Involvement: a Potentially Important Mediator of Consumer Behavior

ABSTRACT - Although "involvement" has the potential of being an important mediator of consumer behavior, our current understanding of its effects are limited. The primary reason for this seems to be our general failure to develop a publicly acceptable conceptual definition of "involvement", valid measures of it and procedures for manipulating it in the laboratory. Until this is accomplished the quality and quantity of empirical research in this area will be limited. The papers in this session, in general, reflect this problem.

Citation:

Andrew A. Mitchell (1979) ,"Involvement: a Potentially Important Mediator of Consumer Behavior", in NA - Advances in Consumer Research Volume 06, eds. William L. Wilkie, Ann Abor, MI : Association for Consumer Research, Pages: 191-196.

Advances in Consumer Research Volume 6, 1979      Pages 191-196

INVOLVEMENT: A POTENTIALLY IMPORTANT MEDIATOR OF CONSUMER BEHAVIOR

Andrew A. Mitchell, Carnegie Mellon University

ABSTRACT -

Although "involvement" has the potential of being an important mediator of consumer behavior, our current understanding of its effects are limited. The primary reason for this seems to be our general failure to develop a publicly acceptable conceptual definition of "involvement", valid measures of it and procedures for manipulating it in the laboratory. Until this is accomplished the quality and quantity of empirical research in this area will be limited. The papers in this session, in general, reflect this problem.

INTRODUCTION

Based on the number of papers on the subject over the last couple of years, interest in "involvement" and its effect on consumer behavior has been growing at an increasing rate (e.g. Robertson, 1976; Rothschild, 1978). There seems to be general agreement that "involvement" is a potentially important mediator of consumer behavior, however, "involvement" remains an elusive concept. Precisely what is involvement? How do we measure it? How do we manipulate involvement in the laboratory? Until we can answer these questions, the quantity and quality of empirical research on the subject will remain limited.

In this discussion paper, I would like to elaborate on these issues after briefly discussing the three papers in this session. Of the three papers, the one by Gr°nhaug and Kangun seems to have only a tangential relationship to "involvement." Therefore, I will discuss it first, followed by the Newman and Dolich and the Lastovicka papers, respectively. These latter two papers present the results of empirical studies using different measures of "involvement." In the discussion of these papers, I will only mention how "involvement" was measured in each study and reserve discussion of these measures until later.

GR+NHAUG AND KANGUN PAPER

This paper examines differences in smoking related beliefs between smokers and nonsmokers in Norway. Although there is some question as to the projectability of the results, the study does indicate that nonhealth related beliefs may be important determinants of smoking behavior. In addition, the overall measurement approach and the cross cultural nature of the study make it an interesting paper.

Belief Hierarchy

In the study, the authors use Fishbein's hierarchy of beliefs or different levels of beliefs (Fishbein, 1977) to analyze smoking behavior. These different levels are awareness, general acceptance and personal acceptance. They form a hierarchy because a belief at the personal acceptance level, for instance, implies a belief at the awareness and general acceptance levels. This conceptualization of different belief levels has two important implications. First, it suggests that information at the different levels may be organized in different structures in secondary or long term memory. For instance, there may be a memory structure for the general concept of smoking which contains general beliefs about smoking. There may also be a separate memory structure containing beliefs about an individual's smoking behavior. This conceptualization is consistent with some of our preliminary findings using elicitation procedures to examine the content of stored information about specific brands in secondary memory (see Olson, 1979 for a discussion of these procedures). Basically we find that placing individuals in a purchase mode produces a different set of associations about a particular brand than if the subject were not in this mode. As might be expected, the associations elicited from a purchase mode are generally evaluative (i.e. brand x gets clothes cleaner).

In general, we would expect that the beliefs organized into the memory structure concerning the general concept of smoking would determine a general attitude about smoking, while beliefs organized into the memory structure concerning the individual's smoking behavior would determine the attitude toward the act of smoking for the individual. Consequently, we would expect that personal beliefs would be better predictors of behavior. This study provides some evidence of this since more differences were found between smokers and nonsmokers for beliefs at the personal acceptance level than the general acceptance level.

A second implication is that different cognitive processes are required to form beliefs at the different levels. The formation of beliefs at the awareness level seems to involve processes similar to those found with incidental learning (Postman, 1964; McLaughlin, 1965). Here information is learned at the nonsense syllable level (i.e. little meaning is attached to the information). At both the general acceptance and the personal acceptance levels the information acquires semantic meaning. At the personal acceptance level, however, the information has been directly related to the individual's behavior and the outcomes of this behavior. A similarity should be noted here with Krugman's measure of "involvement" as the number of "connections" an individual makes between a message and the "content of their life" during exposure to a message (Krugman, 1967). Cognitive processes that result in the generation of these "connections" would be essential in forming beliefs at the personal acceptance level. However, these "connections" could also be made after exposure to a message. Individuals may form beliefs at the general acceptance level and then, at some later point in time, reprocess this information to form beliefs at the personal acceptance level.

In summary, the concept of different levels of beliefs is an interesting one and is applicable to other areas of consumer research. As we become more concerned with measuring precisely what information consumers store in secondary or long term memory about products, the distinction between different belief levels should prove useful. This distinction may also prove useful in communication research. It may be determined, for instance, that the type of copy strategy used in an advertisement may have an important affect on whether individuals form beliefs at the personal acceptance level.

Criticisms

My criticisms of the paper are relatively few in number. First, in the presentation of the results, it would be helpful if there was an indication in the tables of which variables were significantly different. It is also not clear from the paper, exactly what statistical tests were used to determine statistical significance. Second, I believe that the awareness level should be measured by a general awareness of information as opposed to knowledge of which particular agency disseminated it. Someone may remember seeing a message that "smoking causes cancer," but may not remember the sponsoring agency. In using the proposed framework, recognition of the information would seem to be a more important measure than recall of the sponsoring agency. Finally, although the paper stresses the importance of understanding the causes of smoking behavior and how it may be changed, the thrust of this research does not appear to be in this direction.

Future Research

In using Fishbein's attitude theory (Fishbein, 1967) to examine how smoking behavior may be changed, the first step is to examine precisely which beliefs determine smoking behavior. Here we may find considerable heterogeneity in the population with respect to these beliefs, which may form a basis for segmenting the market. Second, it must be determined if getting smokers to form beliefs about the health hazards of smoking at the personal level will change their smoking behavior. Again, we should expect to find that the formation of these beliefs may cause some smokers to change their behavior, but not others. Finally, if it is determined that getting smokers to form these beliefs will cause a significant number of smokers to change their behavior, then a communication program must be developed to accomplish this goal. This, of course, is probably the most difficult step. If it is determined that getting smokers to form beliefs about the health hazards of smoking at the personal level will not cause a change in the behavior of smokers, then, other beliefs will need to be attacked or formed.

This discussion raises one final political question that does not seem to have been resolved, at least in the United States. This concerns to what extent the government should attempt to deter cigarette smoking. At one level the government may simply be concerned with informing its citizens about the health hazards of smoking. The goal would be to have significant proportion of the population form beliefs about the health hazards of smoking at the general acceptance level. At the second level, the government would want a significant proportion of the population to form these beliefs at the personal acceptance level. At the next level, the government would actively try to deter cigarette smoking by any means possible short of banning cigarettes from the market. For instance, attempts may be made to change nonhealth related beliefs if this proves more effective in changing smoking behavior than changing or forming health related beliefs. A final level, would be the outright ban of cigarettes. I suspect, however, that the government will not want to make this decision until an understanding of the relative effectiveness of programs based on these different levels is determined. This, I think, emphasizes the importance of this research.

NEWMAN AND DOLICH PAPER

This paper examines "ego involvement" as a mediator of communication effects using Social Judgment Theory (Sherif, Sherif and Nebergall, 1965) as a framework. It is somewhat unique in that the communication was a demonstration drive in an automobile. Clearly, an understanding of how "involvement'' mediates communication effects is needed, however, conceptual and methodological problems in this study place into question the validity of the results.

Methodological Problems

From a methodological perspective, it should be noted first that this basically is a correlational study. Consequently any differences found between groups may always be attributed to any other characteristic on which the groups differ. Second, the use of Social Judgment Theory as a framework means that a control group cannot be used to control for possible measurement effects due to the necessity of a before treatment measure. Because of this, the resulting design is subject to demand characteristics. Although different experimenters were used for each part of the study, I suspect that this did not conceal the purpose of the study from many of the subjects. Finally there are always problems in using change scores as a dependent variable. Space does not permit a thorough discussion of this problem, however, it occurs because the magnitude of a difference score is due to both actual or treatment differences and measurement error (Cronbach and Furby, 1970). Frequently, these two effects will be systematically related as in regression effects (Lord, 1963). These problems can usually be controlled by using the after measure as the dependent variable and the before measure as a covariate. In this study, however, the use of this procedure is not possible since the before measure will be highly correlated with whether the Pinto is an acceptable or unacceptable alternative. It is difficult to assess the effect of these methodological problems on the results of the study, however, it should be noted that each one of them provides one or more alternative explanations of the results.

Conceptual Problems

There are also conceptual problems in the interpretation and application of Social Judgment Theory in this situation. To properly use Social Judgment Theory, the authors should have measured the respondents' prior evaluation of the Pinto, the latitudes of acceptance and rejection of messages concerning the Pinto on an evaluative scale, the respondents' perception of the position advocated by the message and the respondents' evaluation of the Pinto after receiving the message (Figure 1). In this particular situation, however, the message was a demonstration drive in a Pinto, so it would be difficult to obtain the latter two measures separately. Consequently, a confirmation/ disconfirmation of expectations framework (Olson and Dover, 1979; Anderson, 1973) may be more appropriate in this situation.

FIGURE 1

SOCIAL JUDGMENT THEORY APPLIED TO MESSAGES EVALUATING THE PINTO

It should also be noted that the theoretical predictions given in Table 1 of the paper are not based on Social Judgment Theory. Without knowledge of the subject's latitude of acceptance and rejection for messages providing an evaluation of the Pinto and the subject's perception of the message, no prediction is possible as to either the direction or amount of attitude change. In addition, the terms assimilation and contrast are misused. In Social Judgment Theory these terms refer to distortions in perceptual judgments and not the direction of attitude change (Sherif and Hovland, 1953). Finally, ego involvement with the Pinto should be measured by the relative sizes of the latitudes of acceptance and rejection and not by the relative number of acceptable and unacceptable brands (Sherif and Sherif, 1967). This latter measure may be confounded by the location of the individual in the purchase decision process. For instance, someone just beginning the process may have a large number of acceptable brands, while someone near the end of the process may only have a small number of acceptable brands.

Aside from these methodological and conceptual comments, I also have a couple of comments concerning the presentation of the results. It would be helpful if the authors gave the number of subjects per cell, discussed precisely how attitudes were measured and mentioned what statistical tests were used, if any.

In summary, methodological and conceptual problems place into question the validity of the results. If the authors' main goal is the testing of Social Judgment Theory in a marketing context a different experimental situation will probably be required. If the authors are primarily interested in examining the effects of a demonstration drive as a method of attitude change a confirmation/disconfirmation of expectations framework may be more appropriate.

LASTOVICKA PAPER

This paper examines "involvement" as a basis for a product classification system. The author suggests that in order for "involvement" to form such a basis, it must be first demonstrated that "involvement" is an important mediator of consumer behavior. In general, I agree with this, however, we also need a publicly agreed upon conceptual definition of "involvement" and a valid measure of it. Then, in order to use "involvement" as a basis for a product classification system, it needs to be demonstrated that there is little within product variance across individuals on a valid measure of "involvement" and that different product classifications based on this measure imply different marketing programs.

Dependent Variable

In examining "involvement" as a mediator of consumer behavior, the author defines "consumer behavior" on a uni-dimensional scale anchored by a description of extensive problem solving and routinized response behavior. In responding to this scale, the respondent first reads a general description of extensive problem solving and routinized response behavior and then indicates for a number of different product purchase situations his or her hypothesized shopping behavior. This dependent measure is suspect for a number of reasons. First, I question whether shopping behavior can be validly measured on a undimensional scale. When amount of stored information, usage of stored information, amount of information search from different sources and brand choice behavior are included as aspects of shopping behavior, the scale must be multidimensional. It is also not clear conceptually why this measure of "consumer behavior" was selected. All purchases of "low involvement" products should not necessarily exhibit routinized response behavior and all purchases of "high involvement" products should not necessarily exhibit extensive problem solving. Second, we should be suspicious of all self-report measures of shopping behavior. The study by Newman and Lockeman (1975), for instance, suggests that individuals generally give biased estimates of their information search behavior. This problem is compounded in this study since the measure is concerned with shopping behavior in a hypothetical as opposed to an actual situation. There is an indication that there are problems with this scale since according to Figure 1 in the paper, thirty to forty percent of the sample reported that they would use extensive problem solving to purchase lightbulbs, bread and toothpaste for their personal use.

First Analysis

In this analysis a repeated measure ANOVA is used to examine individual and product-in situation effects on the self-report measure of shopping behavior. Three of the product-in-situation are hypothesized to be "high involvement'' situations. Although the classification of the product-in-situation has face validity, it is somewhat surprising that the author did not actually test whether these product-in-situations actually differed on the two scales that the author uses to measure "involvement".

There is also some discussion in the paper about the difficulty in separating the effect of the interaction from the error in the analysis. It should be noted, however, that all of the remaining degrees of freedom in the analysis are accounted for by the interaction.

The results of the analysis indicates that approximately 17% of the variance in the dependent variable is accounted for by individual effects, approximately 20% by the pro-ducts-in-situation factor and the remainder by the interactions. The author interprets this result as an indication that the respondents were "relatively homogeneous in behavior within a product class and heterogeneous across product". Based on the reported results this conclusion is difficult to justify. The results seem to indicate considerable heterogeneity within a product-in-situation. Unfortunately, it is difficult to ascertain how much of this is due to measurement problems and how much is due to actual behavioral differences.

Second Analysis

In this analysis, multiple regression analysis is used to examine the effect of "involvement" on the self-report shopping behavior scale after controlling for "knowledge about the product class", "amount of personal experience with the product class" and various demographic variables. In the analysis, "involvement" is hypothesized to consist of two components: (1) the "importance" of the purchase in attaining personal values and (2) the amount of "commitment'' that the individual has to a particular brand within the product category. The validity of these components as a measure of "involvement" seem to be based on the results of a previous study (Lastovicka and Gardner, 1978). In this study the two dimensions from a multidimensional scaling of the relative importance of a number of different product categories were interpreted as "importance'' and "commitment". The resulting configuration, however, indicates that the solution should probably have only a single dimension and the authors admit that this solution would be acceptable based on the resulting stress value. It should also be noted that according to the authors' interpretation of the two-dimensional solution, automobiles and houses would be lower on "commitment" than tissues. Consequently, there is some question as to whether or not these are valid measures of product class "involvement".

The results of the second analysis indicates that the set of independent variables explain approximately 16% of the variance in the dependent variable. The variable that is the most important predictor of the dependent variable, noncommitment, explains approximately 3% of the variance. This is especially disturbing because we would intuitively expect a large negative correlation between noncommitment and the dependent variable. An individual that is committed to a particular brand in a product category should exhibit routinized response behavior. This appears to be another indication that there are probably serious validity problems with the measures used in this study.

In summary, although I generally agree with the overall direction of this research, the author has not been as careful as he might have been in developing valid measures of his constructs. This appears to be especially true of the dependent variable. Consequently, the results of this study, which seems to indicate considerable heterogeneity of shopping behavior within product-in-situations and little relationship between "involvement" and shopping behavior, should not be considered conclusive. With better measures of "involvement" and valid multidimensional measures of shopping behavior, I suspect that we may find some rather strong relationships.

FUTURE RESEARCH DIRECTIONS

There seems to be general agreement on the potential importance of the construct "involvement" as a mediator of consumer behavior. However, there seems to be little agreement on precisely what "involvement" is or how it should be measured. In this section, I would like to present my conceptualization of "involvement" and then discuss future research directions using field and laboratory methods.

Involvement: What is it? Briefly, I view "involvement" as an individual level, internal state variable that indicates the amount of arousal, interest or drive evoked by a particular stimulus or situation. Involvement, therefore, has two dimensions, intensity and direction. Intensity concerns the level of arousal, interest or drive and direction concerns the evoking stimulus object and/or situation. In addition, the direction component may concern situations and stimulus objects at different levels of generality. Consequently, we may discuss "involvement" with respect to a product category, a particular brand and the purchase of a product for a particular reason.

For cognitive psychologists who are not used to thinking in terms of arousal or interest, "involvement" might be thought of in terms of a goal hierarchy (Simon, 1967). Here "involvement" might indicate the location of the particular stimulus and/or situation in a goal hierarchy at a particular point in time. A high "involvement" situation would have a reasonably high priority in the hierarchy and a low "involvement" situation a low priority. Consequently, we might think of "involvement" as mediating the amount of cognitive effort that will be expended in a given situation.

In the literature, two different types of definitions/measures of "involvement" have appeared. In one case, "involvement'' has been viewed as a state variable and in others as a process. As discussed previously, I prefer to think of "involvement" as a state variable, so I will, therefore, discuss alternative definitions of this type first.

Most of the state variable measures of "involvement" that have appeared in the consumer behavior literature have their roots in social psychology. Here, starting with Sherif and Cantril (1947), it has been conceptualized in terms of the relationship between an individual's values and an issue or object. The more the issue or object becomes integrated with the individual's values, the higher the level of "involvement". This definition of "involvement'', termed "ego involvement", has generally been viewed in the social psychological literature as a mediator of persuasion effects. More recent definitions of "ego involvement'' have stressed "commitment" to the issue (Freidman, 1964) or the relationship of the issue or object to other elements of an individual's cognitive structure (Ostrom and Brock, 1968).

In developing measures of "involvement", researchers in consumer behavior have tended to simply apply social psychological measures of "ego involvement" with a particular issue to marketing situations involving products. In many cases, however, there appear to be conceptual problems in making this transition. For instance, Newman and Dolich and others (e.g. Lastovicka and Gardner, 1978), have adapted the Our Category Procedure (Sherif and Sherif, 1967), to measure "involvement" with a product category. Under this adaptation, the relative number of acceptable and unacceptable brands within a product category are used to measure "involvement" with the product category. Conceptually, however, this is not a measure of product class "involvement" as defined here. As mentioned previously, this measure may be influenced by where the individual is in the purchase decision process if the product category is a consumer durable. If the product category is a consumer nondurable, an individual may have only one or two acceptable brands in order to reduce the amount of effort required in decision making in a "low involvement" product category. Consequently, the relative number of acceptable and unacceptable brands in product category is not a conceptual measure of product class "involvement." It may be a consequence of it, however, this is an empirical question.

Similarly, the "commitment" measure used by Lastovicka does not appear to be a conceptual measure of product class "involvement." Instead, it appears to be closer to a measure of the cognitive dimension of brand loyalty (Mitchell and Olson, 1975). Conceptually, I see little relationship between product class "involvement" and brand loyalty, however, again an empirical relationship may exist.

The second measure of "involvement" used by Lastovicka seems to be similar to "ego involvement" as defined by Sherif and Cantril (1947), however, it is different from my conceptualization given earlier. I view "ego involvement'' as a possible antecedent condition of "involvement." As such it may be a sufficient condition for "high involvement,'' but it is not a necessary condition. For instance, a refrigerator may not be high on "ego involvement'', however, for many people it may be a "high involvement" purchase situation.

Process definitions/measures of "involvement" usually involve information acquisition processes and decision processes. For instance, Krugman (1967), measures "involvement'' in terms of the number of "connections" made between the product being advertised and the individual's personal life during exposure to an advertisement. Houston and Rothschild (1978) suggest a measure of "involvement" in terms of decision processes and Ray et al (1973) define "involvement" in terms of the orders or mental states after exposure to information. However, if we conceptualize "involvement" as discussed here, defining "involvement" as a process may be misleading. Although, the level of "involvement'' may have a strong effect on the process used to make decisions or acquire information, I do not believe that it is the only determinant of these processes. For instance, in acquiring information from advertisements, the modality, the structure and content of the advertisement may also have an effect on these processes Consequently, I don't believe it is useful to define "involvement'' as a process.

Once we have arrived at a conceptual definition of involvement, there are a number of research directions that are important to pursue. These are best discussed under the classifications of field and laboratory studies.

Field Studies: The first priority here would be to develop a scale for measuring "involvement". Following the conceptual definition given here, a bank of questions might be developed to measure the level of arousal or interest in a particular product class or brand. "How likely are you to read a magazine article about _____________?" or "Do you generally read advertisements for __________?" might be examples of these questions. Standard procedures, of course, should be used to insure the internal consistency, reliability, and the validity (Cronbach and Meehl, 1955; Campbell and Fiske, 1959) of the scale.

Once a reliable and valid scale is developed, there are a number of possible research directions. All of these involve the construction of a nomological net relating "involvement" to other aspects of consumer behavior (Cronbach and Meehl, 1955). First, the antecedent and consequent conditions of product class "involvement" should be examined. For instance, how exactly does "involvement" affect purchase behavior? Second, what is the relationship between "involvement" and other constructs in consumer behavior such as perceived risk? Finally, how much between respondent variance is there on "involvement" within different product categories?

Laboratory Studies: The first priority here would be to examine how involvement affects information acquisition and decision processes. The central problem in these types of studies is how to create different "involvement levels in the laboratory. This, I'm afraid, may prove to be a difficult task. In our current studies examining advertising effects at the individual level, we have, instead, tried to hypothesize how "involvement" may affect the type of processing that occurs during exposure to an advertisement (Mitchell, Russo and Gardner, 1978). We have hypothesized, for instance, that "involvement" may affect the strategy that an individual uses in processing information and the likelihood that competing stimuli will reduce attention levels. We have then attempted to create experimental conditions where these different processes will occur. This allows us to examine the effects of these different processes on the evaluations formed and the amount of information recalled about the brand and the advertisement.

SUMMARY

The concept of "involvement" seems to be potentially an important mediator of consumer behavior. However, before its potential can be determined empirically we need a publicly acceptable conceptual definition of "involvement, a valid scale for measuring it and methods for manipulating "involvement" in laboratory settings. In this paper, it has been defined as a individual level, state variable that measures the amount of arousal or interest in a stimulus object or situation. As such, "involvement" has two dimensions, intensity and direction. Consequently, we may talk about the amount of "involvement" with a product class, a brand or a purchase situation. Different measures of "involvement" that have appeared in the literature were then examined using this definition. Problems were indicated with each method.

Future research directions were also discussed. These involve the development of a scale for measuring "involvement,'' an examination of the antecedents and consequences of product class involvement and an examination of the effect of involvement on information acquisition and decision processes.

References

Anderson, R. E. "Consumer Dissatisfaction: The Effect of Disconfirmed Expectancy on Perceived Product Performance," Journal of Marketing Research, 10(1973), 38-44.

Campbell, D. T. and D. W. Fiske, "Convergent and Discriminant Validation by the Multitrait-Multimethod Matrix," Psychological Bulletin, 56(1959), 81-105.

Cronbach, L. J. and P. E. Meehl, "Construct Validity in Psychological Tests," Psychological Bulletin, 52(1955), 281-302.

Cronbach, L. J. and L. Furby, "How we should Measure 'Change'-or Should We?" Psychological Bulletin, Vol. 74 (1970), 68-80.

Fishbein, M. "A Behavior Theory Approach to the Relations between Beliefs about an Object and the Attitude toward the Object," in M. Fishbein, ed., Readings in Attitude Theory and Measurement, New York: John Wiley & Sons, 1967, 257-266.

Fishbein, M., Consumer Beliefs and Behavior with Respect to Cigarette Smoking: A Critical Analysis of the Public Literature, Staff Report of the Federal Trade Commission, May, 1977.

Fishbein, M. and I. Ajzen, Belief, Attitude, Intention and Behavior: An Introduction to Theory and Research, Reading, Mass.: Addison-Wesley Publishing Co., 1976.

Freedman, J. L. "Involvement, Discrepancy, and Change," Journal of Abnormal and Social Psychology, 69(1964), 290-295.

Houston, M. J. and M. L. Rothschild, "Conceptual and Methodological Perspectives on Involvement" in S. Jain, ed. Research Frontiers in Marketing: Dialogues and Directions, 1978 Educators' Proceedings, Chicago: American Marketing Association, 1978, 184-187.

Krugman, H. E., "The Impact of Television Advertising: Learning Without Involvement," Public Opinion Quarterly, 29(1965), 349-346.

Krugman, H. E., "The Measurement of Advertising Involvement,'' Public Opinion Quarterly, 30(1967), 583-596.

Lastovicka, J. L. and D. M. Gardner, "Low Involvement Versus High Involvement Cognitive Structures," in H. K. Hunt, ed., Advances in Consumer Research, Vol. V, Ann Arbor: Association for Consumer Research, 1978, 87-92.

Lastovicka, J. L. and D. M. Gardner, "Components of Involvement,'' in J.C. Maloney and B. Silverman, eds., Attitude Research Plays for High Stakes, Chicago: American Marketing Association (forthcoming).

Lord, F. M., "Elementary Models for Measuring Change," in C. W. Harris, ed., Problems in Measuring Change, Madison" University of Wisconsin Press, 1963, 21-38.

McLaughlin, B., "'Intentional' and 'Incidental' Learning in Human Subjects: The Role of Instructions to Learn and Motivation," Psychological Bulletin, 63(1965), 359-376.

Mitchell, A. A. and J.C. Olson, "The Use of Restricted and Unrestricted Maximum Likelihood Factor Analysis to Examine Alternative Measures of Brand Loyalty" in E. M. Mazze, ed., Marketing: The Challenges and Opportunities, Chicago: American Marketing Association, 1975, 181-186.

Mitchell, A. A., J. E. Russo and M. Gardner, "Strategy Limitation Effects on the Learning of Advertising Messages," Working paper 49-77-78, Graduate School of Industrial Administration, Carnegie-Mellon University, Pittsburgh, PA. 15213.

Newman, J. W. and B. D. Lockeman, "Measuring Prepurchase Information Seeking," Journal of Consumer Research, 2 (1975), 216-222.

Olson, J.C., "The Stability of Responses Obtained by Free Elicitation: Implications for Measuring Attribute Salience and Memory Structure," in William Wilkie, ed. Advances in Consumer Research, Vol. VI, (forthcoming).

Olson, J.C. and P.A. Dover "Disconfirmation of Consumer Expectations Through Product Trial," Journal of Applied Psychology (forthcoming).

Ostrom, T. M. and T. C. Brock, "A Cognitive Model of Attitudinal Involvement," in R. P. Abelson, et al, eds., Theories of Cognitive Consistency: A Sourcebook, Chicago: Rand McNally, 1968.

Postman, L., "Short-term Memory and Incidental Learning," in W. A. Melton, ed., Categories of Human Learning, New York: Academic Press, 1964.

Ray, M. L., A. G. Sawyer, M. L. Rothschild, R. M. Heeler, E. C. Strong, and J. B. Reed, "Marketing Communication and the Hierarchy of Effects," in P. Clarke, ed., New Models for Mass Communication Research, Beverly Hills, Ca.: Sage Publications, 1973.

Robertson, T. S., "Low Commitment Consumer Behavior," Journal of Advertising Research, 16(1976), 19-27.

Rothschild, M. L., "Political Advertising: A Neglected Policy Issue in Marketing," Journal of Marketing Research, 15(1978), 58-71.

Rothschild, M. L. and M. J. Houston, "The Consumer Involvement Matrix: Some Preliminary Findings," in B. A. Greenberg and D. N. Bellenger, Contemporary Marketing Thought, 1977 Educators' Proceedings, 1977, 95-98.

Sherif, M. and H. Cantril, The Psychology of Ego Involvement, New York: John Wiley & Sons, 1947.

Sherif, M. and C. J. Hovland, Social Judgment: Assimilation and Contrast Effects in Communication and Attitude Changes, New Haven: Yale University Press, 1953.

Sherif, M. and C. W. Sherif, "The Own Categories procedure in Attitude Research," in M. Fishbein, ed., Readings in Attitude Theory and Measurement, New York: John Wiley and Sons, 1967.

Sherif, C. W., M. Sherif and Nebergall, Attitude and Attitude Change: The Social Judgment Involvement Approach, New Haven" Yale University Press, 1965.

Simon, H. A., ,"Motivational and Emotional Controls of Cognition,'' Psychological Review, 74(1967), 29-39.

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