Special Session Summary the Effects of Goals on Consumer Choice

Nathan Novemsky, Yale University
[ to cite ]:
Nathan Novemsky (2002) ,"Special Session Summary the Effects of Goals on Consumer Choice", in NA - Advances in Consumer Research Volume 29, eds. Susan M. Broniarczyk and Kent Nakamoto, Valdosta, GA : Association for Consumer Research, Pages: 5-7.

Advances in Consumer Research Volume 29, 2002     Pages 5-7



Nathan Novemsky, Yale University

Consumer choice has been one of the most researched areas in the consumer decision-making literature. Much of this research has examined issues relating to single choices, such as the rules that consumers employ or the manner in which the set of alternatives under consideration (i.e., the "choice context") and the task characteristics influence preferences (e.g., Bettman, Johnson, and Payne 1991). Although the study of goals has impacted some areas of consumer research, the study of consumer choice has been relatively unaffected. Much of the work in consumer choice has taken a "cold cognitive" view of the consumer, treating consumers as imperfect information processors. However, most choices are motivated by an attempt to satisfy or work toward satisfying a goal. This motivational view has been left out of many treatments of consumer choice because the research has concentrated on the decisions themselves and (sometimes) the immediate context of the decision. Consideration of the goals that drive consumer behavior allows us to make richer theoretical predictions as well as to link consumer choice research to the extant literature in psychology on goal representation and activation.

Goals can affect decisions in three primary ways. Each of our papers addresses one of these aspects:

1. Whether or not a consumer has set goals (and how high the goals are) affects the risks that a consumer is willing to take.

2. The linking of a series of decisions under one goal may affect sequential decision making, so that a particular decision depends on the goals activated (or fulfilled) by a previous decision,

3. In order to be able to drive behavior, one goal must be selected as primary. To this end, the activation of a goal suppresses competing goals. Changes in goal activation lead to preference changes and potential inconsistencies.

Despite different specific objectives, the three papers in this session have some commonalities. They all inform existing theories of consumer choice by providing novel predictions based on a consumer’s activated goals at the time of the choice. Two of the papers look at how goals affect risk-taking, a central topic in consumer choice. A different subset of two papers look at how the level of fulfillment of one goal affects choices related to a different goal. All three papers posit psychological processes that incorporate goals into the mechanisms of exsting theories of choice.

While these papers share common themes, they also offer somewhat different perspectives on the same issue. One paper involves the decision to set a goal, another considers the effects of the current level of fulfillment of a goal on decisions related to that goal, and the third paper focuses on the interaction of multiple goals considered simultaneously. Our discussion leader, Christie Brown, will help bring these themes together and suggest directions to help us further understand the role of goals in consumer choice.



Miguel Brendl, INSEAD

Arthur Markman, University of Texas at Austin

Julie R. Irwin, University of Texas at Austin

Most theories of goals posit that the strength or degree of activation of a goal determines how strongly that goal influences behavior. Goals may become active as a result of chronic personality factors, local contextual factors and physiological needs. One question that has not yet been addressed is the influence of the activation of one goal on competing goals. There are two plausible possibilities for this influence: (1) goals wax and wane in their activation independently, and (2) strong activation of one goal inhibits the activation of competing goals. Our research is the first, to our knowledge, to directly address these two possibilities.

The activation and suppression of goals has a number of implications for decision making. In this presentation we will outline these implications and present new research on the topic. Two of the authors’ previous research has shown that the presence of a strongly activated goal will increase the attractiveness of goal-related items. For example, a habitual smoker who is smoke deprived finds a raffle to win cigarettes in the future more attractive than a raffle to win cash in the future. A habitual smoker who has just had a cigarette shows the opposite preference ordering. So, goal activation affects preference, even when the preference will not satisfy the immediate goal (we can assume all of the smoke deprived smokers had a cigarette before they were due to receive the lottery prize).

More importantly for the current research, this finding suggests that one strongly active goal (i.e., to smoke a cigarette) inhibited competing goals (e.g., to acquire cash). To explore this issue more deeply, we first ran a field study in which college students were approached either right before or right after they had eaten lunch. They were given a brief vignette describing a lottery their university was considering that would give them an opportunity to win $1,000 in cash. They were asked to indicate how much they would be willing to pay for this lottery using a scale ranging from $0 to $3 in 25 cent increments. On average, students who had not yet eaten lunch were willing to pay significantly less for this lottery than were students who had already eaten lunch. This result further supports the hypothesis that an active goal inhibits competing goals.

Laboratory versions of this research have also been run. In this work, a variety of manipulations are used to induce a goal (such as eating) and items are evaluated either before or after that goal is satisfied. The results of these laboratory studies are also generally consistent with the view that strongly active goals inhibit competing goals, thereby making objects that are related to those competing goals less attractive. In fact, we have explored the possibility that contextual effects on decision making may be manipulated changes in goal activation.



Richard P. Larrick, University of Chicago

Chip Heath, Stanford University

George Wu, University of Chicago

One of the most widely-documented findings in psychology is that goals improve task performance (Locke & Latham, 1990). In general, having a specific, challenging goal increase effort and persistence compared to a vague intention, such as "doing one’s best." Recently, we have proposed that research on individual decision making provides an important link between goals and motivation (Heath, Larrick, & Wu, 1999). Specifically, we proposed that a goal serves as a reference point, dividing performance outcomes into regions of gain and loss. The Prospect Theory value function (Kahneman & Tversky, 1979; Tversky & Kahneman, 1992) predicts that risk preferences will be different in these regions: The principle of "diminishing sensitivity" implies that decision makers will be risk averse for choices involving gains (where the value function is concave) but risk seeking for choices involving losses (where the value function is convex). If goals serve as a reference points, they will systematically change risk preference.

Three Experiments

The following three studies (Larrick, Heath, & Wu, 2001) used the standard goal-setting manipulation of instructing some participants to set a "specific, challenging goal" and others to "do their best" (Locke & Latham, 1990) to test the effect of goals on risk taking. All studies were played for actual money.

Study 1 gave participants the opportunity to earn money by solving anagrams that varied in difficulty and monetary value. Hard anagrams were riskier than easy anagrams because they were worth more (40 versus 20 cents each) but were also much harder to solve. Participants were given a cover sheet with examples of hard anagrams and easy anagrams, and then were told to set a "specific, challenging" monetary goal or "to do their best" at making money. They were then instructed to select a mix of hard and easy anagrams that would constitute a pool of 15 items to solve. As expected, participants who set a specific, challenging goal opted for a riskier pool of anagrams (5.67 hard anagrams and 9.33 easy anagrams) than did participants in the do your best condition (3.51 and 11.49, respectively), p<.01, and as a result, they earned less money ($1.50 and $1.81, respectively), p<.05.

In a second study, MBA students were asked to choose among risky gambles either under "do your best" or "specific, challenging goal" instructions. They were presented with 15 gambles that increased in payoff by 50 cents, ranging from a 100% chance of winning $3 to a 21% chance of winning $10. The expected value was shown next to each gamble. Expected value decreased by approximately 5 cents with each 50 cent increase in the absolute payoff. Participants selected riskier gambles in the goal condition than in the do your best condition, sacrificing the expected value of their payoffs (Median EVs=2.52 and 2.75, respectively), p<.01.

In a third study, participants played a simple two-party distributive bargaining exercise known as an ultimatum game in which a "Proposer" offers a division of $7.00 and a "Responder" simultaneously sets a minimum acceptable payment. The modal response for both players is an equal division (Camerer & Thaler, 1995). Thus, a risky strategy in this game is to demand more than an equal division, especially for Responders. As predicted, participants who set a specific, challenging goal were more likely to engage in a risky strategy. Specifically, nearly 30% of Responders in the goal condition requested more than half of the $7.00, compared to 5% in the do your best condition. As a result, the market was much less efficient in the goal condition, for which only 53% of all possible transactions would be completed, than in the do your best condition, for which 68% would be completed, t(74)=-2.13, p<.05, twotailed. This 15% difference translated into an expected payoff that was $1.05 less for goal dyads than for do your best dyads.

Goal-Induced Risk Aversion

It is important to note that the Prospect Theory value function does not predict that goals will always increase risk taking. Although the convexity of the loss function leads people to be risk-seeking when they are deep in the domain of losses, loss aversion will make them extremely conservative if their current strategy is sufficient to lift them from their current "loss" position. For example, consider the scenario below, which was presented to 147 MBA students:

Terry works in sales and has set a goal of making 30 sales [doing his best] this month. With two days left, he has completed 25 sales. He’s considering two strategies:

A) He knows that if he concentrates his remaining two days on the 5 clients most ready to buy, he can close those deals.

B) He can spend his time spot-calling his entire remaining base of 20 clients. On average, this strategy in the past has produced between 2 to 8 sales with equal probability.

Which strategy will Terry prefer?

1                  2                  3                  4                  5                  6                  7

Strongly prefer A                          Indifferent                                   Strongly prefer B

In this case, the certainty of reaching the goal makes the safe option much more attractive than the gamble (which risks leaving the salesperson in the domain of losses), and leads to greater risk aversion than in the do your best condition, t(146)=-2.09, p<.05 .


We used the Prospect Theory value function to predict that people would be significantly more risk-seeking when they set a specific, challenging goal. Consistent with this prediction, we found that goal participants were more likely to choose risky strategies than participants who were trying to do their best. By pursuing riskier strategies in these tasks, goal participants accepted lower expected values for their final outcomes. In using goals to motivate employees, organizations must manage the process so that they realize the advantages of innovation and creativity (Longswirth, 1991) while avoiding the potential damage of reckless actions (Maremont, 1995).



Ravi Dhar, Yale University

Nathan Novemsky, Yale University

Although most research on consumer decision making has focused on isolated choices, many products are purchased and consumed in conjunction with other products in order to satisfy a single goal. For example, consider a person who is thinking about spending an evening out which involves a number of decisions and experiences related to entertainment, transportation, and food. This person’s choices during the evening will be affected by their goals for that evening (e.g. having a good time, eating healthy). However, surprisingly little is known about how the achievement of goals as a sequence unfolds alters later choices in the sequence. Some recent work has looked at which sequences of outcomes are evaluated as more favorable either before the sequence has begun (prospective evaluation) or after the sequence has ended (retrospective evaluation). In contrast, our objective is to examine choices made in an ongoing sequence. We examine the effect of one consumption experience on risk preference for a second consumption experience. Consider a consumer who has just finished her entrTe at a restaurant. Further imagine that on one occasion the entrTe was enjoyable and on another occasion was not enjoyable. This consumer is now considering what kind of dessert to orderBa dessert that she has tried before and is moderately enjoyable or a different one that is equally likely to be much worse or much better than the first dessert. We predict and find that if the first outcome helped the individual meet her goal, she is more likely to seek riskier options that can provide peak goal experiences than if the first outcome does not move her toward her goal.

Our explanation centers on goal achievement and the motivation to fulfill those goals. For example, if consumption of the first item moves one toward a goal (enjoyed the entrTe), the positive goal experience leads that individual to shift her aspiration level upward and strive for a higher goal. Reaching for this higher aspiration level increases the risk seeking tendencies of such individuals in the second choice (risky dessert). Those individuals who did not enjoy the first experience risk not reaching their initial aspiration level (an enjoyable meal) and therefore are not motivated to reach a higher goal. As a result, they are less risk seeking.

We provide further support for the goal achievement account by showing that the effect of a positive initial outcome on a subsequent choice between two items depends on the implicit goal achievable in the second choice. When the second choice involves risk along a dimension unrelated to the achievement realized in the first consumption choice, the success of the first consumption experience has no effect on the second choice. The motivation to fulfill a goal impacts risk-taking in the second choice only when the risk in that second choice is related to fulfilling the same implicit goal. This risk-taking is not a function of the risk, but rather it depends on which option (risky or riskless) offers the highest level of goal achievement. We observe risk aversion after a positive initial outcome, when the riskless option offers the highest possible level of goal achievement.

Increased risk-taking can be seen in choices and consumption that is not adjacent in time as long as both events relate to the same goal. In fact, we find that people will link two temporally distant events using a spontaneously generated goal and exhibit the risk preferences described above. This pattern holds across several domains, including choices about hedonic consumption, health behavior, and helping behavior.

We also find that the framing of the goal has important consequences for this sequential effect. An explanation based on the effects of positive affect on risky choices is found to be incomplete. Implications for consumer choice are discussed.