Involvement and Other Variables Mediating Communication Effects As Opposed to Explaining All Consumer Behavior

Michael L. Ray, Stanford University
ABSTRACT - The papers in this session reflect an advanced stage of use of involvement in consumer research. Serious questions about reliability and particularly validity and usefulness of the concept are being asked. The answers seem to be that involvement measures can be reliable, valid and useful only within certain situations. At present, as the title indicates, involvement has been shown to be measured well and to be valuable only in the communication effects situation. And even there it must be used along with other variables. Some general implications for the development and use of variables such as involvement are discussed.
[ to cite ]:
Michael L. Ray (1979) ,"Involvement and Other Variables Mediating Communication Effects As Opposed to Explaining All Consumer Behavior", in NA - Advances in Consumer Research Volume 06, eds. William L. Wilkie, Ann Abor, MI : Association for Consumer Research, Pages: 197-199.

Advances in Consumer Research Volume 6, 1979      Pages 197-199

INVOLVEMENT AND OTHER VARIABLES MEDIATING COMMUNICATION EFFECTS AS OPPOSED TO EXPLAINING ALL CONSUMER BEHAVIOR

Michael L. Ray, Stanford University

ABSTRACT -

The papers in this session reflect an advanced stage of use of involvement in consumer research. Serious questions about reliability and particularly validity and usefulness of the concept are being asked. The answers seem to be that involvement measures can be reliable, valid and useful only within certain situations. At present, as the title indicates, involvement has been shown to be measured well and to be valuable only in the communication effects situation. And even there it must be used along with other variables. Some general implications for the development and use of variables such as involvement are discussed.

INTRODUCTION

The modern introduction of ego involvement as a social psychological variable probably can be traced to the book by Sherif and Cantril (1947) published over three decades ago. Although the concept appeared sporadically in the consumer research literature since then, it was introduced most effectively to our field only about thirteen years ago by Krugman (1965) when he explained the effects of television advertising with a low involvement viewer hypothesis. Since that time the concept has been expanded for the consumer field by Ray et al (1973, Ray 1976), Robertson (1976), Rothschild (1974), and by a series of authors who participated in an American Marketing Association Attitude Research conference devoted primarily to the involvement variable (Maloney and Silverman, 1978).

Given its brief history in consumer research, the involvement concept has only just started to be questioned and refined for its application to consumer behavior issues. The papers in this session reflect this stage in the application of the concept.

THE PRESENT AND POTENTIAL STATE OF INVOLVEMENT IN CONSUMER RESEARCH

The main questions now seem to be ones of validity. What is the best measure of involvement? What are the components of involvement? Does involvement affect consumer behavior in the way involvement "theorists" say it should? What are the alternative concepts and measures of concepts which might operate to explain consumer behavior phenomena in conjunction with or in place of involvement?

It is a basic theme of this discussion paper that the ultimate answers to these questions about involvement will be quite conservative and limiting. That is, there will be no one best method to measure involvement. The components and effects of involvement will be governed primarily by the consumer decision situation. And involvement alone will not explain consumer phenomena. Other mediating variables will have to be invoked. In short, involvement will not be the whole answer to the fascinating problems of consumer research. Involvement will provide part of the answer only in certain situations. At this point, as the title of this paper indicates, we can be somewhat confident in the operation of involvement along with other variables only in the communication situation.

This may seem to be a pessimistic view. Actually it is realistic. Consumer research tends to be an applied field. We borrow concepts and techniques from several sciences. Early in the process we have unrealistic expectations for a new concept. Then, because these expectations cannot possibly be met, there is a period of disillusionment. That period could lead to the concept being dropped or being refined. Fortunately, as illustrated by the papers in this session, the period of disillusionment for the involvement concept is leading to refinement rather than rejection.

THE PAPERS

Lastovicka

Lastovicka asks all the right questions about involvement and provides evidence of its nature, importance and interaction with other variables. It is clear that involvement as conceptualized by Lastovicka is an important variable. It is also just as clear that involvement alone will not explain consumer behavior; other variables must be used.

There are limits to the inferences we can make from the Lastovicka research, however. These limits are caused basically because his key measures have not been tested for reliability or validity. While Lastovicka is in the majority of consumer research in this regard (e.g. Ray, 1979) and has actually done some previous validity testing of measures related to involvement (Lastovicka and Gardner, 1978), the study is based entirely on responses on a five-point scale to two previously untested descriptions. That yeasaying is present is clear from the mean results which show that even the low involvement product situations produced scores around the mid-point of the scale. It appears that the respondents as a group didn't really want to admit that their behavior was totally to the routinized behavior side of the continuum.

This lack of validation for the key measure could be dismissed if this were only the first of a series of studies. But in consumer research we do not have a tradition of replications.

It could also be argued that the possible yeasaying or role-playing decreased variance so that the strong effect of involvement shown in the study was actually conservative. But this argument ignores the fact that, as mentioned above, the two descriptions consumers rated for seven product categories were untested and, I believe, biased toward some effect of involvement. My belief should have less credibility than Lastovicka's here since this is his study. But on the other hand anybody's guess is still a guess since his key measure was not validated.

What is clear is that there is large residual error variance which cannot be explained by either subject or product (involvement). My guess is that an error term of this size is probably due to the interaction of the products with the yeasaying or role playing biases that probably exist in this study. This is not to say that Lastovicka's results are worthless. But they must be followed with much further work if they are to have lasting value.

Some hint as to the character of this further work is seen in the limits to Lastovicka's "Implications" section. While I agree that his research can support his statement that "... no one classification device is a very satisfactory device for totally accounting for differences in acquisition behavior," I find his speculations about the information processing behaviors of people new to a product class to be completely without foundation in his research and counter to results my coworkers and I have obtained. Specifically, I find it hard to believe that there are substantial numbers of people purchasing in any product category which, as Lastovicka implies, never do any extensive problem solving. I guess I would be classified as a proponent of "low involvement theory" as Lastovicka put it. Even though I believe this is a misuse of the word "theory," I do believe that we "theorists" never said that a consumer's whole experience with a product would be in a low involvement mode.

In sum, Lastovicka's study is a nice demonstration of the power of involvement and a first step which points to the need for further research. That further research should consist of experimental studies which determine how people in various stages of the decision process behave with regard to various sources of information. Notice it will be necessary to determine how people actually behave rather depending on how they say they behave. Thus far such studies have been done only with regard to advertising, with the exception, perhaps, of the next study to be discussed.

Newman and Dolich

This is an interesting application of Sherif's ego-involvement notions to consumer behavior; particularly in light of the fact that this was a high-involvement communication (a live product demonstration of an automobile) and that the study was experimental in nature. The author's conclusion that the operation of the ego-involvement was as expected (if it can be assumed that the demonstration was a negative experience) is both provocative and full of implications for marketers.

Unfortunately there are several shortcomings to the Newman and Dolich research which limits the implications we can draw from it. Like Lastovicka, they use scales that have not been tested for reliability and validity for this application. The sample sizes are extremely small, apparently ranging from about five to 12 in the six cells of the design represented in Table 1. Since respondents were arbitrarily divided into two groups on the basis of their pretest scores, there is the slim possibility of artifactual results caused by "regression toward the mean" (Campbell and Stanley, 1966). The curves in Figure 2 which we are to compare with the Sherif predictions of Figure 1 are not truly curves at all since they are based on only two data points. Finally, and most importantly, there is no statistical analysis so we have no way of knowing whether the differences reported in the paper are true differences in a statistical sense.

None of these deficiencies in the paper are serious in themselves. But, taken together, they weaken my faith in Newman and Dolich's conclusions. This is unfortunate because unlike recent papers in this area (e.g. Lastovicka and Gardner, 1978, Rothschild and Houston, 1977) this study is experimental and observes actual response rather than just consumer opinion about their response. Also it is one of the first experimental studies of this sort using a high involvement communication (another example: Swinyard and Coney, 1978). What can be said, however, is that the results do fit theory, and the method should be used again with the corrections implied above.

Gr°nhaug and Kangun

This paper is both more and less than a study on involvement. It is "more" in the sense that it offers non-American data on a socially-important issue in consumer behavior. It is "less" in the sense that we have to assume that smokers and non-smokers and those who have tried to stop actually differ in involvement. Gr°nhaug and Kangun would not, I believe, presume that this difference truly exists. They would probably agree that smokers and non-smokers differ on many more variables than involvement. The authors cannot be blamed for the placement of their paper in a session on involvement. Nor can I be blamed for having difficulty in discussing it in this regard. Perhaps our program chairman, William Wilkie, is the person we should blame. I certainly wouldn't do this, however.

At the opportunity of being repetitive, let me say that this paper, like the previous two, suffers because measures are of unknown reliability and validity. For example, the results on "awareness" which are given much importance by the authors, are based on a recognition measure. It has long been established (e.g. Wells, 1964) that recognition measures lead respondents to give subjective estimates of the probability that they were exposed to some communication rather than their true "awareness" of it. So when Gr°nhaug and Kangun find, to their surprise, that smokers and nonsmokers are equally aware of statements on smoking and health, they are simply replicating the biased sort of result that has been found with recognition measures through the years.

Finally, in either reanalysis of the present data or in future studies in this area, Gr°nhaug and Kangun should consider the sort of involvement measure proposed by Rothschild and Houston (1977). The results would probably indicate both high and low involvement segments.

IMPLICATIONS

Involvement was introduced to consumer research by Krugman to explain the effects of television advertising. Although he presented no direct data to support his hypothesis, subsequent experimental studies (e.g. Ray, et al., 1973, Ray, 1976, Ray and Webb, 1976, Rothschild, 1974, Swinyard and Coney, 1978) indicate that advertising and particularly television advertising operates the way Krugman said it does, partially because of involvement but also because of a variety of other factors including the nature of the medium, the perceived degree of brand differentiation, the sources of information most important in decision-making, and the degree of experience consumers have had with the product category.

Nothing in the present set of papers or any others I've seen recently on involvement in consumer research convinces me that a single measure of involvement will be found for all consumer research situations, that involvement alone can be used to explain any consumer behavior phenomenon or that even in that limited role it can do anything more than explain communication-advertising effects.

Measures and applications of involvement should be developed in individual consumer research application situations. An excellent example of such development of the concept in the political campaigning area was done by Rothschild (1974, Rothschild and Ray, 1974). He based his work on Krugman's ideas, political voting studies and the results of a large scale multiple measures study of political attitudes and behavior (see a report in Heeler and Ray, 1972). Recently his conceptualization was upheld in a study by Swinyard and Coney (1978). But Rothschild and Houston's attempt to find a more general measure of involvement produced less clear-cut results.

Another exemplary use of involvement was by Webb (1978, Ray and Webb, 1976) in the area of television clutter. Using Krugman's "connections" measure of involvement he selected commercials that were high and low in involvement. He then used these commercials in an experimental study with variations of television clutter. He found that the high involvement commercials (which probably scored high on the connections measure for a variety of reasons other than "pure" involvement) were much less affected by the clutter conditions than were the low involvement commercials. While this study may not contribute a great deal to the "theory" of involvement, it does provide valuable information to those who are attempting to develop advertising for a difficult television environment. And here Webb used a measure of involvement which, while imprecise, was developed by Krugman for use with advertising and which could gauge all the important aspects of involvement for this situation of interest. And I believe that Webb's experiment provides some reliability (in the prestudy) and validity information on the measure and concept precisely because the identified advertisements operated the way Krugman would say they would.

This discussion reflects trends that are obvious in consumer research and should be applied in the development and use of the involvement concept. First, consumer behavior is highly situational. Involvement up to now has been shown to work in one broad type of consumer situation: information processing of communications, particularly advertising. Successful uses of the concept have been careful long-term efforts in particular communication-advertising situations.

Second, the causes of consumer behavior are multivariate. Involvement has a number of components depending on the situation, and involvement alone will not explain consumer behavior. Other mediating variables are necessary.

Third, reliability and validity testing in consumer research is inadequate, although it is improving. The papers in this session, although sometimes weak in measures, provided some experimental evidence of the validity of the involvement concept and its measures.

Fourth, it is better to observe consumer responses in experiments than to ask consumers to guess how they would respond. And the experimental method is a good one for obtaining validity indications for concepts to be used in consumer research.

REFERENCES

Donald T. Campbell and Julian Stanley, Experimental and Quasi-Experimental Designs for Research (Chicago: Rand-McNally, 1966).

Roger M. Heeler and Michael L. Ray, "Measure Validation in Marketing," Journal of Marketing Research, 9 (November 1972), 361-71.

Herbert E. Krugman, "The Impact of Television Advertising: Learning without Involvement," Public Opinion Quarterly, 29 (Fall 1965), 349-56.

John L. Lastovicka and David M. Gardner, "Components of Involvement," In John C. Maloney and Bernard Silverman (Eds.) Attitude Research Plays for High Stakes (Chicago: American Marketing Association, 1978).

John C. Maloney and Bernard Silverman (Eds.) Attitude Research Plays for High Stakes (Chicago: American Marketing Association, 1978).

Michael L. Ray, "Attitude as Communication Response," In Deborah Johnson and William D. Wells, Attitude Research at Bay (Chicago: American Marketing Association, 1976).

Michael L. Ray, "Measurement and Marketing: Is the Flirtation Going to Lead to Romance?" Journal of Marketing Research, 16 (February 1979).

Michael L. Ray, Alan G. Sawyer, Michael L. Rothschild, Roger M. Heeler, Edward C. Strong, and Jerome B. Reed, "Marketing Communication and the Hierarchy of Effects," In Peter Clarke (Ed.) New Models for Mass Communication Research (Beverly Hills, Ca.: Sage Publishing, 1973), 147-76.

Michael L. Ray and Peter H. Webb, Experimental Research on the Effects of Television Clutter: Dealing with a Difficult Media Environment (Cambridge, Mass.: Marketing Science Institute, 1976), Report No. 76-102.

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

Michael L. Rothschild, "The Effects of Political Advertising on the Voting Behavior of a Low-Involvement Electorate," unpublished Ph.D. dissertation, Graduate School of Business, Stanford University, 1974.

Michael L. Rothschild and Michael J. Houston, "The Consumer Involvement Matrix: Some Preliminary Findings,'' In Barnett A. Greenberg and Danny N. Bellenger (Eds.) Contemporary Marketing Thought (Chicago: American Marketing Association, 1977), 95-8.

Michael L. Rothschild and Michael L. Ray, "Involvement and Political Advertising Effect: An Exploratory Experiment,'' Communication Research, 1 (July 1974), 264-84.

Muzafer Sherif and Hadley Cantril, The Psychology of Ego Involvements (New York: Wiley, 1947).

William R. Swinyard and Kenneth A. Coney, "Promotional Effects on a High- versus Low-Involvement Electorate," Journal of Consumer Research, 5 (June 1978), 41-8.

Peter H. Webb, "Consumer Information Acquisition in a Difficult Media Environment: The Case of Television Clutter," Unpublished Ph.D. dissertation, Graduate School of Business, Stanford University, 1978.

William D. Wells, "Recognition, Recall, and Rating Scales," Journal of Advertising Research, 4 (September 1964), 2-8.

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