Toward Understanding the Dynamics of Household Decision Conflict Behavior

William J. Qualls, The University of Michigan
ABSTRACT - The underlying structure of household decision conflict and conflict resolution behavior is examined using the structural modelling technique partial least squares. Household conflict and conflict resolution are treated as multivariate constructs across multiple levels of a household automobile purchase decision. A process oriented framework of household decision conflict behavior is developed and tested. The results of the study indicate that there are multiple levels of conflict in a typical joint household purchase decision rather than an overall level of household conflict behavior found in previous research.
[ to cite ]:
William J. Qualls (1988) ,"Toward Understanding the Dynamics of Household Decision Conflict Behavior", in NA - Advances in Consumer Research Volume 15, eds. Micheal J. Houston, Provo, UT : Association for Consumer Research, Pages: 442-448.

Advances in Consumer Research Volume 15, 1988      Pages 442-448

TOWARD UNDERSTANDING THE DYNAMICS OF HOUSEHOLD DECISION CONFLICT BEHAVIOR

William J. Qualls, The University of Michigan

ABSTRACT -

The underlying structure of household decision conflict and conflict resolution behavior is examined using the structural modelling technique partial least squares. Household conflict and conflict resolution are treated as multivariate constructs across multiple levels of a household automobile purchase decision. A process oriented framework of household decision conflict behavior is developed and tested. The results of the study indicate that there are multiple levels of conflict in a typical joint household purchase decision rather than an overall level of household conflict behavior found in previous research.

INTRODUCTION

For many family decisions, joint decision behavior is becoming the norm (Haas 1980) rather than the exception (Davis and Rigaux 1974) in today's households. One result of the increase in joint family decision making (FDM) between husbands and wives has been an increase in the potential for conflict arising in household decisions. Yet, most of FDM research has dealt with the decision role structure of the household focusing primarily on "who decides" (Davis 1976); seldom examining the role of conflict. Even studies which have examined the nature of conflict behavior specifically, tend to ignore the process orientation in which conflict exists, emphasizing more how conflict is resolved (Sheth 1974, Davis 1976, Burns and Granbois 1977, Belch et al. 1980, Seymour and Lessne 1984). Research is still needed to understand the process of conflict behavior and the role it plays in explaining how families make decisions. The purpose of this paper is to examine the process of conflict and the nature of conflict resolution in household decision behavior.

The Nature Of Household Conflict Behavior

Research to date has failed to adequately capture the process of FDM conflict even though researchers have called attention to the need for models of FDM that take conflict into account, (Granbois 1971, Belch et al. 1980, Spiro 1983). Granbois (1971) has long contended that household conflict behavior better reflects the dynamic process of FDM than more traditional input-outcome models of FDM. To date the majority of FDM conflict research has focused on outcomes of the process; specifically conflict resolution.

Sheth (1974) in an adaptation of the March and Simon framework for intergroup conflict proposed four modes of family conflict resolution including: problem solving, bargaining, persuasion, and politics. As March and Simon, Sheth contends that FDM conflict arises as a result of 1) differences in purchase goals and 2) differences in perceptions regarding decision alternatives. As such in a typical household decision the potential for conflict is possible at two levels or stages during the same decision.

Conversely, Davis (1976) contends that most FDM is either consensual or accommodative in nature, based on predetermined roles and areas of decision responsibility. As such even for decisions involving both the husband and wife, very little conflict would be expected or perceived. According to Davis (1976) consensual FDM takes place when there is agreement about purchase goals and decision alternatives; while accommodative behavior occurs when there is disagreement regarding purchase goals or decision alternatives or both. Such a framework is appropriate for classifying types of conflict, yet it is implied that conflict is likely to occur at various levels or stages during the decision process. Combining both the conceptual frameworks of Sheth (1974) and Davis (1976), Belch et al. (1980) constructed a seven category framework to describe household conflict resolution behavior. They found that couples have preferred strategies for resolving conflict. With few exceptions such studies have averaged individuals' conflict perceptions to derive some overall measure of conflict Unfortunately, most of the FDM conflict research has only examined the issue from an overall level of household conflict behavior.

Sheth and Cosmas (1975) attempted to empirically test the use of the four alternative modes of conflict resolution proposed by Sheth (1974). They found that households used some form of conflict resolution more frequently (persuasion) than others (problem solving). Conversely, Belch et al. (1980) found that problem solving was the conflict resolution mode most frequently employed with bargaining and persuasion being used less frequently. One possible explanation for the discrepancy between the two studies results could be related to the fact that each study was tapping a different level of conflict behavior. The possibility also exists that such studies may have unintentionally confounded the perception of conflict resolution by employing a single summative measure of conflict. Perceptions of conflict resolution may represent the outcome of a process in which the husband and wife compares their conflict behavior during earlier stages of the FDM process. Finally, the findings from both studies could have resulted from the tendency for husbands and wives to behave in the manner expected by society.

While other studies have examined the issue of FDM conflict, they have focused primarily on the determinants of conflict behavior. Burns and Granbois (1977) found that involvement, empathy, and authority moderate the relationship between FDM and conflict resolution behavior and the degree of these constructs presence in FDM in most cases reduces the need for conflict resolution behavior. More recently, Seymour and Lessne (1984) have focused on examining the arousal of conflict in FDM. They contend that better measures are needed in order to examine a process oriented perspective of household conflict behavior. Thus our knowledge to date regarding FDM conflict has done a good job of identifying the determinants of conflict behavior and the outcomes (modes of conflict resolution), but needs to be further developed in order to capture it in process orientation. The purpose of the present paper is to test such a framework. The model that is proposed in the next section and tested in the paper is meant to be relational in nature and is not meant to indicate specific cause and effect relationships. Specifically the issues to be addressed in the current study include:

1) To what extent does conflict behavior occur at different levels (stages) of a joint household decision.

2) To test a process oriented theoretical framework which explains the network of relationships found household decision conflict behavior.

A Model of Household Decision Conflict Behavior

The review of previous research confirms the need to develop and test models of FDM that explicitly examines the role of conflict behavior in such models. The model of conflict behavior presented here integrates several compatible perspectives of FDM research to examine the nature of conflict based on a process orientation. Consistent with previous research in FDM, the model assumes that husbands and wives will have different perceptions regarding their influence in the outcomes of a household decision. Figure 1 illustrates the proposed model to be tested in the present study. In this model there are three main classes of variables which influences the degree of conflict and conflict -resolution. The perception of influence perceived by the husband and wife affect the nature of conflict and conflict resolution both directly and indirectly, (Burns and Granbois 1977, Qualls 1984). The perceived influence held by husbands and wives is captured by measures of their sex role orientation, proportion of shared influence, involvement in the product, and the importance of the decision. The second set of variables affecting conflict behavior is the influence attempts (Spiro 1983) by the husband and wife. Based on the husband and wife perceptions of influence and their perception of their attempts to influence their spouses' decisions, conflict will arise. Conflict is conceptually defined in the present study as consisting of disagreement between husbands and wives over household purchase goals, decision alternatives, or the decision process itself. Finally the mode of conflict resolution is a function of the degree of conflict, husband and wife influence attempts, and the perceptions of influence held by the husband and wife. Each of the major constructs hypothesized in the model is examined in detail below.

FIGURE 1

A MODEL OF HOUSEHOLD DECISION CONFLICT BEHAVIOR

Household decision role structure as a function of the level of perceived influence is the most explored aspect of FDM, (Davis and Rigaux 1974, Davis 1976, Qualls 1984). The degree of influence in household decisions is reflected in who decides or perceptions of how a decision is made. Influence is defined in the present study as the perception of the action taken by one spouse to obtain his or her most preferred decision outcome while simultaneously stopping the attainment of their spouses' most preferred outcome. Structurally, husband and wife influence is determined by an individual's perception of influence held for a particular decision, the importance of that decision, their involvement with the product, and the sex role orientation of the spouse.

Yet, mere perceptions of influence is ineffectual, unless one actually attempts to influence the other spouse's behavior. Spiro (1983) contends that it is the influence attempts by a spouse that serves as a more accurate indicator of the process of FDM. Specifically, six different influence strategies have been identified as those being the most- often employed by husbands, and wives include:

Expert- refers to the spouse who has superior information regarding decision alternatives.

Legitimate- refers to the spouse who attempts to influence the other based upon their position in the household.

Bargaining- involves an attempt by a spouse to gather influence presently to be reciprocated to the other spouse at some future date.

Reward/Referent- is an attempt by a spouse to influence the behavior of an another by offering a reward.

Emotional- involves a spouse using some emotional laden reaction in the hope of influencing the other's behavior.

Impression Management- involves any persuasive attempts by one spouse to influence the behavior of the other.

In the present study the emphasis is placed on the influence strategies used by both spouses and not the individual influences strategies themselves.

The use of a particular influence strategy itself may be sufficient to cause disagreement between husbands and wives, but in addition conflict is a function of the influence perceptions of both the husband and wife. As such conflict is defined as the level of disagreement that exist between husbands and wives regarding decision alternatives. - It's our belief that conflict exists in two forms, perceived conflict and manifest conflict. As a perception, conflict is affective in nature referring to household disagreement that is felt. As manifest conflict, it is behavioral in nature, referring to actual disagreement over household decisions between husbands and wives.

In the proposed model manifest conflict is structurally determined by the level of preference discrepancy (Bums and Granbois 1977), while the perception of cognitive conflict (Hammond 1973), and conflict arousal (Seymour and Lessne 1984) serve as -determinants of perceived conflict. Preference discrepancy refers to actual disagreement over several different decision alternatives. Cognitive conflict is the awareness of inconsistent preferences between husbands and wives when based OD the same set of decision alternatives. As such cognitive conflict exists when one spouse perceives that their individual preferences are different from that of their spouses. While the former represents an actual measure of disagreement, the latter is a perceived discrepancy based on the spouses' rating of their partners desired set of decision alternatives. Finally, conflict arousal is defined as the perceived level of dyadic discordance during a joint household decision (Seymour and Lessne 1984). It refers to a husband and wife orientation or tendency towards conflict.

When conflict exists in an household, it must be resolved in order for the household to continue to function. As such, alternative modes of conflict resolution must be considered in order for a decision to take place. The method by which husbands and wives resolve decision conflict is determined by the level of husband and wife influence, their attempts to influence each other, and the degree of conflict. The framework used here draws heavily from the work by Thomas (1973), which combines two dimensions cooperation (attempting to satisfy the other spouse's concerns) and assertiveness (attempting to satisfy one's own concern), as the foundation for the framework for handling conflict. The four modes of conflict resolution used in the present study include:

Competition-represents an attempt by a spouse to completely dominate a decision by force. The objective is to satisfy one's own personal preference regardless of the preference of their spouse.

Concession-involves a spouse adhering to the preferences of the spouse unconditionally or conditionally in return for future considerations

Avoidance-Withdrawal- refers to inaction to avoid an argument by either spouse to resolve the conflict in the hopes that time will help to sway the other spouse from their original position to one closer to their own.

Bargaining- is characterized by spousal behavior in which there is an effort by both spouses to achieve a balance in their original preference positions through compromise.

The measures of conflict resolution, conflict, influence attempts, and perception of influence is summarized in Table 1. The next section presents the methodology employed and a brief discussion of the analysis technique.

METHODOLOGY

In-home personal interviews were conducted for 63 husband-wife households in a medium size midwestern city with a population of about 100,000 people. Couples for the study were solicited through a circular distributed throughout the complex from a large cooperative family unit apartment. Each respondent was paid a ten dollar honorarium for their participation in the study. Primarily a convenience sample, the respondent group is not meant to be generalizable to the population, but it is clearly representative of the potential target market for automobiles. The average age of husbands and wives was 33, with at least one spouse employed full-time; Average household income was approximately $21,000. Seventy Five percent of the respondents had at least a four year college degree.

Household couples were told they were participating in a survey by a major automobile firm who wanted information on the preferences of consumers regarding future automobile purchase behavior. A debriefing session held with each couple revealed that respondents were not aware of the true purpose of the study which was to examine conflict and conflict resolution behavior. Husbands and wives were first interviewed individually to obtain their responses on a series of paper and pencil tasks. Upon completion of the individual questionnaires, husbands and wives were brought together in a room in which they jointly participated in an automobile purchase decision task.

The automobile purchase was chosen as the decision task because it is commonly accepted as a joint decision involving several subdecisions (Davis 1970, Green and Cunningham 1975). Couples were told that they were participating in an hypothetical decision in which they had to make several subdecisions regarding their purchase of an automobile. Specifically, these subdecisions included:

1) What should the make/model of the automobile be?

2) What color automobile should be purchased?

3) Should the automobile be bought or leased?

4) What options should be included on the automobile?

For each of the subdecisions (excluding the buy/lease subdecision) at least six alternatives were listed. The survey took approximately 50 minutes to complete. Respondents were to pretend they were in the process of buying a new automobile. The only restriction was that they could only be approved for a loan up to 520,000. Based on the subdecisions listed previously they were to rank order the alternatives for each of the subdecisions.

The questionnaire contained measures to obtain information on six different latent variables hypothesized in the model. Separate measures were obtained for both husbands and wives for each manifest variable, and jointly for ten manifest variables. A summary description of each of these measures are listed in Table 1.

TABLE 1

SUMMARY OF LATENT CONSTRUCT MEASURES

The objective of the analysis was to examine the significance of the relationship of the conceptual network of household conflict behavior. Data from the study were analyzed via partial least squares (PLS) a structural modelling technique in which hypothesized latent constructs are estimated indirectly through a set of observables manifest variables (Wold 1982). PLS offers some advantages over the more popular maximum likelihood estimation procedure LISREL. PLS is not as restrictive with respect to sample size and distribution requirements. PLS maximizes the degree of explainable variance for a network of hypothesized relationships, while allowing interactions between the models structural variables, and observable constructs. Hypothesized constructs are estimated by way of a series of interactive regressions, in which each latent variable is predicted as a function of its observed indicators. PLS as an analytical technique is gaining popularity as a method for developing and testing theory. (Fornell and Larcker 1982, Fornell and Robinson 1983, Fornell et al 1983). The interested reader is referred to these sources for a more detailed explanation of PLS. Lohmoller (1981) LVPLSC program of latent variable path analysis with partial least squares estimation is used to derive the models' path estimates. Tables 2 and 3 present the results of PLS analysis of the measurement and structural models respectively and are discussed in the next section.

RESULTS

While the benefits of capturing a process oriented view of household decision conflict are many, the difficulty lies in measuring such a complex process. In PLS there is no true overall test of the hypothesized model or an assessment of the fit of the model to the data. Fornell et al. (1983) have developed a set of descriptive statistics useful in evaluating PLS models. Based on the data analysis suggestions by Fornell et al. (1983), the results of the analysis of the structural model and measurement model hypothesized are discussed below.

Table 2 lists the results of the analysis of the - measurement model. The measurement model exemplifies the meaning of the measures that is gained from being hypothesized in its present theoretical network of household relationships. Specifically, the PLS estimates of the factor loadings, the reliability of each measure, and the average variance extracted (ave) is illustrated. Nunnally (1978) criteria of .5 for reliabilities, and Fornell and Larcker (1981) suggestion of .5 is used to evaluate the acceptability of the measurement modes.

Examination of the measurement model results indicates that the hypothesized measures are at best suspect. Only 13 of the 32 measures exceed the minimum criterion suggested by Nunnally for exploratory research. Low loadings provide further evidence that the measurement model could be improved. The presence of low loadings for the measurement variables can be traced to the multidimensionality of latent construct, or the unreliability of the measures themselves. The latter reason is the most likely explanation. Similarly, the average variance extracted by each of the hypothesized latent constructs fails to exceed the suggested minimum criterion of .5 by Fornell and Larcker (1981). The solution to the low reliabilities is found in the development of improved measures, while low loadings can be handled by a revised model. Yet the focus of the present study goes beyond the individual measures to the network of hypothesized relationships. The structural model provides results which are more appropriate for evaluating the system's view of household conflict behavior.

A more appropriate evaluation of the hypothesized model is the explanatory power or the ability of the model to predict based upon the hypothesized theoretical network. In addition, specific hypothesized relationships in the model can be tested by examining the size, sign, and significance of the path coefficients. The results of the structural model analysis are presented in Table 3. The results of this analysis are much more encouraging when 47% of the variance in household conflict resolution is accounted for by the present network of hypothesized relationship. Similar findings are found for both husbands' and wives' influence attempts accounting for 45 and 30 percent of the variance respectively. Conversely minimum explanatory power is captured by the latent construct household conflict. Such results are encouraging for exploratory research of the nature presented here.

Further examination of the models' path coefficients reveal conflicting findings. Although the signs of the path coefficients are in expected direction, most are relatively small and, there were more nonsignificant paths than there were significant paths.

DISCUSSION

Underlying the model developed and tested in the present study is the contention that household decision behavior can be better understood by examining conflict behavior and how it's handled. While the explanatory power of the proposed model is encouraging, the size of the path coefficients could be due to the measurement problems discussed earlier.

TABLE 2

RESULTS OF PLS ANALYSIS MEASUREMENT MODEL

TABLE 3

RESULTS OF PLS ANALYSIS STRUCTURAL MODEL

Even with these measurement difficulties, some interesting and encouraging findings emerge. For example, the measurement of husband and wife influence revealed substantially different perspectives regarding the foundation of the influence perceived by husbands. It was their perception of the distribution of influence that keyed its impact on other household behaviors, whereas the wives sex role orientation, product involvement, and importance of the decision contributed more to the impact of wives' influence on the decision. Such findings suggest that husbands' preconceived notions of the traditional household roles has changed very little. The loadings indicate that with respect to the automobile purchase decision, that husbands are still traditional in their orientations, while wives would like to be more egalitarian.

A second encouraging result was the ability to explain what could be classified as more active or process oriented household behavior than specific decision acts. Specifically, for household behaviors such as husband/wife influence a tempts and conflict resolution, a larger portion of the explainable variance was captured than for household behavioral acts such as conflict.

Similarly, the ability of the model to capture conflict behavior over a series of automobile subdecisions overcame earlier problems which only addressed the issue of overall conflict. The explanatory power of the conflict resolution construct shows consistent results with past findings (Belch et al. 1980, Burns and Granbois 19471). The decision process, or subdecisions in the process, affect the method in which conflict is resolved. The antecedents to household conflict resolution cross all levels/stages of the decision process. As such decisions made during one stage impact and affect decision behaviors in different stages.

A comparison of husband and wife responses shows some differences, but generally husbands and wives view conflict behavior and the handling of conflict in a similar way. What is evident is that household conflict behavior is more complex than the model proposed here. Although the model proposed here takes a process-oriented view and is more complete than others, it is not surprising that there are other factors which may account for the unexplained variance. For example, Burns and Granbois (1977) have shown the effect of certain moderating factors in reducing the need for conflict resolution.

The limitations of the present study may also account for the findings. Beyond the measurement issues, the sample size and its cross sectional makeup contribute to the lack of stronger findings. As such, generalizations to other populations or causal conclusions drawn from the hypothesized model should be limited. In addition, the design of the study which is based upon both perceptions and behaviors may raise a few questions. Yet, it is viewed as a particular strength of this study over previous studies, that perceptions and behavior were measured simultaneously.

The nomological validity of the proposed process model of household decision conflict behavior is the most significant contribution of the present study. If the degree to which predictions are confirmed based upon the theoretical framework containing the hypothesized constructs, the high- R2 obtained for the model is just one indicator of the model's predictive capabilities. While other researchers have called for the need to examine household behavior from a systems perspective, the present study is one of the few attempts which has accomplished this task. The findings, although weak, are suggestive of a theoretical framework which can be developed through further research. Thus, the general framework developed and tested here can serve as a guide to a program of research using process perspectives in examining household decision behavior of particular interest. Future research should address such questions as: 1) Does conflict differ during different stages and/or do different factors affect conflict and conflict resolution during these stages?

Answers to such questions will move us beyond the complaints of researchers that knowledge of FDM is limited to just knowing "who decides."

REFERENCES

Belch, M., G. Belch, and D. Sciglimpaglia (1980) "Conflict in Family Decision Making: An Exploratory Investigation," Advances in Consumer Research Vol VII, J. Olson (ed) Chicago; Association for Consumer Research p. 475-479.

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

Davis, H. (1976) "Decision Making Within the Household" Journal of Consumer Research, 2 (March) 241-260.

Davis, H. and B.P. Rigaux (1974) "Perception of Marital Roles in Decision Processes" Journal of Consumer Research, 1, (June), 51-42.

Fornell C. and D. Larcker (1981) "Evaluating Structural Equations Models with Observable and Measurement Error" Journal of Marketing Research, (February) 39-50.

Fornell C. and W. Robinson (1983) "Industrial Organization and Consumer Satisfaction/Dissatisfaction" Journal of Consumer Research (March) 443-448.

Fornell, C., G. Tellis, G. Zinkhan (1982) "Validity Assessment: A Structural Equations Approach Using Partial Least Squares" An Assessment of Marketing Thoughts and Practice, B. Walker et al. (eds) Chicago: American Marketing Association 405-409.

Granbois, D.H. (1971) "A Multi-Level Approach To Family Role Structure Research" Advances In Consumer Research Vol II, D. Gardner (ed) Association For consumer Research;, 99-107.

Green, R. and I. Cunningham (1975) "Feminine Role Perception and Family Purchasing Decisions" Journal of Marketing Research, 12, (August), 325-332.

Haas, L. (1980) "Role Sharing Couples: A Study of Egalitarian Marriages," Family Relations, 29, (July) 289-296.

Hammond, K. (1973) "The Cognitive Conflict Paradigm" by L. Rappoport and D. Summers (Eds.) Human Judgment and Social Interaction, New York: Holt, Rhinehart, Winston, p. 188-205.

Lohmoller, J.B. (1981) LVPLS 1.6: Latent Variable Path Analysis with Partial Least Squares Estimation, University of Federal Armed Forces.

Qualls, W. (1984) "Sex Roles, Husband-Wife Influence, and Family Decision Behavior" Advances in Consumer Research Vol XII. T.C. Kinnear (Ed) Ann Arbor MI: Association For Consumer Research.

Seymour D., and G. Lessne (1984) "Spousal Conflict Arousal: Scale Development" Journal of Consumer Research, 2, (December), 810-821.

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

Sheth J. and S. Cosmas (19753 "Tactics of Conflict Resolution in Family Buying Behavior" presentation at American Psychological Association, Chicago.

Spiro, R. (1983) "Persuasion In Family Decision Making" Journal of Consumer Research, 4, (March) 393-402.

Thomas, K. and R. Kilmann (1973) "Some Properties of Existing Conflict Behavior Instruments" Working Paper #73-11, UCLA.

Wold, H. (1982) "Systems Under Indirect Observation Using PLS" A Second Generation of Multivariate Analysis Methods: Vol I. C. Fornell (Ed) New York: Praeger.

Zaichowsky, J.L. (1985) "Measuring the Involvement Construct" Journal of Consumer Research, 12, (3), 34 1-52.

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