Consumer Decision Making and Perceived Decision Freedom

ABSTRACT - This experimental study explores the decision consequences of varying levels of perceived decision freedom. The results indicate that the number of equally valued alternatives is the most important antecedent of perceived decision freedom. Further, these results suggest different cognitive, affective, and behavioral consequences for different levels of perceived decision freedom. Differences between this view of perceived decision freedom and reactance theory are also discussed.


John R. Walton and Eric N. Berkowitz (1985) ,"Consumer Decision Making and Perceived Decision Freedom", in NA - Advances in Consumer Research Volume 12, eds. Elizabeth C. Hirschman and Moris B. Holbrook, Provo, UT : Association for Consumer Research, Pages: 461-464.

Advances in Consumer Research Volume 12, 1985      Pages 461-464


John R. Walton, Miami University

Eric N. Berkowitz, University of Massachusetts


This experimental study explores the decision consequences of varying levels of perceived decision freedom. The results indicate that the number of equally valued alternatives is the most important antecedent of perceived decision freedom. Further, these results suggest different cognitive, affective, and behavioral consequences for different levels of perceived decision freedom. Differences between this view of perceived decision freedom and reactance theory are also discussed.


Clee and Wicklund (1980) attempted to extend the concept of psychological reactance to consumer decision king. Reactance was described as the motivational state of the person whose freedom is threatened" (Clee and Wicklund 1980, p. 389). The authors identified (1) expectation of freedom and (2) threats which infringe on that freedom (i.e., social influence and/or external barriers) as antecedents of reactance and suggest behavioral and attitudinal consequences aimed at reasserting this threatened freedom.

This model implies that for consumer behavior consequences to be manifested, perceived decision freedom must be threatened. While these authors cite several research studies that support this view, other research has demonstrated that perceived decision freedom has consumer behavioral consequences independent of the perception of threatened loss of freedom (Reibstein, Youngblood, and Fromkin 1975; Walton et al. 1979). It may be equally fruitful, therefore, for consumer researchers to study the effects of varying levels of decision freedom, as opposed to the more restrictive situation in which freedom is only perceived to be threatened.

Accordingly, the objectives of this paper are twofold. First, perceived decision freedom is conceptualized by (1) defining the term within the context of the decision process and (2) reviewing research on important antecedents of perceived decision freedom. Second, particular antecedent conditions are manipulated experimentally for a consumer decision involving non-carbonated soft drink alternatives. Hypothesized consequences of these antecedents manipulations are then specified and tested.


The individual consumer decision process begins with the identification of a consumption problem. The consumer will then engage in internal, and, if necessary, external search activities (Engel, Kollat, and Blackwell 1973, p. 375). If successful, these activities will result in the identification of an alternative or see of alternatives which may conceivably solve the problem. Decision freedom relates to the perception of volition the consumer believes herself/himself to exercise when he/she decides (1) whether or not to choose among the alternatives identified and/or (2) to choose one alternative over the others in the set (Harvey 1976; Steiner 1970). In recent literature, the term perceived control has been considered synonymous with perceived decision freedom (Harvey, Harris, and Lightner 1979).

The amount of empirical research on perceived decision freedom has been limited despite three extensive theoretical discussions in the last decade (Harvey 1976; Harvey, Harris, and Lightner 1979; Steiner 1970). This contrasts sharply with the number of research efforts that have tested theoretical propositions that assume the presence of some level of perceived decision freedom. These efforts include not only reactance studies, but also numerous attribution and dissonance studies. Given the potential importance of perceived decision freedom in both marketing and nonmarketing decision situations, this lack of effort is surprising.

Antecedent Conditions

Extant research on perceived freedom has been concerned primarily with the identification of important antecedent conditions. Most of these efforts have focused on the relative attractiveness of and number of alternatives in the choice set. These findings have consistently supported the hypothesis that greater decision freedom is manifested when alternatives are perceived as being relatively equal in attractiveness (Brehm and Cohen 1959; Harvey and Jellison 1974; Harvey and Johnston 1973; Jellison and Harvey 1973; Monty et al. 1979; Walton et al. 1979).

Given relative equality of preference, some studies have assessed the relationship between perceived decision freedom and the number of alternatives in the choice set. Both monotonic and nonmonotonic results have been observed. Walton et al. (1979) reported a positive monotonic relationship when 2 and 4 equally valued choice alternatives were considered. In a second study, Harvey and Jellison (1974) found nonmonotonic relationships for 3, 6, and 12 alternatives, with the highest decision freedom associated with the 6 alternative condition. While these are seemingly inconsistent, they do not necessarily conflict when the range in the number of alternatives is considered.

Knowledge relating perceived decision freedom to another antecedent variables is sparse. Some preliminary efforts have investigated predecision uncertainty, valence of available alternatives, and quantity of available alternatives.

Predecision uncertainty refers to the probability that the expected consequences of choosing a particular alternative will, in fact, be experienced by the decision maker. Two attribution studies that have investigated the relationships between predecision uncertainty and perceived decision freedom report contradictory findings. Subjects in Kruglanski and Cohen's (1974) study attributed more decision freedom when outcomes were more certain, while their counterparts in Harvey and Johnston's (1973) study demonstrated the opposite relationship. While this conflict cannot be resolved in this review, it is noteworthy that both studies measured subject attributions toward identified actors, and not actual perceptions of, and responses to, personal choice situations. Since the perception of decision freedom in personal decision situations may differ from attribution situations, efforts to investigate predecision uncertainty in the former appear justified.

Assuming alternatives in the choice set are perceived as relatively equal in attractiveness, the valence of these alternatives may still vary from very positive to very negative. Harvey and Harris (1975) reported a positive relationship between the perception of decision freedom and the valence of alternatives in the choice set. This finding suggests that any experiment attempting to investigate decision freedom by constructing relatistic choice sets should select sets that are positive, or at least neutral, in valence. This selection criterion should reduce an important source of error variance in the measurement of perceived decision freedom and related constructs.

A final potential antecedent of perceived decision freedom is the quantity of alternatives available in the choice set. Walton et al. (1979) observed that in certain choice situations, increasing the number of alternatives in the choice set also increases the quantity of alternatives available to subjects. Specifically, these authors investigated the amount of consumption of a chosen soft drink when subjects chose from 2 to 4 alternative choice sets. They argued that the greater quantity of drink available to the 4 alternative subjects may affect their subsequent soft drink consumption. Their findings, however, were equivocal on this point.

Two issues are apparent from this review of perceived decision freedom research. First, few consumer research studies have investigated perceived decision freedom. Second, studies that have been performed on this concept have focused on the relationship between various antecedents and perceived decision freedom and have ignored the various consequences of experiencing decision freedom. The study described in the remainder of this paper is concerned with both of these issues.


Subjects were 206 male and female introductory marketing students. The decision task facing these subjects was to choose a particular flavor of a noncarbonated soft drink from a set of equally valued alternatives.

Independent Variables

The antecedent variables served as the independent variables in the study. Three levels of number of soft drink flavor alternatives (2, 4, and 6) and two levels of quantity (I or 2 two-quart pitchers of each soft drink flavor) were included. Two levels of predecision uncertainty (high or low) were specified and operationalized as described in the experimental procedure discussion. Combining these independent variables into a factorial design resulted in 12 distinct treatment conditions.

Pretest and Subject Assignment

Subjects were pretested in their scheduled class period to determine their preferences for 12 different flavors of noncarbonated soft drinks. These flavors were: black cherry, cherry, grape, apple cider, lemonade, raspberry, fruit punch, orange, lemon-lime, peach, strawberry, and pink lemonade. Preferences were measured on a five point scale ((1) Dislike a lot, (2) Dislike Somewhat. (3) Neutral. (4) Like Somewhat. and (5) Like a lot). After completion of the pretest, subjects were asked to sign-up for a 1/2 hour individual session. These individual sessions started one week after the pretest and continued for three weeks.

For inclusion in the second phase of the study, a subject's pretest data must have shown at least two flavors rated equally and the preference rating for these similar alternatives in the neutral or like categories. This latter requirement was necessary because of the valence research cited earlier. Six subjects failed to meet these criteria and were eliminated from the study.

A second subject assignment consideration was that a subject must have rated at least 4 flavors equally (either neutral or like categories) to be assigned to a 4 alternative treatment or 6 flavors equally (either neutral or like categories) to be assigned to a 6 alternative treatment. Subject to this restriction, subjects were randomly assigned to the twelve treatments. [The authors recognize the potential for bias in this assignment method. It may be that subjects rating more alternatives equally were more equivocal or less discriminatory in their judgment. Unfortunately, given the objectives of the study, this potential bias had to be accepted.]

Experimental Procedure

Before a subject's arrival, the appropriate number of flavors and pitchers of soft drink were prepared and placed on a table. A chair on which the subject would sit was placed directly in front of the table. The pitchers were covered with a large piece of white paper so that as they were seated, the subject could not see the specific alternatives available to him/her. A partition was placed around the area so that the table could not be seen by the subject until the experimenter was ready to begin.

Upon arrival, the subject was seated in the briefing area and his/her thirst was measured on a 4 point scale ((1) extremely thirsty, (2) very thirsty, (3) somewhat thirsty, (4) not thirsty at all). After current level of thirst was assessed, the following instructions were read to each subject.

"Let me briefly describe what we want you to do. Behind the divider are several different flavors of powdered, noncarbonated soft drinks like Kool-aid. They have already have been mixed and are in pitchers ready to drink. The names of the flavors appear on a card in front of the appropriate pitcher.

In a minute, you will be seated in front of these pitchers. At the time I want you to choose the one flavor that you would most like to drink. You may take as long as you wish to make your choice. However, you may not touch any of the pitchers, nor taste the contents of any of the pitchers.

As soon as you have chosen a flavor, please say, out loud, the name of that flavor. I'll move that pitcher (or pitchers) to you and give you a cup. At that time you may drink as much of your chosen flavor as you wish. We do ask that you don't leave any liquid in your cup. So, if you pour it, please drink it.

While you're drinking I'll be removing the pitchers you didn't choose. I'll be back in a few minutes. Even if you're finished drinking, don't leave until I come back. When I come back, I'll give you a short questionnaire to fill out and then you'll be finished."

At this point, depending upon the treatment condition to which the subject had been assigned, two versions of the remaining instructions were read. The intent of these two versions was to manipulate predecision uncertainty. Version A read as follows: "One more thing. All of the flavors you will choose from were mixed according to package directions. Therefore, none should be too sweet or too tart. Any questions?" Since it was determined on the pretest that all subjects were familiar with noncarbonated soft drink products, it was expected that the subjects would perceive a low level of predecision uncertainty as to the taste outcomes of these alternatives.

Version B read as follows: "One more thing. e person who mixes the soft drinks has not had a lot of experience. He just told me that he doesn't think he used enough sugar. Therefore, all the flavors you have to choose from might be too tart. Any questions?" It was expected that subjects who received this version would exhibit higher levels of predecision uncertainty as to the taste outcomes of their alternatives. It should be noted that all drinks were mixed according to package directions.

The subject was then taken behind the partition and seated in the chair. The experimenter stood to the left of the subject and removed the paper with her left hand and started a concealed stopwatch with her right hand. The experimenter stopped the watch as soon as the subject indicated his/her choice.

After the subject consumed as much of the room temperature drink as he/she wanted, he/she was asked to complete a questionnaire. After the subject completed the questionnaire, he/she was thanked for their participation. After the subject left, the experimenter recorded the elapsed time on the stopwatch and inspected the cup to insure no liquid remained. In each case the cup was virtually empty. Each pitcher contained 32 ounces of liquid and was graduated in ounces so that the amount that had been poured out by the subject could be easily determined. The experimenter recorded this amount and the room was Prepared for the next subject.

Dependent Variables

In addition to the amount of soft drink consumed and response time t the questionnaire measured a variety of affective and cognitive response variables. Table 1 lists these additional dependent variables and the scales used to measure them. Also, the questionnaire contained the same flavor preference scales as the pretest. From these data, a preference change score for the chosen flavor alternative was calculated. The scale ranged from 1 (negative change of 4 scale units) to 7 (positive change of 2 units).



Hypotheses and Analysis Methods

Four hypotheses were formulated and subsequently tested. In order to assess the consequences of perceived decision freedom, it was necessary that perceived decision freedom vary across treatment conditions. Therefore, the first hypothesis was:

H1: Perceived decision freedom will vary across the antecedent treatment conditions.

Subsequent hypotheses investigated cognitive, affective, and behavioral consequences of perceived decision freedom. These hypotheses were:

H2: Measures of cognitive consequences will vary across antecedent treatment conditions.

H3: Measures of affective consequences will vary across antecedent treatment conditions.

H4: Measures of behavior and behavioral intention will vary across antecedent treatment conditions.

Nonorthogonal analysis of variance (regression method) was used to analyze each of these hypotheses. When significant effects were observed, multiple comparisons were accomplished using the Newman-Kuels procedure.


Any thirst differences across treatments might seriously confound the results of the study. Therefore, as a manipulation check, the four point thirst measure was analyzed. No significant interactions or main effects were indicated.



Table 2 summarizes the analysis of variance results for each hypothesis. For H1, alternatives and predecision uncertainty main effects were significant, while the quantity main effect and all interaction terms were not. Perceived decision freedom means for the 2, 4, and 6 alternative conditions were 4.16, 5.45, and 6.05, respectively. Multiple comparison tests documented significant differences .between each pair. For predecision uncertainty, perceived decision freedom was greater when outcomes were more certain (X = 5.32) as compared to less certain (X = 4.66).

Hypothesis 2 concerned two cognitive variables, thirst-quenching and difficulty in participating. Only a significant alternatives main effect was observed for thirst-quenching. Specifically, the 2 alternative condition was seen as less thirst-quenching (X = 3.03) than either the 4 (X = 3.60) or 6 (X = 3.54) alternative conditions.

Regarding Hypothesis 3, a significant alternatives main effect was indicated for flavor preference change, how good, and friends like. As shown in Table 3, the pattern of means for each of these variables was identical. For each variable, the two alternatives value was significantly lower in affect than either the four or six alternative values. The four and six alternative comparisons, however, were not significantly different.

While all other effects for these three variables did not reach conventional significance levels, the predecision uncertainty main effect for flavor preference change approached the minimum standard (p = .06). Given this proximity, this finding is reported. Specifically, greater preference change was evidenced when outcomes were more certain (X = 4.26) as compared to less certain (X = 4.08).



The final hypothesis concerned behavior or behavioral intention differences across treatments. As seen in Tale 2, soft drink consumption and response latency exhibited such differences.

The alternative - predecision uncertainty interaction was the single significant effect for soft drink consumption. This effect is graphically displayed in Figure 1. For two alternatives, a significantly greater number of soft drink was consumed when outcomes were more certain. The effect became less pronounced as the number of alternatives increased. Indeed, for six alternatives, more consumption was found in the less certain group.



Significant alternative and quantity main effects were observed for the response latency measure. As the number of alternatives increased, response latency increased accordingly. Mean seconds to decision for the 2, 4, and 6 alternative treatments were 7.73, 11.63, and 13.56, respectively. Differences between each pair were significant.

A similar relationship was exhibited with quantity levels. That is, response latency was greatest for two pitchers (X = 10.96) as compared to one (X = 9.64). Willingness to participate was the intention measure and did not vary across treatments.

All of the research hypotheses in the study received some degree of support. The importance of these results are discussed in the concluding section of this paper.


The principal objective of this study was to assess the effects of varying levels of perceived decision freedom on selected cognitive, affective, and behavioral consequences. Overall, the results suggested a range of such effects, with affective and behavioral being the most pronounced. Specific findings indicated that the number of similarly valued alternatives is the antecedent condition most strongly related to the perception of decision freedom. Relationships between the other antecedents and perceived decision freedom were much less pronounced.

While the decision task in this study was admittedly atypical, some generalizations to other types of decision situations seem appropriate. For example, the bulk of all consumer decisions may be classified as routine response behavior (Haward 1977). Since the consumer has typically made this decision frequently, it is reasonable to assume that one or more preferred alternatives have been identified. When a single alternative is clearly preferred to the others for this type of decision, we would expect decision freedom to be perceived as low. Based on the findings in this study, cognitive and behavioral differences and lessened affect could be predicted. Note that reactance theory would predict consequences only if the consumer perceived their freedom to be threatened. This is an unlikely occurrence for this type of decision.

When two or more alternatives are equally preferred for a routine response decision, perceived decision freedom should- be greater. In this situation, cognitive and behavioral differences and heightened affect could be predicted. Similar outcomes should be expected for most nonroutine decision situations. In these limited or extensive problem solving situations (Howard 1977), we would expect greater external search to result in the identification of a set of potentially attractive alternatives. Once again, recognizing the presence of heightened levels of perceived decision freedom would allow predictions about likely consequences that are not obvious with reactance theory.

This study represents only an initial effort and many issues remain unresolved. For example, the use of a homogeneous subject pool ignores any relationship between individual difference variables and perceived decision freedom. Demographic, socio-economic, personality, and cognitive style variables undoubtedly have some impact on the perception of decision freedom and these relationships should be assessed.

Furthermore, the relationship between perceived decision freedom and the quality of the resulting decision was ignored. It may be that when perceived decision freedom is greater, consumers feel better about their decision but do not actually choose the best alternative available to them. Clearly, more research efforts are required to clarify these and other research issues. Based upon the results of this study, these additional efforts seem warranted.


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

Engel, James F., Kollat, David T., and Blackwell, Roger D. (1973), Consumer Behavior, New York: Holt, Rinehart, and Winston, Inc.

Harvey, John H., (1976), "Attribution of Freedom," in New Directions in Attribution Research, eds. John H. Harvey, William J. Ickes, and Robert F. Kidd, Hillsdale. N.J.: Lawrence Erlbaum Associates.

Harvey, John H., and Harris, Ben (1975), "Determinants of Perceived Choice and the Relationship Between Perceived Choice and Expectancy about Feelings of Internal Control," Journal of Personality and Social Psychology, 31, 101-106.

Harvey, John H. and Harris, Ben, and Lightner, Jean M. (1979), "Perceived Freedom as a Central Concept in Psychological Theory and Research," in Choice and Perceived Control, eds. Lawrence C. Perlmuter and Richard A. Monty, Hillsdale, N.J.: Lawrence Erlbaum Associates.

Harvey, John H., and Jellison, Jerald M. (1974), "Determinants of Perceived Choice, Number of Options, and Perceived Time in Making a Selection," Memory and Cognition, 2, 539-544.

Harvey, John H., and Johnston, Shawn (1973), "Determinants of the Perception of Choice," Journal of Experimental Social Psychology, 9, 164-179.

Howard, John A. (1977), Consumer Behavior: Application of Theory, New York: McGraw-Hill.

Jellison, Jerald M., and Harvey, John H. (1973), Determinants of Perceived Choice and the Relationships Between Perceived Choice and Perceived Competence'," Journal of Personality and Social Psychology, 28, 376-382.

Kruglanski, Arie W., and Cohen, Menashe (1974), 'Attributing Freedom in the Decision Context: Effects of the Choice Alternatives, Degree of Commitment, and Predecision in Uncertainty,' Journal of Personality and Social Psychology, 30, 178-187.

Monty, Richard A., Geller, E. Scott, Savage, Ricky E., and Perlmuter, Lawrence C. (1979), "The Freedom to Choose is Not Always So Choice," Human Learning and Memory, 5, 170-178.

Reibstein, David J., Youngblood, Stuart A., and Fromkin, Howard L. (1975), "Number of Choices and Perceived Decision Freedom as a Determinant of Satisfaction and Consumer Behavior," Journal of Applied Psychology, 60, 434-437.

Steiner, Ivan D. (1970), "Decision Freedom," in Advances in Experimental Social Psychology; ed. Leonard Berkowitz, New York: Academic Press.

Walton, John a., Berkowitz, Eric N., Ross, Ivan, and Cvar, Margaret (1979), 'Consumer Behavior and Perceived Decision Freedom: A Reexamination," Journal of Applied Psychology, 64, 472-476.



John R. Walton, Miami University
Eric N. Berkowitz, University of Massachusetts


NA - Advances in Consumer Research Volume 12 | 1985

Share Proceeding

Featured papers

See More


P11. A Price Premium on A Trivial but Weak Preferred Attribute Increase Choice: The Roles of Scarcity, Arousal and Perceived Risk

Yueyan Wu, Hunan University, China
Chunyan Xie, Western Norway University of Applied Sciences
Zhi Yang, Hunan University, China
Luluo Peng, Hunan University, China

Read More


Product Transparency in Online Selling Mechanisms: Consumer Preference for Opaque Products

Lucas Stich, Ludwig-Maximilians-University Munich
Martin Spann, Ludwig-Maximilians-University Munich
Gerald Häubl, University of Alberta, Canada

Read More


Crime and Punishment through the Political Lens: How Liberals Forgive, and Conservatives Punish Ethical Brand Users

Thomas Allard, Nanyang Technological University, Singapore
Brent McFerran, Simon Fraser University, Canada

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.