Advances in Consumer Research Volume 12, 1985 Pages 487-491
HUSBAND-WIFE DECISION MAKING: AN EXPLORATORY STUDY OF THE INTERACTION PROCESS
David Brinberg, Baruch College, CUNY
Nancy Schwenk, University of Maryland
The present study presented a preliminary analysis of interaction data in couple decision making. A detailed discussion of the methodology used to examine interaction processes was presented as well as a discussion of the limitations and benefits of interaction analysis to examine couple decision making.
Husband-wife interaction in family decision making is an area of consumer behavior that is of growing interest (Kassarjian 1982). In recent years, consumer researchers (e.g., Bonfield 1978; Burns and Ortinau 1978; Davis 1976; Filiatrault and Ritchie 1980; Granbois 1978; Rollins and Bahr 1976; Szybillo, Sosanie, and Tenenbein 1979; Woodside and Motes 1979) have focused husband-wife research on the following questions: (1) What is the relative power of each spouse over the other? (2) Does relative power vary by product, stages in the family life cycle, and decision style? and (3) What is the influence of marital roles in the decision making process?
Most of the research on husband-wife decision making has focused on outcomes, rather than process. A great deal of potential information is lost concerning the decision process by using input-output static models. The current paper is an initial attempt to incorporate a process an lysis to study couple decision making. We will begin this paper with a brief overview of past research on couple decision making, then provide a detailed description of the methodology used in the present study to examine the interaction process between couples. We will then present some findings associated with this process analysis and discuss their implications for future research.
PREVIOUS RESEARCH FINDINGS
Two approaches have dominated the study of husband-wife interaction in family decision making: (1) power allocation in the marriage and (2) role differentiation.
Blood and Wolfe (1960), in their classic work, found the power position to be the most important aspect of family structure; where power was defined as the potential ability of one partner to influence the other's behavior. Blood and Wolfe found that the partner who contributes the greater amount of resources to the marriage (where a resource is defined as anything that can be transmitted to the other to help the latter satisfy his/her needs or attain her/his goals), the greater the power of that individual over the other. These authors found the relative power of one spouse over the other varied as a function of product type (e.g., husbands were more influential in decisions related to the car and career whereas wives were more influential in decisions related to food budgeting). These findings have been criticized by Herr (1963), however, for examining power at a global, rather than a specific level, as well as by Safilios-Rothschild (1970).
Power in family decision making has also been related to social class (Komarovsky 1961), self-concept of power (Woodside 1972) and the impact of children (Filiatrault and Ritchie 1980) on the distribution of power. In addition, Schaninger and Allen (1981) examined the relation between the wife's occupational status and her power in the family.
A consistent theme throughout previous research on power allocation is the focus on static, outcome oriented variables (e.g., the wife or husband estimating their relative influence over the other with regard to some decision). The empirical research we will discuss in this paper may be used as an alternative strategy for assessing power, based on the patterns of interaction, rather than post-hoc self-report estimates.
The second major approach toward the study of husband-wife decision making is the identification of the roles and their relation to the decision process. Past work in this area has focused on role structure (i.e., husband-dominant, wife-dominant, etc.) as well ag the extent of role specialization (see Engel and Blackwell 1983; Schiffman and Kanuk 1983 for a more detailed discussion).
Recent work (e.g., Scanzoni and Szinovacz 1980), however, has focused on traditional vs. modern sex roles. Traditional sex roles have been found to be sharply different and rigid, and tent to make family decision making quite simple; that is, the expected behaviors of the husband-wife are quite clear. In the more modern sex role, the behaviors expected of each spouse are becoming less predictable because of the increased flexibility and freedom associated with the role. Consequently, the decision process for those couples has become more complex. In a study of changing ferule roles on family purchase patterns, Green and Cunningham (1975) found that husbands of modern wives tend to make fever purchase decision than husbands of traditional wives. A more detailed review of the literature on role differentiation may be found in Schwenk (1983).
The research on role differentiation has also focused on static, outcome oriented models. One aspect of that research which we will examine further in the current study is the relation between traditional vs. modern couples and their patterns of interaction.
PROCESS ANALYSIS IN COUPLE DECISION MAKING
Interest in couple decision making also has a research history in clinical and social psychology. One outgrowth of that interest has been the development of coding schemes to record the exchanges between two (or more) people. Bales (1950) provided an early effort in the use of a coding scheme to categorize the interaction among group members, That scheme (described as the Interaction Process Analysis) used an instrumental and expressive dimension as the bases for the 12 categories in the IPA. Several researchers (e.g., Waxler and Mishler 1965; Winter and Ferreira 1967; Raush, Barry, Hertel and Swain 1974) adapted Bales IPA to study the interactions in families. Unfortunately, all these researchers found that high levels of interjudge agreement could not be established because of the ambiguity in the categories.
McGrath and Kravitz (1982) noted that the late 1970's brought a resurgence of interest in group interaction. Perhaps the most exciting research in this area has focused on verbal and nonverbal communication processes. New category systems were developed (Gottman 1979; Bales and Cohen 1979) as well as new technology for the analysis of process data. In the current study, we will use the Couple Interactive Scoring System (CISS) developed by Gottman (1979) as well as the analytic techniques he described for the analysis of process.
The main focus of the current study is the adaptation of the CISS to study the interaction process between couples who are making a decision. We examined two categories of decisions: (1) product related decisions and (2) interpersonal decisions. Because of the recent interest in family decision making research on traditional vs. modern roles, we also examined how the patterns associated with these two types of roles differ in regard to their decision process. This study was designed as a hypothesis-generating rather than a hypothesis-testing project. Our plan is to use the patterns we identify in this research to form hypotheses that can be tested in future work.
Our purposes in this paper are two-fold: (1) to introduce consumer researchers to a set of conceptual and analytic tools that will allow them to examine process data, and (2) to present some preliminary findings of such data.
The Couple Interactive Scoring System (CISS) developed by Gottman (1979) was used to categorize the interactions of the couples in the current study. When using this scheme, the couple's conversation is taped and transcribed verbatim. Statements are then divided into "thought units," which provide the basic unit of analysis for the CISS. The "thought unit" can bc a phrase, a sentence, or a speech fragment and is usually separated by pauses, commas, ands, buts, or periods. Each thought unit is placed into one and only one of the categories in the CISS. The categories are as follows:
(1) Agreement - which includes direct agreement, accepting responsibility for a past or present problem, accepting modification and compliance with a preceding request or command,
(2) Disagreement - which includes direct disagreement, denial of responsibility, qualified agreement or apology, disagreement with rationale or justification supplied,
(3) Communication Talk - which is a comment on the process of communication, directing the conversation back to a given topic or toward a resolution of a problem, an evaluation of a conversation, or a request for clarification,
(4) Mind Reading - which is an inference about feelings, attitudes, opinions or motives of the spouse or couple, or which attributes past, present, or future behaviors to the other person,
(5) Problem Solving and Information Exchange which includes suggesting a plan or method for how to solve a problem and presenting information or behavioral facts that may relate to the couple,
(6) Summarize Other - which is summarizing the previous statements of the other person, or reviews or summarizes the couple's conversation,
(7) Summarizes Self - which is a review or summary of the speaker's previous statements, and
(8) Expressing Feelings about a Problem - which is a concern about the problem the couple is discussing or a statement of feelings being expressed by the speaker.
A question is placed by the appropriate content code.
The original coding scheme developed by Gottman (1979) had 28 categories and was .00 cumbersome to be used reliably by coders. The eight categories were a refinement of that earlier scheme. In addition to the content coding, Gottman (1979) has also coded affect and nonverbal responses contiguous with the content code. We did not include affect coding or nonverbal coding in the current study but anticipate including these additional channels of information in future work.
Both the husband and the wife were coded separately. We expanded the 8 category coding system because we believed: (a) coders could reliably place the information derived from interaction in the appropriate category and (b) useful and interesting insights may be gained into the decision process with the expanded set of categories. The following are the fifteen content categories used in the present study:
2. Agreement - accepts modification
4. Disagreement - Yes, but...
5. Disagreement - with rationale
6. Disagreement - Command
7. Communication Talk
8. Mind Reading
9. Problem Solving
10. Information Exchange
11. Information exchange - Most important issue
12. Information exchange - Least important issue
13. Summarize other
14. Summarize self
15. Express feelings about decision
Four undergraduates were trained to use the revised CISS. Initially, each coder was given a booklet containing definitions of the CISS codes, abbreviations and examples. Their first task was to memorize the codes, apply then to a set of example., and to generalize to other examples. Each coder was given sample transcripts to divide into thought units, and then place in the appropriate category. Reliability estimates were obtained by comparing the responses of two of the coders as well as with the main experimenter. Training the coders to meet the criterion of 90: reliability took approximately 5 weeks (with 2, 2 hour sessions per week). We used Cohen's (1960) Kappa to estimate the percent of agreement between coders. The Kappa statistic was computed separately for each transcript. In addition to measuring inter-coder reliability, the main experimenter coded randomly selected pages from a transcript for additional reliability checks.
The following criteria were used to select the sample: (1) the couple had to be married, (2) living together, (3) of child bearing age (because of questions concerning the possibility of having and disciplining children). A total of 41 couples took part in the study and all (but one) were interviewed in their home. The interviews lasted between 1 and 2 hours.
The initial set of instructions described the general purpose of the study as we 11 as the use of the audiotaping equipment. The first task for each participant was the completion of three, short questionnaires. The first questionnaire was used to identify the major decision that would be pertinent for that couple. Each person indicated "how likely is it that you and your spouse will make a decision about each of the following within the next year." The product related decisions were: a house, a car, a major furniture purchase (such as living room, dining room, or bedroom furniture), a major appliance, home entertainment equipment (such as TV, stereo, VCR) and making investments (such as stocks, bonds, or real estate). This particular set of products were selected because of past research (Cox 1975; Burns and Ortinau 1978; Woodside and Motes 1979; Park 1982). The interpersonal related decisions were: changing one's job or career, moving, having a child, disciplining a child, taking a vacation and taking in an elderly parent. Each list contained an "Other" category which gave the participant the opportunity to add a decision possibility not included in the original set. Only 3 of the 41 couples used the "other" category. We will present the methodology used to collect both the product and interpersonal decisions, although only the results of product related decisions will be discussed. The product both spouses indicated had the highest likelihood of being decided on in the next year was chosen as the topic of conversation.
The second questionnaire assessed marital roles (Scanzoni and Szinovacz 1980). Each participant was presented a set of 22 statements concerning family roles and asked to respond on a 5 point agree-disagree scale. This questionnaire was used to assess the traditional to modern orientation of each spouse in the marriage.
The third questionnaire assessed marital adjustment (Spanier 1976). Each participant responded on a 6 point agree-disagree scale to a set of 34 statements that assessed four aspects of adjustment: (1) dyadic consensus, (2) affect expressiveness, (3) dyadic satisfaction and (4) dyadic cohesiveness. For the purposes of the present study, all 34 items were used for an overall measure of adjustment (all questionnaires may be obtained from the first author upon request).
After responding to these questionnaires, the couples were informed which product decision they would probably be making in the next year. We selected the product based on the responses to the first questionnaire. The couples were instructed
"To imagine they had decided to purchase ___. I (the experimenter) am interested in finding out how you will choose this product that will best meet your needs and that you both can be happy with. I would like you to discuss with each other, for no more than 30 minutes, the issues you need to consider in purchasing ___. Please do not assume that your spouse is aware of what you think and feel about purchasing ___. You should also not assume that you know what your spouse thinks and feels. It is important that you respond naturally, not at you think you ought to respond. At the end of the discussion, you should hopefully be ready to go out and purchase ___.
At the completion of the first conversation, the individuals were asked to indicate what they considered to be their most (and least) important points. This information was then used in the coding of the interactions. The researcher then repeated the same procedure for the second conversation. The order of presentation for the two decision classes was counterbalanced. At the end of both conversations, the couple was given a brief questionnaire to assess their background (e.g., years married, age, educational background, number of children, income).
A total of 41 couples participated in the study. The average length of marriage was 7.6 years, after having known their spouse an average of 3.3 years before marriage. The husband was two years older than the wife and had one more year of schooling. There was a sizeable difference in income between the husband and the wife, with the husband reporting an average salary of $28,000 and the wife reporting an average salary of $7,000. A total of 68% of the couples had children. There were no significant differences in the dyadic adjustment scores for the husband and the wife. On the marital roles questionnaire, however, women preferred contemporary roles significantly more than the husband. This finding is consistent with Scanzoni and Szinovacz (1980).
Distribution of Decisions
In the product category, the frequencies of each product are as follows: purchasing a car (n-9), purchasing a house (n-9), purchasing home furnishings (n-9), purchasing home entertainment equipment (n-7), making investments (n-4), and "other" (n-3). There was little difference in the husbands' and wives' perceived likelihood of making a decision for each product (range was .3 to 1.1 on a 7 point likely-unlikely scale).
Inter-coder and coder-experimenter reliability coefficients were calculated for each conversation. Inter-coder reliability was calculated across an entire conversation, while coder-experimenter reliability was calculated over randomly selected samples of each conversation. The median inter-coder reliability coefficient was .92 (with a range from .90 to .95). The median coder-experimenter reliability coefficient also was .92 (with a range from .90 to .94).
The primary focus of the present study was to examine the sequential nature of the decision process. Two forms of analyses were conducted. The first examined the relation between: (1) the proportion of occurrence of each of the categories at lag 0 (i.e., the occurrence of each category independent of the other categories) and (2) the perceived marital roles for that couple (i.e., whether the couple perceived themselves to be contemporary or traditional). This relation was determined by correlating the proportion of each category and the couples' score on the marital roles questionnaire. Several findings should be noted, all of which are significant p <.05:
- the more contemporary the couple, the more likely the wife is to direct the conversation back toward a resolution of the problem (i.e., communication talk) (r-.27)
- the more contemporary the couple, the more likely the husband is to agree with the wife (r-.31)
- the more contemporary the couple, the less likely the husband is to disagree with the wife (r=.34)
- the more contemporary the husband, the less likely the wife is to give commands to him (r=.27)
These findings suggest the following trends: (a) there seem to be more positive interactions (i.e., agreement) among contemporary couples, (b) the greater "communication talk" among contemporary couples suggests their decision process is more complex than traditional couples.
The second set of analyses examined the relation between the lag 1 proportions of the categories with the marital roles. The term lag 1 refers to a strategy where each category serves as the "criterion behavior" for each of the other categories. For each category, the conditional probability of that category given the criterion category is determined. The relation between these conditional probabilities and the marital roles was determined by correlating the proportion derived from the conditional probability with the score on the marital roles questionnaire. Because of the large array of data, we will only highlight some of the findings (A more detailed description of these findings may be obtained from the first author).
- the more contemporary the wife, the more likely the husband will agree with her after she disagrees with him (r-.29)
- the more contemporary the husband, the less likely the wife will provide a rationale when she disagrees or make an inference about his feelings after he expresses qualified agreement (i.e., yes, but. .) ( r- .36)
- the more contemporary the couple, the more likely the wife is to exchange information about the issue most important to her after the husband suggests a solution to the problem (r-.28)
- the more contemporary the couple, the less likely the wife is to express her feelings about a problem after her husband has summarized himself (r-.62)
These findings suggest the following trends: (a) as before, there is more positive interaction in the discussions of contemporary couples, (b) contemporary couples are less likely to become involved in cross-complaining, a term used by Gottman (1979) to describe disagreement followed by disagreement, (c) contemporary wives are more likely to emphasize the issue most important to them during the conversation, perhaps indicative of their feelings of equality with their husband and (d) contemporary wives seem less likely to express their feelings about a problem.
The development of theories capable of explaining patterns of interaction between couples making decisions is in very early stages of development. The data collected in the present study are an initial attempt to identify patterns that may provide a basis for such theories. The analyses presented in this paper are quite preliminary. Subsequent work in this area (and with this data set) will provide a richer, more thorough, description of the interaction process.
Several limitations in the present data set and in interaction research should be noted: (a) only a verbal content cote was used to categorize the interactions in the present study. Future research should use additional channels of information, that is, nonverbal and affective sources of information that can be coded concurrently with the verbal categories, (b) the present study only describes patterns that emerged from the interaction, that is, it was hypothesis-generating research. Future research should test the robustness of these patterns across decision classes and using confirmatory, rather than exploratory analytic techniques, (c) the technology Deeded to study interaction processes is more complex than simple self-report research (e.g., video-equipment to tape two individuals interacting, split screen playback equipment that can synchronize the tape of each participant), (d) interaction research is labor intensive, where the training of coders can take between 20-30 hours, and the analysis of a 10 minute interaction can take 2-3 hours (e.g., transcribing the interaction, having 2 (or more) coders categorize the interactions, and (e) the amount of information generated from a sequential analysis is vast (e.g., with 30 codes, there are 900 pieces of information at lax 1).
In spite of these limitations, there are several important advantages in the use of process analysis to study couple decision making: (a) the interaction between the couples can be examined explicitly. In this way, researchers no longer will need to assume certain types of interactions but will be able to examine them directly, (b) there will be less reliance on self-report measures in collecting information and greater use of direct observation and (c) a richer, more complex, and more accurate nomological network of relations will emerge to explain couple decision making.
Bales, Robert F. (1950), Interaction process analysis: A method for the study of small groups, Cambridge, Mass. Addison-Wesley.
Bales, R.F. and Cohen, Stephen P. (1970), SYMLOG, New York., Free Press.
Blood, Robert and Wolfe, Donald. (1960), Husbands and wives: The dynamics of family living, Glencoe, Ill., Free Press.
Bonfield, Edward H. (1978), "Perception of marital roles in decision processes: Replication and extension." Advances in Consumer Research, 5, 300-307.
Burns, Alvin, C. and Ortinau, David J. (1978), "Underlying perceptual patterns in husband and wife purchase decision influence assessments." Advances in Consumer Research, 5, 372-376.
Cohen, Jacob. (1960), "A coefficient of agreement for nominal scales." Educational and Psychological Measurement, 20, 37-46.
Cox, Eli P. (1975), "Family purchase decision making and the process of adjustment," Journal of Marketing, 12, 189-195.
Davis, Harry L. (1976), "Decision making within the household," Journal of Consumer Research,2, 241-257.
Engel, James F. and Blackwell Roger, D. (1982) Consumer behavior. New York., Dryden Press.
Filiatrault, Pierre and Ritchie, J.R. Brent. (1980), "Joint purchase decisions: A comparison of influence structure in family and couple decision making decision units," Journal of Consumer Research, 7, 131-140.
Gottman, John, M. (1979), Marital interaction: Experimental investigations, New York., Academic Press.
Green, Robert T. and Cunningham, Isabella. (1975), "Feminine role perception and family purchasing decisions," Journal of Marketing Research, 12, 325-332.
Herr, David H. (1963), "The measurement and bases of family power: An overview," Journal of Marriage and the FamilY, 25, 72-81.
Kassarjian, Harold (1982), "Consumer Psychology," Annual Review of Psychology, 33, 619-649.
Komarovsky, Mirra (1961), "Class differences in family decision making on expenditures." In Nelson Foote (Ed.) Household Decision Making, New York., New York University Press.
McGrath, Joseph E. and Kravitz, David (1982), "Group research," Annual review of Psychology, 33, 195-230.
Park, C. Whan (1982), "Joint decisions in home: A muddling through process," Journal of Consumer Research, 9, 151-162.
Raush, H.L., Barry, W.A., Hertel, R.K., and Swain, M.A. (1977), Communication, conflict and marriage. San Francisco., Jossey-Bass.
Safilios-Rothschild, Constantins (1970), "The study of family power structure: A review 1960-1969," Journal of Marriage and the Family, 32, 539-553.
Scanzoni, John and Szinovacz, Maximiliane (1980), Family decision making. Beverly Hills, Ca., Sage Publications.
Schaninger, Charles, M. and Allen, Christopher S. (1981), "Wife's occupational status as a consumer behavior restraint," Journal of Consumer Research, 8, 189-196.
Schiffman. Leon and Kanuk, Leslie (1983), Consumer behavior. Englewood Cliffs, N.J., Prentice-Hall.
Schwenk, Nancy (1983) "Exploratory analysis of husband-wife decision making " Unpublished Master's thesis. University of Maryland.
Spanier, Graham (1976), "Measuring dyadic adjustment: New scales for assessing the quality of marriage and similar dyads," Journal of Marriage and the Family, 1, 15-28.
Szybillo, George, J., Sosanie, Arlene, R., and Tenenbein, Aaron (1979), "Family member influence in household decision making," Journal of Consumer Research, 6, 312-316.
Waxler, N.E. and Michler, E.G. (1965), "Scoring and reliability problems and interaction process analysis: A methodological note," Sociometry, 29, 32-49.
Winter, W.D. and Ferreira, A.J. (1967), "Interaction process analysis of family decision making," Family Process, 6, 155-172.
Woodside, Arch G. (1972), "Informal group influence on risk taking," Journal of Marketing Research, 9, 223-229.
Woodside, Arch G. and Motes, William H. (1979), "Husband and wife perceptions of marital roles in consumer decision processes for sis products," Educators Conference Proceedings, American Marketing Association, 214-219.