Marketing Applications of Intentions Data


C. Joseph Clawson (1972) ,"Marketing Applications of Intentions Data", in SV - Proceedings of the Third Annual Conference of the Association for Consumer Research, eds. M. Venkatesan, Chicago, IL : Association for Consumer Research, Pages: 522-525.

Proceedings of the Third Annual Conference of the Association for Consumer Research, 1972      Pages 522-525


C. Joseph Clawson, University of Southern California

It is always a pleasant change of pace for a researcher, just as it is for an actor, to switch from the role of performer to that of critic. How sweet to dish it out with lofty condescension or humorous derision! How bitter to have to take it with open mind and dazzling smile, while hiding clenched teeth and murderous heart!

That is why Bob Pratt's invitation to spend part of my time evaluating the Granbois-Summer paper and the Suh paper at today's meeting was so appealing. However, he also invited me to play the performer role a little too, by offering some thoughts on the marketing applications of intentions data. In this schizoid role, it is my purpose to suggest some criteria that marketing managers can--and some do--apply in evaluating consumer intentions data. These thoughts will be interspersed with illustrations taken from the two papers just presented to you, but are not confined to them.

The marketing manager's intention to use intentions data appears to be directly related to the number of "Yes" answers he can find to five crucial questions in his mind. Let us examine these questions as criteria for marketing applications.


This issue really breaks down into two sub-issues. First, "Do we really need consumer sales forecasts of any sort, at this time, in our company?" Second, "If so, should we obtain intentions data in making those forecasts?"

Many companies can honest}y answer either or both of these questions with a "No." To save time, we will not inquire when and why this is the right answer, but will assume we are talking today only about marketers who can answer both questions affirmatively.


That is, do they measure "true" intentions--at least as of the time they are expressed? This is a technical question, but it is appropriate because marketing applications hinge upon the answer.

For instance, is the sample size large enough to give reasonably accurate measurements of all the variables? Neither of today's studies presents the confidence limits for its sample size. However, this is not too serious an oversight for our purposes. The Granbois-Summers sample of 77 married couples must be large enough because it gives clearcut findings! The Suh sample of 230 couples may be too small because it yields disappointing results.

What about bias? Granbois and Summers have been careful to avoid suggesting products to their respondents, by letting them originate their own lists of items they would consider. However, this systematically excludes just about everyone who would have assigned a 0 or .1 probability of purchasing these same items if they had been presented on a predetermined list. Thus, it inflates all the predicted and actual purchase rates, so they could not be used for estimating the magnitude of future durable goods demand in any larger population, as the authors carefully admit. Dr. Suh does use a predetermined list. However, he provides data here only- on one of the product classes, living room furniture.

As for scaling, both studies utilize an 11-point scale to measure strength of intentions. However, the Suh report first praises the superiority of such a scale over simpler ones, then promptly abandons his 11-point scale and converts it into a 2-point scale, intenders vs. non-intenders. While such a split may be required by multiple discriminant analysis, it blurs the entire picture and no doubt contributes to his disappointing results.


Professors Granbois and Summers were trying to predict purchases by measuring intentions, whereas Dr. Suh went further upstream, hoping to predict intentions with 20 prior influences. However, we can still ask in both cases if they predicted well, and if they measured all of the truly important factors that influence the outcomes.

The Granbois-Summers project confirms, once again, the familiar positive correlation between strength of intentions and actual purchase rates. They also report what I consider very worthwhile, interesting, but not overwhelming effects arising from age, sex, income, wife's working status, and joint E decision-making. However, their substantial discrepancies between expected and actual purchase rates point up the basic difficulty in predicting behavior 12 months ahead. Too many changes take place in the main predictor variable, intentions, as the months slowly roll by.

The Suh study neatly sidesteps the time-lapse problem by measuring 20 predictor variables and one criterion variable-- intentions--as of the same time. However, as Dr. Suh sorrowfully but manfully confesses, it may be necessary in the future to select different, more relevant variables. The ones he employed just did not give clearcut, significant overall predictions as between intenders and non-intenders.

One moral for all of us, I believe, is that we should measure changes in purchases, changes in intentions, and changes in causal factors in successive, reasonably short periods of time. Another moral is that we need to discover . and use a broader range of genuinely causal variables, such as motives, attitudes, social pressures, in-store promotions, competitors' prices, and de-emphasize the use of the remote influences, such as general personality traits, and "proxy" variables, such as demographics. Of course, if good theory suggests that specific traits and specific demographics are causal, that is another matter. Kassarjian has argued this point well in the November 1971 issue of the Journal of Marketing Research.


The marketing manager also wants to know if he can take any action whatever based on intentions data, or if can only say, "Right on!" or "No way"' as he passively files them away.

One way we can make intentions data more useful is to relate them to the types of consumer choices that a given marketing manager has the best chance of being able to influence. For instance, I would assume that the marketing manager for General Electric television sets can surely do more to change GE's brand share within the industry- than he can to change overall demand for television sets in general.

Yet most published studies concentrate on product class intentions--as in today's papers, the University of Michigan studies, the Bureau of the Census reports, and- yes--in my own article in the Journal of Marketing for September 1971. Why don't we research the best ways to use intentions data to predict brand choices, store choices, feature choices, timing choices, quantity choices, And so on?

Another way to make intentions data more actionable is to select the right time period to cover in the forecast. We should not overlook, for instance, that even very short periods, such as two weeks or three months can be long enough for many companies to take compensatory action. While unable to change the predicted sales level, they can adjust to it by shifting their production rates, inventory size, overtime, local advertising budget, and so on. However, if our management is only concerned with the need for remedial or aggressive actions to change the predicted sales level itself, then of course the planning period should be six months, a year, or five years--whatever period is affected by long-range commitments in new plant and equipment, investment in R & D, and so on.


Guiding action is not the same as alerting action, but intentions data could do both if tied to certain other facts. In fact, the potential pay-off multiplies as they are added.

Of course, even at the bottom rung of this pay-off ladder, where intentions data are completely unadorned with supplementary information, they can at least alert marketing managers to their need to consider possible action. However, the matters of "What," "To whom," "How," and "Whether" are left blank, for them to fill in.

Rising to the second rung, as both of today's studies do, for example, we add demographics like sex, age, area, income, and so on. Such data can actually help managers pinpoint the most fertile market segments, identify problem categories requiring attention, and abandon hopeless situations. This leaves only "What," "How," and "Whether" unanswered.

So to reach the third rung, we can add data on attitudes, motives, confidence, awareness, and other variables that hopefully determine the intentions, as the Suh study commendably tries to do. Here the data become diagnostic. As such, they provide ideas for what needs to be changed, such as price, advertising copy, product features, and so on. The managers, however, must still create their own answers to "How" and "Whether."

The fourth rung on the ladder may be achieved when we include "contingency intentions," Our consumers are asked for their purchase probabilities under various hypothetical conditions, such as a proposed change in product, package, retail outlets, advertising theme--or even imaginary changes in competitors' efforts, personal income, taxes. We still need to study the best ways of getting valid contingency intentions and testing them under controlled conditions. However, if he had them, the marketing manager could be greatly assisted in selecting the right marketing mix for reaching any desired sales level.

Finally, the top level of potential usefulness comes into sight, at least, when the different marketing mixes that we derive from contingency intentions are matched against internal cost figures. Combined, the manager can come closer to picking the optimum mix, assuming anyone cares about profitability or return on investment! There, I think, is a real challenge to the producers and users of intentions data.

In conclusion, I hope that all of us who work with intentions surveys can remember to match them to the needs of our own customers, the marketing executives. We can thank the authors of today's two papers, and more generally, we can be proud of the practical marketing applications of today's intentions data, which are already valuable. However, the untapped potential is far greater.



C. Joseph Clawson, University of Southern California


SV - Proceedings of the Third Annual Conference of the Association for Consumer Research | 1972

Share Proceeding

Featured papers

See More


Trust the Polls? Neural and Recall Responses Provide Alternative Predictors of Political Outcomes

Samuel B Barnett, Northwestern University, USA
Andres Campero, Massachusetts Institute of Technology, USA
Ronen Zilberman, Massachusetts Institute of Technology, USA
Chris Rose, New York University, USA
Aaron Robinson, Northwestern University, USA
Moran Cerf, Northwestern University, USA

Read More


Q7. Desire in Performed Consumption: Examining the Case of Korean Beauty Vlogging

Marie-Eve Jodoin, HEC Montreal, Canada
Marie-Agnès Parmentier, HEC Montreal, Canada

Read More


A Rational Model to Predict Consumers’ Irrational Behavior

Vahid Rahmani, Rowan University

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.