The Time and Outcome Valuation Model: Implications For Understanding Reactance and Risky Choices in Consumer Decision Making

John C. Mowen, Oklahoma State University
ABSTRACT - The time and outcome valuation (TOV) model (Mowen and Mowen 1991) was described and used as a metatheory to account for the effects of reactance and risk perception on consumers. A key assumption of the model, that negative outcomes are discounted more rapidly than positive outcomes over time, was successfully tested within the context of individuals making choices between risky or conservative bets. Implications of the model for understanding consumer decision making and for future research are identified.
[ to cite ]:
John C. Mowen (1992) ,"The Time and Outcome Valuation Model: Implications For Understanding Reactance and Risky Choices in Consumer Decision Making", in NA - Advances in Consumer Research Volume 19, eds. John F. Sherry, Jr. and Brian Sternthal, Provo, UT : Association for Consumer Research, Pages: 182-189.

Advances in Consumer Research Volume 19, 1992       Pages 182-189

THE TIME AND OUTCOME VALUATION MODEL: IMPLICATIONS FOR UNDERSTANDING REACTANCE AND RISKY CHOICES IN CONSUMER DECISION MAKING

John C. Mowen, Oklahoma State University

ABSTRACT -

The time and outcome valuation (TOV) model (Mowen and Mowen 1991) was described and used as a metatheory to account for the effects of reactance and risk perception on consumers. A key assumption of the model, that negative outcomes are discounted more rapidly than positive outcomes over time, was successfully tested within the context of individuals making choices between risky or conservative bets. Implications of the model for understanding consumer decision making and for future research are identified.

INTRODUCTION

Consumer decision making occurs within a world filled with risk and uncertainty. Researchers generally regard risk perception in terms of an expectancy-valuation model that is based upon two factors: (a) the likelihood that a bad outcome could occur and (b) the degree of negativity of the outcome should it occur (Dowling 1986). Thus, risk perception can be conceptualized as based upon the integration of a probability estimate and a valuation of an outcome in order to form an overall evaluation of risk.

Researchers have begun to investigate risk perception by analyzing separately the factors that influence the estimation of probabilities and the valuation of outcomes (e.g., Kahneman and Tversky 1979). Valuation occurs when an individual assesses the extent of the goodness or badness of a possible or actual outcome, event, or thing. The concept of valuation is central to many mid-range theories used to describe consumer behavior. For example, valuation is a critical element of the many expectancy-value models employed by researchers to describe various consumer processes. In such models the overall evaluation of a course of action is predicted by multiplying the likelihood of each of the various outcomes occurring by the value of each outcome should it occur and then summing across the outcomes. From the compensatory models of choice and attitude formation (e.g., Fishbein and Ajzen 1975), to models of risk perception (Bauer 1960), to subjective expected utility models (e.g., prospect theory, Kahneman and Tversky 1979), one finds valuation as a central element.

Mowen and Mowen (1991) developed a model, that focuses on the role of time on outcome valuation. Called the time and outcome valuation model (the TOV), it proposes that the distance in time between when a decision is made and when potential positive or negative outcomes occur systematically influences the perception of the goodness/badness of the outcomes. Derived from approach-avoidance conflict theory (Miller 1959), prospect theory (Kahneman and Tversky 1979), and discounted utility theory, the TOV acts as a metatheory, which proposes that a family of "time and outcome valuation" phenomena exist. Some of these phenomena include: risk aversion in the present, future optimism, individual traps and fences, speed-up costs, and delay charge effects. See Mowen and Mowen (1991) for a description of these phenomena.

The TOV proposes that depending upon situational circumstances, losses and gains can occur at different points in time relative to when a decision is made. Based upon when in time the losses or gains occur relative to the decision point, divergent decision biases may result. For example, when situational circumstances cause the negative outcomes of a decision to occur in the present and the positive outcomes of a decision to occur in the future, the individual fence (Platt 1973) is predicted to occur. An investment decision is a case in which an individual fence would impact consumer decision making. In an investment decision, the money or time required to make the investment is taken from the consumer's expendable income in the present causing a short-term loss. Only in the long-term will the benefits of the investment be felt. Because losses are overvalued in the present and the future gains are discounted, consumers will have a strong tendency not to make long-term investments unless the positive benefits in the long-run are extremely salient.

The TOV model is based upon a series of assumptions, which are summarized below. (See Mowen and Mowen, 1991, for the empirical basis for these assumptions.)

A1. The psychological reaction to outcomes is described by the decreasing marginal valuation of gains and losses.

A2. For outcomes that occur in the present, losses are valued relatively more heavily than gains.

A3. The net valuation of a prospect results from the difference in the valuation of gains and losses.

A4. Outcomes are discounted in time with losses discounted faster than gains.

A5. Time and outcome framing exist. The loss or postponement of a gain is framed as a loss. The avoidance or postponement of a loss is framed as a gain.

A6. Gains and losses may occur at different points in time.

The present paper has three purposes. First, it presents a method of graphing the model in a manner that allows for the derivation of decision phenomena from the TOV. Second, the paper extends the TOV model to account for reactance in consumer settings. Third, the paper presents an empirical study testing a critical assumption of the TOV, that gains are discounted more rapidly than losses.

FIGURE 1

THE TOV MODEL

A GRAPHICAL REPRESENTATION OF THE TOV

Figure 1 presents the graphical representation of the TOV found in Mowen and Mowen (1991). In the three dimensional figure, the x axis represents the actual value of a stimulus and the y axis the psychological value of the stimulus. These axes correspond to those found in prospect theory. The third axis depicts time moving from the present to the future. In the present, the curves match those found in prospect theory resulting in a heavier weighting of losses than gains. Thus, when equivalent gains and losses occur in the present, risk aversion is predicted to occur. However, as time passes, gains and losses are discounted so that the planes begin to approach the x axis. Because losses are discounted more rapidly than gains, the loss plane more rapidly approaches the x axis. As a result of the differential discounting of gains and losses, gains are weighted more heavily than losses when they occur far into the future. Thus, when evaluations occur long prior to the outcomes occurring, the psychological value of a gain may be greater than the psychological value of a loss of the same amount. Mowen and Mowen (1991) called this effect "future optimism."

FIGURE 2

TWO-DIMENSIONAL REPRESENTATION OF THE TOV MODEL AND THE DERIVATION OF PSYCHOLOGICAL REACTANCE

A problem with the representation of the TOV shown in Figure 1 is that it is relatively difficult to depict the changes in valuation that occur as gains and losses occur at different points in time relative to when the decision is made. Figure 2 portrays in 2-dimensional space the same relationships found in Figure 1. The curve labeled "Outcome in Present" is the classic prospect theory curve of Kahneman and Tversky (1979). The curve labeled "Outcome in Future" is the curve that results for the valuation of gains and losses when they occur far into the future (e.g., several years away). Note that the differential discounting of gains and losses results in the curve for future value being steeper in the gain domain than in the loss domain. As a result, the overweighting of losses that occurs in the present has been replaced by an overweighting of gains that occur in the future.

Figure 2 can now be used to depict visually the phenomena that can be derived from the TOV. The next section discusses how the phenomenon of reactance can be derived by using Figure 2.

Reactance Theory

Brehm (1969) in his development of reactance theory suggested that people have a need to maintain their behavioral freedom. If their behavioral freedom is violated, they will react against the threat in order to restore that freedom. Thus, if an option or possession is taken from them, a negative affective state is created, and they will reevaluate it so that it becomes relatively more attractive. Clee and Wicklund (1980) identified a number of situations in which reactance theory has implications for consumer behavior.

Reactance theory can be used to explain post-decisional regret (Walster 1964). Post-decisional regret occurs when, after making a choice between two positively valued alternatives, individuals may experience displeasure and reevaluate the alternatives such that the unchosen option is liked more than the chosen alternative. The TOV proposes that the regret is caused by a reactance state in which the buyer gives up all of the positive attributes of the unchosen alternative (i.e., he loses the freedom to obtain these positive qualities) while they accept all of the negatives of the chosen alternative. Mowen (1990) has suggested that what researchers call cognitive dissonance (i.e., the discomfort that may occur after a purchase) results from the cognitive imbalance that occurs from reactance causing the unchosen alternative to be valued more than the chosen alternative.

The TOV proposes that the negative feelings that occur in a reactance state result from the high weighting of losses taken in the present. That is, the loss of behavioral freedom to possess the desired attributes of the unchosen alternative is tantamount to the loss of any resource. According to Assumption 5 of the TOV, the loss of a positive outcome in the present will be framed in the loss domain and will create highly negative feelings.

The effect is diagramed in Figure 2. Let us assume that Option 1 (the chosen option) is rated originally as having +110 utiles and Option 2 (the unchosen option) as having +100 utiles. Thus, both options are viewed positively, but option 1 was chosen because the overall preference for it was higher. Once the person has been committed to the choice, however, the unchosen option is now framed as a loss of 100 utiles. When the loss is extended down to the curve that represents its psychological value in the present, one finds that the line segment O-L results. Now, consider the chosen alternative--Option 1. When the gain is extended up to obtain its psychological value, one finds that the line segment O-G results. Because of the greater valuation of losses than gains in the present, line segment O-L is longer than O-G. That is, the negative feelings resulting from the loss are stronger than the positive feelings resulting from the gain. As a result, the buyer experiences the overall negative affective state called reactance.

Linder and Crane (1970) performed a study that supports the TOV explanation of reactance theory. These authors asked subjects to rate the attractiveness of two people either of which could interview them for a job. The attractiveness was rated at different points in time prior to when a final choice had to be made. Linder and Crane (1970) predicted that reactance forces would tend to make the attractiveness of the choices converge as the time approached to make the choice. The results supported the prediction. That is, as the time for the final choice grew closer, the subjects increasingly rated the interviewers as similar in attractiveness.

The TOV accounts for the results of the Linder and Crane (1970) study by virtue of the discounting function for gains and losses. That is, as the outcome draws closer, the effects of gains and losses have increasing psychological value with the losses growing more rapidly. Because the inferior (but still positively evaluated option) is the one that the person is considering for elimination, the loss of its positive elements loom large. As a result, it will tend to increase in attractiveness, because of its potential loss. Thus, the TOV model predicts that the attractiveness of the two options will tend to converge.

In sum, the TOV model accounts nicely for the effects of reactance under the same rubric that handles a variety of other mid-range theories. A critical issue for the model, however, concerns the evidence for Assumption 4, that the discount rates for gains and losses diverge.

On the Differential Discounting of Gains and Losses

A key assumption in the TOV is that the discount rate is greater for losses than gains. Both Wright and Weitz (1977) and Jones and Johnson (1973) found indirect evidence that supports the assumption. In each study, the researchers found that people tended to make conservative decisions when outcomes occurred soon after the decision. This result is consistent with a greater weighting of the negative outcomes in the present. However, when outcomes were to be experienced further into the future (i.e., a week or more), riskier decisions were made. This result is consistent with the derivation that when gains and losses occur in the future, gains may be weighted more heavily than losses. As a result people move from a risk averse state in the present to a risk seeking state in the future.

The assumption that losses are discounted more rapidly than gains is controversial, however. Based upon the results of a carefully conducted study, Benzion, Rapoport and Yagil (1989) concluded that the discount rates "...are smaller for losses than for gains" (p. 282). In the Benzion et al. study, subjects gave an estimate of the amount of money that they would receive or pay in order to make them indifferent between a present and future outcome. For example, in the "postpone receipt" condition, the subjects might expect to receive $1,000 right now. They would then indicate the amount of money that would make them indifferent between getting the $1,000 now and getting $X a year from now. Similarly, in the "postpone payment" condition, they might learn that they must pay $1,000 now. They would then indicate how much they would pay a year from now that would make them indifferent between paying now or later. Supporting the previous findings, Benzion et al. (1989) found that the discount rate for "postponing a receipt" was greater than the discount rate for "postponing a payment."

The question is, "Do these results provide evidence for the greater discounting of gains as Benzion et al. suggested?" Their conclusion was derived from the finding that postponing receipts has more psychological impact than postponing payments. Thus, the researchers equated postponing a receipt with a gain and postponing a payment with a loss. This author suggests, however, that in fact subjects frame these situations in a far different way. That is, in accordance with Assumption 5, subjects will frame postponing a receipt as equivalent to taking a loss. (That is, the individual is losing the current use of the money.) Conversely, the postponement of a payment will be framed as a gain. (That is, the individual is gaining the current use of the money.) By recognizing that individuals will frame postponing a receipt as a loss and postponing a payment as a gain, the study provides strong evidence for the assumption that losses are discounted more rapidly than gains.

THE EXPERIMENTS

Study 1: Method.

The goal of each study was to independently manipulate when the potential gains and losses from a decision would occur in time. In each study subjects were asked to rate, on an eight-point scale, their preference for which of two bets they would prefer to play. One bet was conservative with a high likelihood of winning a small amount of money. The other bet was risky with a low likelihood of winning a much larger amount of money. In each study they read:

"On the next several pages, you will find a series of pairs of bets. We would like you to examine each pair of bets and to rate your preference for which you would prefer to play. In the study imagine that you are at a wheel of fortune. Imagine that after you choose each bet, the wheel is spun; and, depending upon where it lands, you will either win or lose based upon the probabilities given in the problem.

Subjects were given a practice problem. They were then asked to look through the entire questionnaire. Any questions that they had were answered. The twenty-five subjects in the study participated either alone or with one other person. They were paid $10 for participating.

Study 1 employed a 2x3x3x2 mixed, factorial design. The first three independent variables were taken as repeated measures. Thus, each subject made 18 choices. The fourth variable was a between subjects variable--the order of the presentation of the stimuli. One-half of the subjects answered eighteen, randomly ordered bets in forward order, and one-half answered them in reverse order. The independent variables were:

Factor A: Size of bet. Low--.80 to win $5 and .20 to lose $5 versus .20 to win $35 and .80 to lose $5. High--.80 to win $60 and .20 to lose $16 versus .80 to win $246 and .20 to lose $16.

Factor B: When gain occurs--immediately, 12 months, 48 months.

Factor C: When loss occurs--immediately, 12 months, 48 months.

Factor D: Order of receiving the 18 bets. The 18 bets were randomized and one-half of the subjects received them in forward and one-half in reverse order.

An example of a problem is shown below:

.80 to win $5 in 6 months                    .20 to win $35  in 6 months

                                   1 2 3 4 5 6 7 8

.20 to lose $5 in 4 years                      .80 to lose $5  in 4 years

The bets on the left side were always the conservative ones, and expected values were held constant between the conservative and risky bets. A second example of a bet is shown below:

.80 to win $60 in 4 years                     .20 to win $246  in 4 years

                                  1 2 3 4 5 6 7 8

.20 to lose $16 now                            .80 to lose $16 now

The second example shows the high priced bet. It also shows another combination of when gains and losses could occur. Note that the expected value of the risky bet in this condition was slightly lower ($40.00) than the conservative bet (44.80.)

Hypotheses. It was hypothesized that preferences for the bets would shift in the following way.

H1. A main effect for bet size would occur, such that higher cost bets would lead to greater preference for the conservative option.

H2. As gains moved further away in time, preferences would shift towards the conservative bet.

H3. As losses moved further away in time, preferences would shift towards the risky bet.

H4. The shift in preferences across time would be greater for losses than for gains.

Results of Study 1

Supporting H1, the results revealed a significant effect for the size of the bet (F=38.6, p<.0001, mean small bet = 5.5, mean larger bet = 4.8.) Subjects preferred the risky option to a greater extent when the bet was smaller. Hypothesis 2 was not supported. No effect was found for the manipulation of when gains occurred (F<1). The results did support H3. As expected, when losses occurred further into the future, subjects increasingly preferred the riskier bet (F=51.5, p<.0001, mean "now" = 4.4, mean 1 year = 5.3, mean 4 years = 5.7). H4 received indirect support from H2 and H3. That is, if no main effect was found for gains and a main effect was found for losses, then the discount rate for losses must be greater than that for gains.

In order to test H4 directly, the conditions in which gains and losses occurred at the same point in time (i.e., now, 1 year, 4 years) were compared. The means revealed a strong linear pattern in which increased preference for the risky alternative occurred as the outcomes moved farther into the future (mean now = 4.2, mean 1 year = 5.2, mean 4 years = 5.8.). (A priori F tests indicated that each mean was significantly different from the others.) This pattern of results could only occur if losses were discounted more rapidly than gains. In sum, the results strongly supported Assumption 4 of the TOV.

Study 2.

Study 2 replicated the first study while extending it in several respects. First, a sample of adults was obtained. Second, three different bet sizes were employed. Third, four levels of time when outcomes for gains and losses would occur were manipulated. Finally, subjects responded to all the bets twice so that the analysis could be performed on an individual subject basis.

Study 2 employed a 2x3x4x4 repeated measures, full factorial design. The independent variables were:

Factor A. Replicate in which subjects answered all questions twice.

Factor B. Three levels of bet: low= .80 to win $5 and .20 to lose $5 versus .20 to win $35 and .80 to lose $5; medium=.80 to win $60 and .20 to lose $16 versus .20 to win $246 and .80 to lose $16; and high= .80 to win $3,000 and .20 to lose $80 versus .20 to win $12,240 and .80 to lose $80.

Factors C. Four levels of when in time gains would be received--now, 6 months, 1 year, 2 years.

Factor D. Four levels of when in time losses would be received--now, 6 months, 1 year, 2 years.

The procedure was the same as in Study 1 except that questionnaires were delivered to 34 white collar professional workers in their offices at a university in the Mid-West. Subjects answered the two questionnaires approximately 1 week apart. As in the first study, subjects rated on an 8-point scale which of the two bets they would prefer to play. However, only 30 of the 34 subjects provided responses adequate for analysis.

Study 2 Results

In the first analysis, the replicate was ignored and an overall repeated measures, analysis of variance was run on the responses to the first set of questions. Supporting the results of the first study, main effects were found for the level of the bet (F=306.4, p<.0001) and for the time when the loss would occur (F=26.4, p<.0001). Means for the bet size variable were: small bet 5.7, medium bet, 4.7, and large bet 3.4). Thus, as the bet increased in size, subjects increasingly preferred the conservative option. Means for the loss variable were: now=4.2, 6 months=4.5, 1 year 4.7, and 2 years=4.9. A prior F tests indicated that a significant difference occurred between each of the conditions except for between the 6 months and 1 year condition. No significant effects were found for the gain variable. These results are consistent with those found in Study 1. Thus, evidence was found for the greater discounting of losses than gains.

In the second study, the replicate was included in order to perform individual analysis on each of the subjects. Because not all subjects completed both questionnaires, the analysis could be run on only 21 of the subjects. The results revealed that for 20 of the 21 subjects the effect for the bet size was significant at p<.10 or less. Only 4 of the subjects revealed a significant main effect for the time when gains would be obtained. In contrast, in nine instances a significant effect was found for time when losses would be paid off. Inspection of the means for each of the subjects showed a consistent pattern in which losses were discounted as they occurred further away in time. In contrast, for those revealing a main effect for gains, in some cases discounting was found. In other cases, gains were viewed as more valuable when they occurred in the future. The inconsistent pattern of responses for gains explains why no overall main effect was found for gains when the analysis was performed over the group rather than over individuals.

DISCUSSION

The TOV acts as a metatheory that can account for a variety of phenomena that involve the valuation of outcomes at divergent points in time. Previously, Mowen and Mowen (1991) developed the TOV and applied it to individual traps and fences, risk aversion in the present, and future optimism, among other phenomena. This paper extends the model to account for the effects of reactance. From a TOV perspective, reactance can be viewed as occurring when a person believes that a potential gain is about to be or has just been lost. Because people frame the failure to obtain a gain as a loss, a negative affective state is created that has been labeled reactance.

The paper also examined the relationship of the TOV to consumer risk perception. The model suggests that all else equal, when outcomes occur close in time to when the decision is made, risk aversion results. Conversely, the TOV predicts that, all else equal, when outcomes occur far away in time from when the decision is made, increased risk seeking behavior results. The results of the two experiments supported this prediction.

The results of Study 2 also revealed large individual differences in the subjects' choice of risky and conservative bets when the outcomes occurred at divergent points in time. When aggregated across individuals, the higher discounting of losses than gains is found. At an individual level, most subjects revealed the expected heavy discounting of losses. However, in some instances subjects discounted gains more. Some subjects even exhibited a tendency to inflate gains across time. Like most behavioral theories, the TOV is a model of "everyperson." Not all individuals will reveal the differential discounting of gains and losses predicted by the model. One avenue for future research is to investigate individual differences in the discounting of gains and losses. Potentially, a scale could be developed that would be effective in predicting the discounting tendencies of individuals. For example, those who are willing to make long term investments should exhibit a tendency to discount losses more quickly than gains. In contrast, those who fall for "get rich quick" schemes may be people who overweight gains in the present and then discount them very quickly when they occur in the future. Similarly, those who engage in compulsive consumption (Faber, O'Guinn, and Krych, 1987) may also reveal a pattern in which they overweight gains and underweight losses in the present.

An important question concerns identifying the mechanism that causes the differential weighting of gains and losses. Is it a memory phenomena? Is it "wired" into our affective response system? Does it relate to a differential ability to imagine the feelings obtained from positive or negative outcomes. These ideas suggest that future research should investigate the differential ability of people to recall positive or negative events from their past. It also suggests the possibility that mood states (Gardner 1985) may influence the impact of time on outcome valuation. Perhaps when people are in a good mood, they discount losses more rapidly. Conversely, when people are in a bad mood, perhaps they discount positive outcomes more rapidly.

If the author is right and individuals differentially discount gains and losses, why would this be the case? I would like to suggest the possibility that it would be advantageous from an evolutionary perspective to have such a valuation process wired into our biological system. The overweighting of negatives in the present and underweighting of negative future outcomes creates an impetus to change and to progress. Further, it would seem that creating a mechanism to cause future optimism would improve our outlook on life. While the model predicts that, all else equal, people are somewhat dissatisfied with their present state of affairs, it also suggests hope for the future. As a result, people may have a "wired-in" tendency to take actions to create a better future.

REFERENCES

Bauer, Raymond A. (1960), "Consumer Behavior As Risk Taking," in Dynamic Marketing for a Changing World, Robert S. Hancock, ed., Chicago: American Marketing Association, 87.

Benzion, Uri, Amnon Rapoport, and Joseph Yagil (1989), "Discount Rates Inferred from Decisions: An Experimental Study," Management Science, 35 (March), 270-284.

Brehm, Jack W. (1969), A Theory of Psychological Reactance, New York: Academic Press, Inc.

Clee, Mona and Robert Wicklund (1980), "Consumer Behavior and Psychological Reactance," Journal of Consumer Research, 6 (March), 389-405.

Dowling, G. R. (1986), "Perceived Risk: The Concept and Its Measurement," Psychology and Marketing, 3, Fall, 193-210.

Faber, Ronald J., Thomas O'Guinn, and Raymond Krych (1987), "Compulsive Consumption," Advances in Consumer Research, Melanie Wallendorf and Paul Anderson (eds), Provo, UT: Association for Consumer Research, 132-135.

Fishbein, Martin and Issaic Ajzen (1975), Belief, Attitude, Intention, and Behavior: An Introduction to Theory and Research, Reading, Mass.: Addison Wesley.

Gardner, Meryl (1985), "Mood States and Consumer Behavior: A Critical Review," Journal of Consumer Research, 12 (December), 113-135.

Jones, Edward and A. Johnson (1973) "Delay of Consequences and the Riskiness of Decisions," Journal of Personality, 10, 613-637.

Kahneman, Daniel and Amos Tversky (1979), "Prospect Theory: An Analysis of Decision Under Risk." Econometrica, 47 (March), 263-291.

Linder, Darwyn and Katherine Crane (1970), "Reactance Theory Analysis of Predecisional Cognitive Processes," Journal of Personality and Social Psychology, 15, 258-264.

Loewenstein, George F. (1988), "Frames of Mind in Intertemporal Choice." Management Science, 34 (February), 200-214.

Loewenstein, George F. and Richard Thaler (1989), "Anomalies: Intertemporal Choice," Journal of Economic Perspectives, 3 (Fall), 181-193.

Miller, N.E. (1959), "Liberalization of Basic S-R Concepts: Extensions to Conflict Behavior, Motivation, and Social Learning," in S. Koch (ed) Psychology: The Study of a Science, Vol. 2, 196-292.

Mowen, John C. (1990), Consumer Behavior, New York: Macmillan Publishing Co.

Mowen, John C. and Maryanne M. Mowen (1991), "The Time and Outcome Valuation Model: Implications for Managerial Decision Making," Journal of Marketing, October, In Press.

Platt, John, (1973), "Social Traps," Science, 28, 641-651.

Thaler, Richard, (1981), "Some Empirical Evidence on Dynamic Inconsistency," Economic Letters, 8, 201-207.

Walster, Elaine (1964), "The Temporal Sequence of Post-Decisional Processes," in Conflict, Choice, and Dissonance, L. Festinger (ed.), Stanford, Calif.: Stanford University Press, 112-137.

Wright, Peter and Barton Weitz (1977), "Time Horizon Effects on Product Evaluation Strategies," Journal of Marketing Research, 14 (November), 429-443.

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