Cue Choice As a Function of Time Pressure and Perceived Risk

ABSTRACT - Cues are used by people when forming beliefs about brands which in turn influence brand selection. To predict and understand why people exhibit particular brand choice behavior it is necessary to understand why people utilize certain types of cues. This paper describes a study which examined the effect of time pressure and perceived risk on people's choice between two types of cues (cues high in predictive value but low in confidence value and cues low in predictive value but high in confidence value). The experiment had a significant main effect for time pressure and a significant interaction term.

Citation:

Douglas A. (Tony) Schellinck (1983) ,"Cue Choice As a Function of Time Pressure and Perceived Risk", in NA - Advances in Consumer Research Volume 10, eds. Richard P. Bagozzi and Alice M. Tybout, Ann Abor, MI : Association for Consumer Research, Pages: 470-475.

Advances in Consumer Research Volume 10, 1983      Pages 470-475

CUE CHOICE AS A FUNCTION OF TIME PRESSURE AND PERCEIVED RISK

Douglas A. (Tony) Schellinck, Dalhousie University

[The author thanks Jagdish Sheth, David Gardner and Russell Belk for their guidance in conducting the research reported in this paper.]

ABSTRACT -

Cues are used by people when forming beliefs about brands which in turn influence brand selection. To predict and understand why people exhibit particular brand choice behavior it is necessary to understand why people utilize certain types of cues. This paper describes a study which examined the effect of time pressure and perceived risk on people's choice between two types of cues (cues high in predictive value but low in confidence value and cues low in predictive value but high in confidence value). The experiment had a significant main effect for time pressure and a significant interaction term.

INTRODUCTION

To date there has been little research into the determinants of cue choice behavior. Considerable research has been conducted on what cues people tend to choose, but not why they choose them (Bettman and Jacoby 1975, Jacoby, Chestnut, Weisl and Fisher 1975, Jacoby, Szybillo and Bersato-Schach 1977, Kohn and Jacoby 1973, Olson and Jacoby 1972, Pincus and Waters 1975). Two rain points become clear from this research. First, people generally attend to less than the full number of cues that are available to them when making a judgement. Second, they tend to consider brand name and intrinsic product cues (cues that cannot be altered without altering the object) when assessing a brand's quality.

This raises the questions of how the consumer selects from the available set of cues and why do they often rely on cues which they themselves admit are likely nonpredictors of the characteristics of the brands under consideration? The purpose of this research is to test the impact of two possible determinants of cue choice behavior.

A cue is defined as a characteristic, event, quality or object, external to a person, that can be encoded and used to categorize a stimulus object (Schellinck 1980). There are hundreds of potential cues that can be associated with any particular object. What is required is some means of categorizing or describing cues such that the determinants of choice among these various types of cues can be studied.

Several taxonomies are used when discussing cues, most of them similar but derived from different fields of study. These include intrinsic/extrinsic (Andrews and Valenzi 1971, Olson and Jacoby 1972), figure/ground (for example, Tversky & Kahneman, 1980), descriptive/behavioral (Sarbin, Taft and Bailey 1960) and significative/ symbolic (Howard and Sheth 1969).

Rather than categorizing cues they can be characterized. Again labels have come from different fields of study, some which essentially describe the same characteristics. Descriptors in cue research include the obviousness and the relevance of the cue (Archer 1962), cue redundancy (Rollings, Bethel and Deffenbacker 1971, Einhorn, Kleinmertz, and Kleinmertz 1979), communality of a cue (Slovic and MacPhillany 1974) the ecological validity of a cue (Brunswick 1949, Vinacke and Arkoff 1957, Kelly and Arrowood 1960), the generalizability of cues (Brim 1954, French and Snider 1959), the judgmental confidence and validity of a cue (Bruner, Goodnow and Austin 1956) and the confidence value and predictive value of cues (Cox 1967, Olson and Jacoby 1972).

Many of these categorization schemes could be used, but the cue sorting rule developed for the study of buyer behavior by Cox (1967) uses predictive and confidence value. Subsequent research by Olson and Jacoby (1972) showed that this scheme of characterizing cues is useful in predicting which cues people will use in evaluating brands.

This cue characterization scheme is particularly appropriate for studying buyer behavior from a consumerism point of view. Consumer advocates tend to insist on making cues available to buyers that are high in predictive value under the assumption consumers will use these cues. However, consumers often ignore these cues. If it can be shown the consumer's choice of cues is influenced by a cue's confidence value under some circumstances than consumerists and government agencies which regulate consumer information requirements may be able to take this fact into consideration. Insight may be rained into this "irrational" behavior.

The study reported here examined the use of cues with varying amounts of confidence value and predictive value. Cox (1967) defined predictive value as the perceived probability with which a criterion category is associated with an attribute category and confidence value as the perceived probability with which each cue can be correctly sorted into a criterion category by an individual.

The work on cue usage cited earlier indicates that people use cues such as brand name, country manufactured in or price, even when the consumer knows they are not particularly predictive of quality. Cox identified other cues (the odor of stockings, indicating wearing quality; the color of clothing detergent boxes, indicating cleaning power; the loudness of electric motors, indicating their power; and the color of cooking oil indicating its thickness) as examples of cues which people use but which have low predictive value. He feels these cues are used, even though they are low in predictive value, because they are high in confidence value relative to other cues available.

Olson and Jacoby (1972) showed that consumers will first choose cues with high predictive value and high confidence value when these are easily accessible. After these cues are encoded and the consumer still has too much uncertainty to make a decision they must choose cues with higher predictive value and lower confidence value or cues with lower predictive value and higher confidence value. The question is, what factors will cause them to choose one type of cue over the other.

Choosing a cue that is strong in one dimension and weak in another has its own type of risk associated with each type of cue. With high predictive value/low confidence value cues there is a higher risk the decision maker will misinterpret the data completely and thus make a wrong choice. However, if the data is interpreted correctly the chances of choosing an excellent brand are better when using high confidence value/low predictive value cues. Thus, the use of high predictive value/low confidence value cues is associated with higher risk but potentially greater payoff.

Choosing a high confidence value/low predictive value cue means there is less chance of incorrectly identifying what, according to the cues, is supposed to be the best brand. But the chances that the cues indicate the best brand are less. Thus, the risk of misinterpreting the cues is less, but the potential payoff is less.

It is hypothesized that when people are highly motivated to pick a good brand they will have a tendency to pick high predictive value/low confidence value cues over high confidence value/low predictive value cues. The buyers are motivated to put in the effort necessary to carefully analyze and interpret the cues. They feel the need to use cues that will maximize their chances of making a good choice. If people are not motivated to pick an excellent brand, or they are faced with a constraint that limits their ability to analyze the cues, they will have a tendency to obtain confidence value; that is, they will tend to pick high confidence value/ low predictive value cues over high prediction value/low confidence value cues.

The purpose of this study was to test these general hypotheses by examining the effect of a motivator and a constraint on the buyer's choice of cues. The motivating factor was perceived risk and the constraint was tine pressure. Specifically it was hypothesized that:

1. As time pressure increases there will be a greater tendency to depend on high confidence value/low predictive value cues.

It was hypothesized that, given the constraint of time pressure, people will feel they don't have the time to spend carefully analyzing high predictive value/low confidence value cues. They may see the value of using high predictive value/low confidence value cues, but feel the risk of misinterpreting the cues is too high and they don't have the time to carefully analyze the cues and reduce this risk. Using high confidence value/low predictive value cues is likely viewed as more time efficient. While the chances of making the best choice are reduced, at least they will more likely make a satisfactory choice.

2. As perceived risk increases there is a tendency to depend on high predictive value/low confidence value cues.

It is hypothesized that as perceived variation in brand performance increases purchasers feel they have to depend more on high predictive value/low confidence value cues in order to identify the better brands. This means that, because performance varies so much, they are more likely to be able to discover significant differences in cue values and therefore identify what is a good brand and what is a poor brand. Thus the chances of misinterpreting the cues seem reduced.

Also, with the high performance uncertainty the chances of picking an unsatisfactory brand are increased. If the cost of making a wrong decision is very high purchasers will be more motivated to analyze the cues and will more likely pick high predictive value/low confidence value cues.

Therefore, the risk of misinterpreting the high predictive value/low confidence value cues is perceived as less, and the potential payoff in terms of not making a wrong choice, is increased, causing purchasers to depend more on high predictive value/low confidence value cues.

METHOD

The study was conducted in two phases. Phase one involved three surveys measuring the perceived predictive and confidence value of cues that might be used in forming a judgement on the relative quality of various kinds of pagers (or "beepers" as they are sometimes called). The questions were based on those used by Olson (Olson and Jacoby, 1972). In order to measure a cue's predictive value, subjects responded on a one to five scale. with one labelled "not at all accurately" and five labelled "extremely accurately" how much they felt "the following types of information accurately indicate the overall quality of rand of paper?" Subjects then indicated their perceived confidence value on the same set of cues. Subjects responded on a one to five scale with one labelled "not at all confident" and five labelled "extremely confident" how confident they felt they knew "what each of the following terms mean?"

The first two surveys identified a list of eight cues that could be used in phase two, four which were perceived as high in predictive value but low in confidence value and four which were perceived as low in predictive value and high in confidence value. The final survey used 50 subjects from the subject pool (students enrolled in an introductory organizational behavior class) to confirm the perceived confidence value and predictive value for these cues, see figure 1).

Phase two employed a 2 X 2 factorial experiment with two levels of time pressure and two levels of risk. Fifty two subjects participated in each of the four possible conditions.

Up to eight subjects participated in the experiment at a time. They sat at separate tables, each with one or two booklets made of cardboard pages with eight tabs on each page. Each page had eight cues on a brand with the brands labelled only by the numbers 1-5. Each tab was labelled as a cue, such as "frequency stability," and the subjects were told they could learn the information about that particular brand by pulling the tab and reading what was written underneath.

The task itself involved the subjects hearing a short lecture on the history of pagers and something about how they operate. They then individually selected the cues, selecting only four of the eight cues available for each brand and then recording on a separate page the order they picket the cues as they proceeded from brand one to brand five. Having selected their cues they could go back and analyze the cues if they wished and then decide on the brand they judged had the highest quality. Since they were not given the price of the alternative brands they could not estimate the relative value and were therefore not asked to choose the pager they would buy. To further control for any price effects, such as subjects assuming that pagers mate in Taiwan are cheaper than pagers mate in the L'.S.A., they were told beforehand that all brands are priced approximately the

FIGURE 1

THE RELATIVE POSITION OF THE FIFTEEN CUES WHEN PLOTTED ACCORDING TO THEIR MEAN PREDICTIVE VALUES (HORIZONTAL AXIS) AND MEAN CONFIDENCE VALUES (VERTICAL AXIS).

Pagers were chosen because it is a product which few people are familiar with. This is important for two reasons. First, familiarity with the product class may be an important determinant of cue choice and usage. By using a product nobody is familiar with (or eliminating those subjects that are familiar with it) familiarity is controlled in the experiment.

Second, in order to manipulate perceived risk it was necessary to quote two very different stories on the performance and average price of the various brands available to consumers. If subjects were unfamiliar with the product there was little chance that they would detect the deception. To further control for familiarity the brand names of the pagers being appraised were not included in the set of cues. Some subjects would possibly be familiar with these brand names, perhaps even having experience with other products from the same manufacturer.

Another reason for using pagers was that the product is technical in nature. Thus, it is reasonable to assume that subjects would want to use technical cues to evaluate the brands. It was assumed that these cues would be viewed as having high predictive value and low confidence value by the subjects.

The main dependent variable in this study was the number of high confidence value, low predictive value cues picket by a subject. A subject picket four cues for each of five brands giving the variable a range from O to 20. The main dependent variable could equally have been the number of low confidence value, high predictive value cues since the sum of the two types of cues used would always equal twenty.

The main dependent variable measured cue choice since cue choice was the primary focus of the research. This was because analyses of cue choice was a logical first step to analyzing the roles of cues in the decision process. Also, from a marketer or consumerist's point of view, getting a buyer to attend to particular cues is the necessary first steps to influencing consumer choice .

However, rather than have the experimental subjects simply choose cues it was decided they should continue the choice process by evaluating the brands and picking the brand they felt was the best one. Analysis could then examine the possible impact of cue choice and the experimental conditions on the-use of the cues once they were picked. Since this analysis was added on for exploratory purposes no hypotheses were developed prior to the experiment. ' However, it was felt that time pressure and perceived risk might further influence cue usage such that the effect of cue choice is magnified or diminished.

The cues people actually use to appraise a product have in past research been identified by regressing the cue values against the subject's ratings of the objects to determine which cues predict the ratings (e.g., Slovic and MacPhillany 1974). This was not possible in the present task as there were not enough observations to allow regression analysis, particularly since many of the cue values were categorical and would have to be converted to dummy variables. Instead, immediately after selecting the brand of pager they thought was the best quality, subjects were asked to consider their thinking process and list the cues they actually used in arriving at their decision. The second set of dependent variables was derived from this question. One was the number of high confidence value, low predictive value cues they said they actually used in selecting the brand, the other the number of high predictive value, low confidence value cues they used. These variables could theoretically range from zero to four for both types of cues for each subject. However, unlike the main dependent variable above, their sum does not have to be constant and therefore both variables were examined during the analysis.

The third dependent variable was the brand judged as being the highest quality. The cue values were such that the best choice using high predictive value cues was different from the choice using the high confidence value cues. If the hypotheses were correct, the distribution of brands judged as the highest quality would change in a predictable direction. For example, pager three had a good user testimonial, was made in England (as opposed to Taiwan and Japan) was sold by Sears (as opposed to Zayre and Ayr-Way), and had by far the largest market share. 'On the other hand, it had the lowest frequency range, the worst paging sensitivity, the second worst adjacent channel selectivity and was tied with four other brands for frequency stability. Thus, subjects using high predictive value cues should not have picket this brand while subjects using high confidence value cues should more often have picked it.

If the hypotheses were correct, pager three should be more often picked under conditions of high time pressure and low risk. Because this was an indirect measure of the impact of the manipulations in cue choice it was reserved as a secondary dependent variable. However it was felt considerable more importance could be accredited to the findings if it was shown that a manipulation not only changed cue choice behavior but that this caused a change in the judgement made by the subject.

Time pressure was manipulated by having the high time pressure group evaluate pagers and then cameras (using a similar booklet) in the time it took the low time pressure group to evaluate pagers. Evaluating cameras after pagers insured that the high time pressure group finished their task in the allotted time. Pretests were used to determine the optimal times-for the two treatment levels based on the point spread in the manipulation checks and discussion with participants concerning their reaction to the time Pressure.

Risk was manipulated by having the high risk level groups read a quote from Consumer's Reports and another from a government study (both fictitious) stating that all pagers tend to perform well in tests and that the chances of buying one that doesn't perform well are minimal. The high risk group was told that several brands did not perform well and the chances of buying one that doesn't perform satisfactorily were considerable. The credibility and impact of these reports had been tested in pretests.

Risk was also manipulated by telling the subjects that the pagers cost about $20.00 in the low risk situation and about $450.00 in the high risk situation.

RESULTS

The manipulation checks showed strong differences in perceived time pressure (p c.0001) and perceived risk (p <.0001). The time pressure main effects for cue choice was significant, F=5.51, (p=.02) and the time pressure X risk interaction was borderline significant at F=3.73, p=.055 (See figure 2). The risk main effects were not significant. A Tukey's HSD test between means indicated a significant difference between the mean for the high time pressure, low risk condition (X=10.48) and the low time pressure, low risk condition (X=7.9) at p<.05. Total variance explained was W2=4.18.

FIGURE 2

AVERAGE NUMBER OF HIGH CV CUES

FIGURE 3

TOTAL NUMBER OF HIGH PV CUES SUBJECTS INDICATED THEY USED TO EVALUATE THE PAGERS

The second dependent variable was the number of high predictive value, low confidence value cues the subjects said they used in the decision. The time pressure main effects was highly significant (p<.001), with the risk main effect and the interaction effect not significantly different (See figure 3). The variance explained was higher for this dependent variable at w2=.086.

The third dependent variable was the pager judged as being the highest quality brand. The distribution of brands was significantly different between the high time pressure and low time pressure conditions X2=10.35, p=.025 (See table 1). Subjects depending on high confidence value/low predictive value cues would be expected to more often pick brand three. As predicted, the percentage picking this brand rose from 42.9% to 62.2% from the low time pressure condition to the high time pressure condition (p-.005). Subjects depending on low confidence value, high predictive value cues would be expected to more often pick brand five. As predicted, the percentage of subjects picking this brand rose from 14.4% to 25.9% from the high time pressure condition to the low time pressure condition (p-.025).

DISCUSSION

Time pressure appears to affect the person's choice of cues and the person's dependence on cues chosen. The cumulative effect has been demonstrated to cause a change in the person's judgements. Subjects in the high time pressure condition that indicated they had experienced considerable time pressure were those who tended to not use the high confidence value cues (v- 9.13). Those who indicated they had experienced little time pressure depended more on high confidence value cues (x- 12.29). This suggests that using high confidence value cues is perceived as a time saving strategy, causing the change in cue choice and cue usage behavior.

This effect might have been even more pronounced if the cues had been characterized by the amount of cognitive effort required to decide on the appropriate category for the object on the attribute of interest, and this measure used as the dependent variable rather than the received confidence value of the cue.

TABLE 1

BRAND SHARE UNDER DIFFERENT TIME PRESSURE TREATMENT LEVELS

The results shed some light on how perceived risk will effect cue choice behavior. In this experiment, the subjects had two types of cues to choose from. They could have chosen cues they felt were less predictive of quality, but cues they felt confident they understood. Understanding the cue would presumably allow them to better evaluate the object. leading to a better choice.

A second strategy would be to choose the cues they felt had high predictive value, but were cues they did not feel confident they understood. This would force them to somehow learn what was, or was not, a good value for a particular object. They could never be sure whether they were classifying the object correctly, but would feel their final choice was better since it was based on cues which they felt were better predictors of quality.

In the experiment it appears subjects preferred to depend more on the latter strategy, even in the low risk situation. 'Only the introduction of considerable time pressure forced them to use the high confidence value cues as a time saving strategy.

The amount of variance explained by time pressure and risk manipulation is not large. However, it appears that variance explained understates the amount of impact time pressure can have on the process since the proportion of people judging brands three and five as being the best quality changed drastically between conditions. If this could be translated directly into market shares (assuming people buy the brand they feel is the best quality) the impact on the firms selling brands three and five would be tremendous. These results indicate the considerable impact that cue choice behavior has on consumer judgement. Until the determinants of cue choice and usage are understood a larger part of the variance in the consumer decision process will not be explained. Time pressure, and to lesser extent perceived risk, were found to be determinants of cue choice.

REFERENCES

Andrews, I. R. and Valenzi, E. R. (1971), Combining Price, Brand and Store Cues to Form an Impression of Product Quality," Proceedings, 79th Annual Convention, APA, 649-50.

Archer, E. James (1962), Concept Identification as a Function of Obviousness of Relevant and Irrelevant Information," Journal of Experimental Psychology, 63, 6, 616-620.

Bettman, J. R. and Jacoby, Jacob (1978), "Patterns of Processing in Consumer Information Acquisition, Advances in Consumer Research, 3, 15-20.

Brim, O. G. Jr. (1954), "The Acceptance of New Behavior in Child-Rearing, Human Relations, 7, 473-491.

Bruner, J. S., Goodnow, Jackeline J. and Austin, G. A. (1956), A Study of Thinking, New York: Wiley.

Brunswick, E. (1949), Systematic and Representative Design of Psychological Experiments, Berkeley: University of California Press.

Cox, D. F. (1967), Risk-Taking and Information-Handling in Consumer Behavior, Boston, Harvard University Press.

Einhorn, H. J., Kleinmertz, D. N. and Kleinmertz, B. (1979), Linear Regression and Process-tracing Models of Judgement," Psychological Review, 86, 465-85.

French, J. R. P., Jr. and Snyder, R. (1959), Leadership and Interpersonal Power, in D. Cartwright (Ed.), Studies in Social Power. Ann Arbor: Univ. of Michigan Press, 118-149.

Howard, John A. and Sheth, Jagdish N. (1969), The Theory of Buyer Behavior, New York: Wiley.

Jacoby, Jacob A., Chestnut, Robert A., Weigl, Karl C. and Fisher, William (1975), "Pre-Purchase Information Acquisition: Description of a Process Methodology, Research Paradigm, and Pilot Investigation,' in Beverlee B. Anderson (Ed.), Advances in Consumer Research, 3, 306-16

Jacoby, Jacob, Szybillo, George J. and Busato-Schach, Jacqueline (1977), 'Information Acquisition Behavior in Brand Choice Situations,' Journal of Consumer Research, 3 (March) 209-16.

Kelly, H. H. and Arrowood, A. J. (1960), "Coalitions in the Triad: Critique and Experiment," Sociometry, 23, 231-44.

Kohn, Carol A. and Jacoby,'Jacob (1973), "Patterns of Information Acquisition in New Product Purchases," in Scott Ward and Peter Wright, (Eds.), Advances in Consumer Research, 1, 427-30.

Olson, Jerry C. and Jacoby, Jacob (1972), "Cue Utilization in Quality Perception Processes," in M. Venkatesan, (Ed.), Association of Consumer Research, Proceedings, 167-79.

Pincus, Steven and L. K. Waters (1975), Product Quality Ratings as a function of Availability of Intrinsic Product Cues and Price Information, Journal of Applied Psychology, 60, 280-82.

Rollings, Harry; Bethel, Barbara and Deffenbacker, Kenneth (1971), Relevant Cue Placement Effects in Concept Identification Tasks, Journal of Experimental Psychology, &7, 1, 9-12.

Sarbin, T. R.; Taft, R. and Bailey, D. E. (1960), Clinical inference and Cognitive Theory, New York: Holt, Rinehart, and Winston.

Schellinck, Douglas A. (1980), Determinants of Cue Choice Behavior, unpublished dissertation, University of Illinois.

Slovic, Paul and MacPhillany, Douglas (1974), "Dimensional Commensurability and Cue Utilization in Comparative Judgement,' Organizational Behavior and Human Performance, II, 172-194.

Tversky, A. and Kahneman, D. (1980), "'The Framing of Decisions and the Rationality of Choice, Department of Psychology, Stanford University.

Vinacke, W. E. and Arkoff, A. (1957), An Experimental Study of Coalitions in the Triad," American Sociological Review, 22, 406-14.

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