New Directions in Family Decision Making Research

ABSTRACT - In this paper, first, a brief review of the literature on family decision making is presented. Having identified the key variables, a simplified depiction of the-process is proposed. The next section proposes some research questions under current investigation by the authors. Finally, an experimental approach suitable for investigating these research questions is discussed.

Citation:

Sunil Gupta, Michael R. Hagerty, and John G. Myers (1983) ,"New Directions in Family Decision Making Research", in NA - Advances in Consumer Research Volume 10, eds. Richard P. Bagozzi and Alice M. Tybout, Ann Abor, MI : Association for Consumer Research, Pages: 445-450.

Advances in Consumer Research Volume 10, 1983      Pages 445-450

NEW DIRECTIONS IN FAMILY DECISION MAKING RESEARCH

Sunil Gupta, University of California, Berkeley

Michael R. Hagerty, University of California, Berkeley

John G. Myers, University of California, Berkeley

ABSTRACT -

In this paper, first, a brief review of the literature on family decision making is presented. Having identified the key variables, a simplified depiction of the-process is proposed. The next section proposes some research questions under current investigation by the authors. Finally, an experimental approach suitable for investigating these research questions is discussed.

INTRODUCTION

Most Consumer Behavior researchers have treated the individual as the unit of analysis. However, the position can be taken, as argued by Samuelson (1956, pp. 8-9) that, "the fundamental unit on the demand side is clearly the family and this consists of a single individual in but a fraction of the total cases." Early research (see Davis', 1976, review) has shown that a substantial proportion of the purchases are made jointly by the husband and the wife. The fact that more than one person is involved in the decision making process renders this phenomenon significantly different from the study of individual decision making. Not only must the analyst study individual preferences, but there is also the need to model how these preferences may be combined to give us a joint decision.

Past research in the area of family decision making (Davis, 1976; Burns and Granbois, 1980) has concentrated on measuring the perceived influence shared by husbands and wives through their recollections of various tactical decisions encountered, mostly, in the purchase of large durable goods. However, as Burns and Granbois (1980) have observed, "the knowledge is restricted by wide reliance on cross sectional data ..." (Similar comments have been made by Davis). This past research has quite clearly established that a great deal of variability in family member roles is to be found both across product categories and within decisions. But, as Davis (1976) concluded, "With few exceptions researchers have not explored why some product categories or subdivisions within product categories are dominated by husbands and others by wives." Additionally, little theoretical or empirical attention has been paid to the significance of the on-going role relationship of a husband and wife and the possibility that future decisions could impact very heavily on and might serve as powerful explanatory variables for current decisions.

In this paper we try to address some of these issues. First, a brief review of the literature dealing with these key issues is presented. Having identified the key variables, a simplified depiction of the process is presented. The next section proposes some research questions under current investigation by the authors. Finally, an experimental approach suitable for investigating these research questions will be discussed.

LITERATURE REVIEW

As has been noted by earlier reviews of this literature (Davis, 1976; Burns and Granbois, 1980), most researchers have investigated the power allocation and role differentiation in various decision categories. Typically these studies have relied upon the respondents' overall perceptions or recollections of the degree to which they participate autonomously or jointly (power allocation) in various decisions or decision phases (product-related factors) and have explored the impact of determinants such as life-cycle stage, cultural environment (organizational/group factors); personality correlates, and demographic characteristics (individual factors) on the incidence of these role structure types. Some recently proposed models, however, have emphasized the potential significance of research going beyond the determination of power structures and exploring the discrepancies between family members' preferences and the means of resolving conflict when discrepancies occur. Sheth (1974) and Granbois (1971), both, predict conflict resolution as a direct outcome of preference discrepancy without intervention of modifying variables. Subsequently Burns and Granbois (1977) have demonstrated the importance of including such modifying variables as interest (they call it involvement) and power (their recognized authority is one component of this). However, they do not propose a model to predict the outcome of a joint decision. Krishnamurthy (1981) has proposed two models which, using interest of the decision participants as the modifying variable, does give specific predictions about preferences of the collectivity for each decision alternative. One of these models (Conflict Resolution Model) does not consider power and the other (Relative Influence Conflict Resolution Model) confounds it with interest. (According to the Conflict Resolution Model each individual's preferences are weighted in proportion to the range of part-worths over the levels of a given attribute. In the Relative Influence Conflict Resolution Model the part-worths are rescaled, before estimating the conflict resolution parameter, such that the standard deviations of the part-worths are proportional to the relative influence of each participant in the joint decision. Since a person's interest in a decision is a function of the variance of his/her utilities, interest and power are confounded).

An area of research concerned with dyadic decision making, yet almost completely ignored by the family decision-making literature is game-theory. Two models that have received some empirical support (Heckathorne, 1978) and a great deal of theoretical investigation are Nash's (1950) solution to the bargaining problem and Kalai and Smorodinsky's (1975) variation on it. [Harsanyi and Selten (1972) have proposed a theoretically appealing solution to the problem. However, the mathematical intractability or the solution concept renders it too difficult to use.] The game-theoretic models are based on preferences discrepancies and the interest of each participant in the decision. This is in direct contrast to the almost exclusive reliance of the extant family decision-making literature on power as the predictor of joint decisions.

As noted earlier, past research has rarely explicitly modeled the on-going relationships of husbands and wives, though the importance of considering this aspect of the process has been emphasized (Davis, 1976). Most past research dealing with this aspect of the process has been theoretical (Pollay, 1969; Coleman, 1973). The basic phenomenon has been labeled the principle of reciprocity by Turner (1970). It is argued that because families make many decisions over time, one member of the family may step out of some decision in which he or she is not particularly interested and seemingly capitulate to another family member. However, the expectation is that the person who prevailed this time will allow the other person to prevail in some future decision. A rigorous theoretical model that explicitly models multiple decisions, interests and authority had been proposed by Coleman (1973). However, this model has not been adequately tested empirically.

It is interesting to note here that Krishnamurthy's (1981) Relative Influence Conflict Resolution model can be considered to be dealing with ongoing decision making. By treating a single decision as a series of compromises to be made on each of the attributes, the dynamic aspects of the decision process are formally incorporated. The reliance of the model on the order in which the attribute levels are discussed and the shortcomings noted earlier are some of the problems with this model.

In summary, we note that there is a need to explicitly study:

i) preference discrepancies among members or the dyad, and

ii) multiple decisions and observe how interest and power interact to give the influence Pattern in a given decision.

THE PROPOSED FRAMEWORK

A framework relating the key variables of the process is shown in Figure 1.

In terms or Figure 1, most of the past research can be seen as dealing with the relation between Individual, Product-Related, and Organizational/Group Factors. Further, most or these studies have looked at single decisions only.

Our points of departure are:

A) explicit differentiation is made between power, influence and authority. In accord with Cartwright (1965) we treat power as the potential to effect some future decision and influence as the actualization of this potential. Following French and Ravens (1956) we treat authority as the 'legitimate' component of power.

B) explicit study of the role of interest. Specifically, we seek to study the effect of having varying amounts of interest in a set of decisions on the exercise of power in those decisions.

C) incorporation of more than one decision in the model.

D) the formal specification of models/mechanisms by which interest and power are transformed into influence patterns and decision outcomes.

Thus, in terms of Figure 1, the thrust of this research deals with the boxes titled Interest, Power, Conflict Resolution Mechanism, and Decision Outcome. Given the early stages of our research we shall not study, in a detailed manner, the post-purchase processes. However, we include them in our description since they are important components of the process.

SOME RESEARCH QUESTIONS

Based on the literature review and the stages t of the decision process chosen for study, we propose the following questions:

A) Does power in a decision predict the influence o f a person in a given decision? The traditional assumption has been that the joint decision will be congruent with the preferences of the more powerful member.

B) Does interest in a decision predict the influence of a person in a given decision?

Two conflicting concepts of the role of interest in a decision have been suggested in the literature, each implying different predictive models. The first concept, proposed by Homans (1974) may be termed the principle of least interest. According to this proposition, "that person is able to dictate the conditions of association whose interest in the continuation of the affair is least" (Thibaut and Kelley, 1959). Nash's (1950) solution to the bargaining problem is an example of this concept. According to this game-theoretic model of the bargaining problem, the dyad should choose the alternative that maximizes the product of the incremental utilities of the two participants. Thus, if two people were to. divide a hundred dollars between them, with neither getting anything if an agreement could not be reached, the Nash solution would prescribe splitting the amount evenly (each gets $50.00). However, if in the event of disagreement, person A is assured $50.00 whereas person B would still get nothing, then the Nash solution would prescribe that A get $75.00 and B get $25.00. (The decision is less critical to A than to B). Thus, the person with least interest would get a larger share.

FIGURE 1

A SIMPLIFIED DESCRIPTION OF FAMILY DECISION MAKING

The alternate concept of the role of interest is that the person with greater interest is the person who would dominate. Krishnamurthy (1981) has suggested that "the greater the involvement, the greater the influence of that individual in the joint decision, and the greater the congruence between the joint decision and his decision if he had to decide by himself." He goes on to suggest two models or what the joint preference would be, based on this premise. A game-theoretic solution which would give importance to the person who has more to gain from a decision has been proposed by Kalai and Smorodinsky (1975). According to this formulation the division of the rewards would be in direct proportion to the best payoffs that the participants could get by being in the group. Thus, if A could expect to get $100.00 and B could only expect to get $75.00 then the payoffs to the two would be in the ratio 4:3.

C) Do people really trade power in a decision of little interest for greater influence over a decision of greater interest? If they do, what is the nature of the interaction between the bases of power and interest in the decision?

D) Can we treat a series of decisions as a 'super game'?

We can think of a series of related decisions as a single 'super' decision. In conjoint analytic terminology we could consider each individual decision as an 'attribute' of the 'super' decision and the particular alternatives in an individual decision as the 'levels' of these 'attributes.' In this context the utility for each alternative is analogous to the Part-worth of the attribute level. In this way, the definition of importance in conjoint analysis coincides with our notion of interest. The advantages of treating the sequence of decisions as a single 'super' game is that we can try to use the game-theoretic models to predict the overall decision. The game-theoretic solutions introduced are really meant for static situations and, typically, require that the payoff space be continuous and, in some instances. convex. The latter two conditions are not usually true in many family decision making situations. However, if we have a super game of, say, 6 individual decisions each with 6 particular alternatives, there are (6 =) 46,656 possible outcomes of the super game. Thus, the space is very nearly complete and convex.

An experimental methodology outlined in the next section which would allow testing explicit hypotheses derived from these issues and questions will now be proposed.. Using the terminology of Calder, Phillips and Tybout (1981) the proposed research deals with theory application rather than effects application. The basic aim is to test theories. We would like to have as much control as possible over the situation so that the main effects of and the interactions among the variables of interest can adequately be tested. The special need to do this in the present context is emphasized by the recommendations made by Kriewall (1980), Krishnamurthy (1981), and the results obtained by Buss (1979). In each of these cases, failure to control the situation adequately led to an inability to say which of the various contending hypotheses were or were not supported. Further, the payoffs of adequate control are demonstrated by Steckel's (1981) research.

THE PROPOSED METHODOLOGY

As stated earlier we would like to study the interaction between interest and power in a given situation on the decision outcome. Further, we would like, explicitly, to see the effect of making multiple decisions on these outcomes. These aims require a carefully designed experiment. Some issues associated with operationalization of the major constructs are reviewed in the next sections.

Interest

It is assumed that the interest of a person in a situation is high if his preferences vary a lot over the possible outcomes. Thus, a given individual should be more interested in the decision represented by Figure 2a than in the decision represented by Figure 2b.

FIGURE 2A

PREFERENCES FOR ALTERNATIVES

FIGURE 2B

PREFERENCES FOR ALTERNATVIES

Ideally dyads should make decisions such that some are very important to one of the members and some others are not. These variations would be reflected as differences in individual preferences over the decisions to be studied. Preference functions could be measured using any of a wide variety of measurement approaches available. (All of the models (Coleman, 1973; Kalai and Smorodinsky, 1975; Krishnamurthy, 1981; and Nash, 1950) presented assume that preferences are measured up to an interval scale. Techniques such as Conjoint Analysis (Green and Srinivasan, 1978) and Multi Attribute Utility Theory (Keeney and Raiffa, 1976) could be used). The drawback of this approach for current needs arises from the possibility that we might not get sufficient variations in the measured preferences. This would preclude adequate testing of the different models. In an early pilot, subjects' preferences for three major decisions were measured using conjoint analysis. Out of the twenty-five subjects who responded, only three had sufficiently different preferences. When preferences are similar, all the models make very similar or identical predictions.

To get the desired control, a 'training method' can be used. Two major options exist here. One is represented by the approach used by Fiorina and Plott (1978). Essentially the method amounts to telling the participants exactly what their preference functions are. In a sense the subjects are deterministically trained to use a particular function. The large number of studies using this approach and the general quality of the results obtained from them show that this is a useful and viable method.

A different training method is used by Hammond and his colleagues (Hammond, Brehmer and Steinmann, 1975), in the area of Social Judgment Theory. Here, the training procedure is probabilistic. Subjects typically make a series of judgments about hypothetical stimuli are given some feedback about what the desired response is, and thus infer the preference function. The uncertainty introduced makes this task more realistic. However, the time taken in training the subjects and the possibility of some subjects not learning the preference functions are some drawbacks of this approach.

The method we are adopting (similar to Fiorina and Plott's, 1978, procedure) involves telling each person in a dyad that his/her payoffs from participation in the experiment are directly related to the utility levels assigned to each of the alternatives for that decision. Thus if a person with the utility levels in Figure 2a were interacting with another person with the utility levels in Figure 2b, we would have one person who is very interested in the decision and another who is relatively indifferent. If alternative A is chosen the former would get zero utils and the latter two utils. In this manner we are able to manipulate interest and preference discrepancy deterministically.

Since, in this research, it is not necessary to study the effects of uncertainty but rather maximum control is most important, this is appropriate. It should be noted that within the basic experimental paradigm the effects of uncertainty can be studied and the actual preferences of respondents could be measured.

Power

Power has been defined as the potential of a person to influence the joint decision. Power can arise as a result of referent status, expertise, reward giving ability, legitimacy. The effects of these several variables on the perceived power of an individual in a relationship is a topic of research itself (Thomas, 1980). These variables could be measured separately (in fact this must be done when we want to predict actual decisions) and the composite used as a measure of power. However, once again the problem of lack of control and possible lack of sufficient variability could arise.

The solution is to explicitly manipulate the power construct. The problem of how to manipulate it still remains. For example, one of the members could be given the right to award a certain amount of money to the other person (reward power), or one of the members could be told that the other person is very knowledgeable in a given decision (expertise power). Pilot tests show that an effective manner of manipulating power is by giving one or the other participant the legitimate right, or authority, to make the final choice in a given decision. This right might always reside with the same person or may change from decision to decision.

Once again, it is important to note that the explicit manipulation of power is suggested to ensure sufficient variability to allow adequate tests of the models. In a more general setting, when a better understanding of the process has been gained, we could actually measure the power correlates.

The Experimental Procedure

g he experimental procedure requires each pair to negotiate, face-to-face, six decisions. In each of the six decisions the preferences are kept highly discrepant. However,the interest of an individual in the decisions varies from situation to situation. In addition, one member of the dyad is designated as the one with authority for some decisions. In the others both are given equal authority. The only restriction imposed is that exact utility numbers not be communicated. Decisions are made sequentially and the sequence effect can be controlled for by using designs balanced for order effects.

The Decision Outcomes and Influence Patterns

We can now compare the predictions made by each of the formal models, the supergame, a random model and the traditional power model against the actual decisions made by the dyads Further, in post-negotiation questioning we can also elicit the perceived influence of each individual for each decision separately and for al-l the decisions as a whole. Also, by tape-recording the bargaining sessions we can further test whether people do in fact bargain over decision situations.

Preliminary Findings

The experimental procedure has been tested using undergraduate students as subjects. These tests show that the subjects were reasonably involved in the experimental task. Deterministic training to vary degree of interest and the explicit authority manipulations were successful. Ten of the fourteen pairs bargained across decision situations. In one case the dyad failed to reach agreements on ali six decisions in the allotted time. Two dyads bargained within decision situations and one agreed to make side-payments at the end of the experiment. The latter case has now been controlled by delaying the payment of earned money and pairing people from different classes. Thus, the results are very encouraging.

SUMMARY

This paper has identified some future avenues of research in the area of family decision making. The point of departure from the existing literature lies in the increased emphasis given to the interaction between power and interest in a sequence of decisions. The introduction of game-theoretic models and their relation to the social psychology literature is a relatively novel extension to the traditional ways of conceptualizing family decision making research. An experimental methodology is proposed and its 'translation' to more 'real' problems suggested.- The suggested approach is likely to be more useful for the study of consumer decision making procedures than the traditionally employed paradigms in gaming experiments.

REFERENCES

Burns, A. C., and Granbois, D. H. (1977), "Factors Moderating the Resolution of Preference Conflict in Family Auto Purchasing," Journal of Marketing Research, 14, 77-86.

Burns, A. C., and Granbois, D. H. (1980), "Advancing the Study of Family Purchase Decision Making," in Proceedings of the Tenth Annual Conference of the Association of Consumer Research, J.C. Olson, ed. San Francisco, California: Association for Consumer Research 221-226.

Buss, W. C. (1979), A Comparison of the Predictive Ability of Group Choice Models. Unpublished Ph. D. Dissertation, Wharton School, University of Pennsylvania.

Calder, B. J., Phillips, L. W., and Tybout, A. M. (1981), "Designing Research for Application," Journal of Consumer Research, (September).

Cartwright, D. (1965), "Influence, Leadership and Control," in Political Power: A Reader in Theory and Research, R. Bell, D. V. Edwards and R. H. Wagners (eds.), New York: The Free Press.

Coleman, J. S. (1973), The Mathematics of Collective Action.. Chicago, Ill: Aldine Publishing Co.

Davis, H. L. (1976), "Decision Making Within the Household," Journal of Consumer Research, 2. 741-260.

Fiorina, M. P., and Plott, C. R. (1978), "Committee Decisions Under Majority Rule: An Experimental Study," American Political Science Review, 72, 575-98.

French, J. R. P., and Ravens, B. (1959), "The Bases of Social Power," in Studies in Social Power, D. Cartwright, ed. Ann Arbor: Research Center for Group Dynamics. Institute for Social Research, University of Michigan, 150-167.

Granbois, D. H. (1971), "A Multilevel Approach to Family Role Structure Research," in Proceedings of the Second Conference of the Association of Consumer Research, D. M. Gardner, ed. College Park, Maryland: Association for Consumer Research, 99-107.

Green, P. E., and Srinivasan, V. (1978), "Conjoint Analysis in Consumer Research: Issues and Outlook," Journal of Consumer Research, 5, 103-193.

Hammond, K. R., Brehmer, B., and Steinmann, D. O. (1975), "Social Judgement Theory," in Human Judgement and Decision Processes, M. F. Kaplan and S. Schwartz, eds. New York: Academic Press, Inc.

Harsany, J. C., and Selten, R. (1972), "A Generalized Nash Solution for Two-Person Bargaining Games with Incomplete Information," Management Science, 18, 80-106.

Heckathorne, D. (1978), "A Paradigm for Bargaining and a Test of Two Bargaining Models," Behavioral Science, 23, 73-85.

Homans, G. C. (1974), Social Behavior: Its Elementary Forms (Revised ed. ), New York: Harcourt, Brace and Jovanovich, Inc.

Kalai, E., and Smorodinsky, M. (1975), "Other Solutions to Nash's Bargaining Problem," Econometrica, 43 (3), 513-8.

Keeney, R. L., and Raiffa, H. (1976), Decisions with Multiple Objectives: Preferences and Value Tradeoffs, New York: John Wiley and Sons, Inc.

Kriewall, M. A. O. (1980), Modeling Multi-Person Decision Processes on a Major Consumption Decision. Unpublished Ph. D. Dissertation, Stanford University.

Krishnamurthy, L. (1981), Modeling Joint Decision Making Through Relative Influence. Unpublished Ph. D. Dissertation, Stanford University.

Nash, J. F. (1950), "The Bargaining Problem," Econometrica, 18, 155-62.

Pollay, R. (1969), "Intrafamily Communication and Consensus," Journal of Communication, 19, 181-201.

Samuelson, P. A. (19;6), "Social Indifference Curves," Quarterly Journal of Economics, 70, 1-22.

Sheth, J. N. (1974), "A Theory or Family Buying Decisions," in Models or Buyer Behavior: Conceptual, Quantitative and Empirical, J. N. Sheth, ed. New York: Harper & Row, 17-33.

Steckel, J. (1981), A Game Theoretic View of Group Choice. Unpublished Ph. D. Dissertation, Wharton School. University of Pennsylvania.

Thibaut, J. W.. and Kelley, H. H. (1959), The Social Psychology of Groups, New York: John Wiley and Sons Inc.

Thomas, R. J. (1980), Correlates of Interpersonal Purchase Influence in Organizations. Unpublished Ph. D. Dissertation, Wharton School, University of Pennsylvania.

Turner, R. (1970), Family Interaction. New York: John Wiley and Sons, Inc.

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