, 1962, 76.

Tucker, L. R., "Implications of Factor Analysis of Three-Way Matrices for the Measurement of Change," in C. W. Harris, Problems in Measuring Change (Madison, Wisconsin: The University of Wisconsin Press, 1963).

Tucker, T, R., "The Extension of Factor Analysis to Three-Dimensional Matrices," in N. Frederiksen and H. Gulliksen (Eds.), Contributions to Mathematical Psychology (New York: Holt, Rinehart, and Winston, 1964).

Tucker, L. R., "Experiments in Multimode Factor Analysis," in A. Anastasi (Ed.), Testing Problems in Perspective (Washington, D.C.: American Council on Education, 1966a).

Tucker, L. R., "Some Mathematical Notes on Three-Mode Factor Analysis," Psychometrika, 1966b, 31, pp. 279-311.

Vavra, T. G., "An Application of Three-Mode Factor Analysis to Product Perception," in Proceedings of the 1971 Fall Conference of the American Marketing Association (Chicago: American Marketing Association, 1972).















Russell W. Belk, Temple University


NA - Advances in Consumer Research Volume 01 | 1974

Share Proceeding

Featured papers

See More


J10. Transnational Consumer Lifestyle

zahra Sharifonnasabi, Queen Mary University of London
Fleura Bardhi, City University of London, UK

Read More


“A Tale of Two Secrets”: Examining the Diverging Effects of Secrecy on Consumption Enjoyment

Xiaojing Yang, University of Wisconsin - Milwaukee, USA
Xiaoyan Deng, Ohio State University, USA
Lei Jia, Ohio State University, USA

Read More


O12. When do People Waste Time? Testing a Mechanism for Parkinson’s Law.

Holly S Howe, Duke University, USA
Tanya Chartrand, Duke University, USA

Read More

Engage with Us

Becoming an Association for Consumer Research member is simple. Membership in ACR is relatively inexpensive, but brings significant benefits to its members.