Perception of Marital Roles in Decision Processes: Replication and Extension

E. H. Bonfield, Temple University
ABSTRACT - A study by Davis and Rigaux (1974) was replicated with two changes: greater specificity in describing decision items and use of a nonconvenience, random (cluster) sample. The earlier research was extended by checking for differences between traditional couples (masculine husband with feminine wife) and all couples. With some exceptions, generally attributable to greater item specificity, the Davis and Rigaux results were supported. Differences between traditional and all couples were found.
[ to cite ]:
E. H. Bonfield (1978) ,"Perception of Marital Roles in Decision Processes: Replication and Extension", in NA - Advances in Consumer Research Volume 05, eds. Kent Hunt, Ann Abor, MI : Association for Consumer Research, Pages: 300-307.

Advances in Consumer Research Volume 5, 1978      Pages 300-307


E. H. Bonfield, Temple University


A study by Davis and Rigaux (1974) was replicated with two changes: greater specificity in describing decision items and use of a nonconvenience, random (cluster) sample. The earlier research was extended by checking for differences between traditional couples (masculine husband with feminine wife) and all couples. With some exceptions, generally attributable to greater item specificity, the Davis and Rigaux results were supported. Differences between traditional and all couples were found.


In a recent study, Davis and Rigaux (1974) explored the self-reported influence exerted at three general phases of the consumer decision process. The general theoretical and methodological notions were useful, but two, recognized, shortcomings reduced the overall value of the study. First, the researchers utilized a convenience sample, making questionable any generalizations. Secondly, the 25 economic decisions were, for the most part, general in nature, leading to a possible masking of role specialization at a more "micro" level. For example, Davis and Rigaux defined housing as a single item with "location, purchase price, or rent" in parentheses. However, previous research (e. g,, Hempel, 19-72) has shown husband dominance, wife dominance, and Joint decisions for various aspects of the housing decision. The present replication attempts to correct for these two potential contaminants, thus adding generalizability to the results.

A related question deals with sex roles. In an environment increasingly subject to changes in the perception of sex roles, it is possible traditional couples--masculine husband with feminine wife--have different perceived marital roles in the decision process than couples, for example, in which both mates are androgynous in sex role orientation.

This study was thus addressed to the same two questions as was the research by Davis and Rigaux (p. 51):

1. Do marital roles in consumer decision-making differ by phase of the process?

2. To what extent do husbands and wives agree in their perception of roles at various phases of the decision process?

In addition, this study was also addressed to the question:

3. Do marital roles and husband-wife perception agreements differ among traditional (masculine husband/ feminine wife) households as compared to the more general population.


Consistent with the replication-extension nature of this study, the methodology closely follows that of Davis and Rigaux (1974).

The Sample

Separately administered questionnaires were successfully completed by both spouses in 60 households selected by means of a cluster sample (Sudman, 1976). Clusters of respondent households were drawn from census blocks in a southeastern SMSA, the blocks and households having been drawn randomly. Only high income, high home value blocks were included in the population. Thus, while the generalizability of the following results is limited to relatively upper social class couples in a single region, the population is specifiable. The relatively high class status of the sample was reflected in the reported occupations of husbands, 61% professional and 25% white collar. While 65% of the wives reported they were homemakers, 16% reported they were professionals while an equal proportion reported they were white collar workers. Nearly half the sample reported annual family incomes of $30,000 or more. Almost 75% of the husbands and 47% of the wives had earned at least one college degree and an additional 40% of the wives had some college education. The median age of couples of approximately 34 years was apparently greater than in the Davis and Rigaux study since only 12% (compared to 36%) were in their twenties. This age difference was reflected in the family sizes in that fewer had no children, 14% to 25%, and more had at least three children, 33% to 28%.

The Questionnaire

As in the Davis and Rigaux study, each respondent gave information about who was the first to recognize a need, who collected the information about possible choices, and who made the final decision to buy, for 20 decisions in this case. The answer categories in each of the resulting 60 questions were Husband, Wife, and Joint; later scored 1, 3, and 2 respectively. While the decision items were in most cases the same or similar to those selected by Davis and Rigaux, an effort was made to describe each item as specifically as possible without making the task too tedious for the respondents. Thus, while Davis and Rigaux asked about living-room furniture, for example, respondents in the present study were asked about a sofa for the living room or family room. The complete list of items is included in Table 2, and a comparison of item names between the two studies is included in the Appendix. Use of more specific items limits the generalizability of the results to those items, but increases the degree of generalizability about those items.


Bem (1974) developed the Bem Sex Role Inventory (BSRI) in such a way as to preclude "an inverse relationship between masculinity and femininity (p. 155)." The BSRI contains 20-item masculinity and femininity scales "selected on the basis of sex typed social desirability (Bem, 1975, p. 635)." With the BSRI, then, individuals can be classified as masculine, feminine, or androgynous. Androgyny is the preferred state since, according to Bem, an individual's behavior is limited if he or she can only perform behavior conforming to his or her perception of the masculine or feminine sex role.

While it was an original hope for this study to compare households with androgynous spouses with households with masculine husbands and feminine wives, the number of total androgynous households was too small (3) for such analysis. Surprisingly perhaps, less than half the households (28) were "traditional" in having masculine husbands and feminine wives.


These analyses follow the same pattern as Davis and Rigaux (1974), but differences in results as well as traditional-total population differences are noted.

A Framework for Classifying Decisions

Davis and Rigaux utilized a diagram, originally suggested by Wolf (1959), which displays all decisions simultaneously. The vertical axis represents the scale of relative influence between husband and wife and results from the average of scores. The horizontal axis represents the degree of role specialization operationalized in terms of proportion of respondents reporting the decision was joint. since the axes are not independent, the area of feasible coordinates is limited by a triangle (see Figure 1). While the definitions of the regions of the triangle as husband dominant and wife dominant are self explanatory--although wife specialization and husband specialization might be more appropriate terms--syncratic and autonomic decisions require further explanation. Most simply, syncratic decisions are those in which more than half the respondents reported the decision was joint, while, in the autonomic case, less than half the respondents indicated the decision was joint, but there was not consensus among respondents as to whether the decision was husband dominant or wife dominant. Thus, autonomic decisions, as well as husband and wife dominant decisions, represent role specialization.

Differences in Roles by Decision Phase

Table 1 includes a classification of the 20 decisions at each of the three phases in the decision process. Phase 1 is problem recognition, Phase 2 is search and evaluation, and Phase 3 is final decision. Percentages are shown so the results may be compared with those of Davis and Rigaux. The earlier results tend to be supported in that there was little difference between problem recognition and search for information, in terms of proportion in each influence pattern, and there was a marked increase in proportion of syncratic final decisions. However, these differences were not so pronounced as in the Davis and Rigaux study, perhaps because of much higher wife dominance throughout the decision process.



A large proportion of the decisions remained in the same influence pattern through all three phases of the decision process (70% in the present study, 64% in the Davis and Rigaux study). There were no differences between traditional couples and all couples among same pattern processes.

Similar to the Davis and Rigaux findings, husband dominance in the three phases was found for insurance on the husband's life (life insurance in Davis and Rigaux), homeowner's or renter's insurance (other insurance), and lawnmower (garden tools) as well as replacement tires for the primary family car which was not included by Davis and Rigaux. Seven decisions remained wife dominant in all stages, including five which support the Davis and Rigaux findings. The supportive, wife dominant decision patterns included replacement or additional pots and pans for the kitchen (kitchenware), beef roast (food and nonalcoholic beverages), slacks for the wife (wife's clothes), and household cleaning products (described the same in both studies). Also, tooth paste and adhesive bandages decisions were found to be wife dominant while the patterns across the three decision phases for the corresponding, more general items, cosmetics and toiletries and nonprescription drugs and first-aid items in the Davis and Rigaux study, were autonomic-wife dominant-autonomic and all autonomic respectively. Only neck tie for the husband was found to have an autonomic pattern over all three phases in the present study. While it might be argued the corresponding, husband's clothes pattern found by Davis and Rigaux was similar in that the third (decision) phase was only barely syncratic, at no stage, and for neither traditional nor total couples, did the percentage of joint decisions reach 15 for neck ties in the present study, while the percentage of joint decisions may have been as low as 30 only in the first (recognition) phase for husband's clothes in the earlier study. The syncratic pattern characterized decisions about movies (concerts, movies, theater) and family vacation (same) in all phases, thus supporting the earlier findings.

The remaining 6 decisions, which changed patterns across phases, were individually at least qualitatively different. In two cases, Davis and Rigaux results were supported. That is, the patterns for living or family room (other furnishings--rugs, drapes) had wife dominant recognition and search phases and autonomic decision phase, and the pattern for TV for living or family room (TV, Hi-Fi, tape recorder) had syncratic-autonomic-syncratic phases in both studies. Washing machine (household appliances excluding TV) decision processes were similar in both studies being autonomic and syncratic in phases 2 and 3 in both studies, but washing machine need recognition appeared to be wife dominant in the present sample compared to autonomic for the more general appliance category of the earlier study. The remaining 3 cases were quite different. While living room furniture and child(ren)'s toys for birthdays and holidays decision processes were characterized as syncratic at all phases in the Davis and Rigaux study, sofa for living room or family room and children's toys for birthdays were found to be wife dominant in both recognition and search in the present study. Further, in the present study, the sofa decision phase was found to be syncratic while the children's toys decision phase was found to be autonomic. Finally, car decisions, which went from autonomic to husband dominant to syncratic in the earlier study, were husband dominant, as primary family car, in the recognition as well as search phases in the present study. For the total present sample, the decision phase was syncratic for primary family car, corresponding to the earlier findings, but among traditional couples the decision phase was autonomic.

Davis and Rigaux pointed out, correctly, (p.55):

While an analysis based upon the pattern of influence gives a general impression of the change in marital roles throughout the decision process, the criterion is such that much information is lost. A decision's position could change rather dramatically along the dimension of relative influence as well as in terms of joint decision making and still remain within the same pattern of influence. A much smaller change along either dimension, on the other hand, can produce a shift in patterns if the decision happens to lie close to the boundaries.

The authors next discussed changes in relative influence and Joint decisions using averages over all item decisions for each phase. The averaging may have been misleading, however, because the averages were likely to have been biased by the sample of decision items selected for the study. Davis and Rigaux did effectively utilize the figures described earlier to graphically summarize changes between phases 1 and 2 and between phases 2 and 3. Figures 1 and 2 provide the same information for the present data on the total sample and Figures 3 and 4 provide this information for the 28 traditional couples.

The Davis and Rigaux findings of right-facing arrows, indicating greater specialization during phase 2, search for information, was not strongly supported since only half the cases in the present study had arrows pointing in the direction of greater role specialization (see Figure 1) compared to 80% in the earlier study. In two cases, replacement or additional pots and pans for the kitchen (kitchenware) and adhesive bandages (nonprescription drugs and first-aid items), nonspecialization movement between phases 1 and 2 found by Davis and Rigaux was supported. Interestingly, the present study supported nonspecialization for neck tie for the husband (husband's clothes), but the direction was toward specialization among traditional couples. While the direction with respect to children's shoes (children's clothes) were opposite in the two studies, the search phase was strongly wife dominant in both, which does support specialization.

In 5 cases, traditional couples differed from the total sample in terms of direction of the arrows in Figures 1 and 3. Traditional couples moved toward nonspecialization in search for drapes for the living room or family room, tooth paste, and insurance on the husband's life, while the total sample moved toward specialization. The opposite was found for beef roast and movies.

The Davis and Rigaux finding of movement toward Joint final decisions--left-facing arrows--was supported as shown in Figures 2 and 4.

Similarity of Roles Perceived by Husbands and Wives

In discussing the analytical advantages of the fact their data were collected from both spouses within the same family, Davis and Rigaux pointed out (p. 55):

. . . it will lead to insights about the more practical issue of which spouse to interview in family research studies . . . Previous research has shown few consistent differences between responses of husbands as a group compared with wives (Wilkening and Morrison, 1963; Davis, 1970; Granbois and Willett, 1970). On the other hand, when intrafamily comparisons of direct questions are made, considerable disagreement is often evident (Scanzoni, 1965; Davis, 1970). While the level of agreement is generally greater than chance, a common finding is that from 10 percent to 50 percent of couples disagree about the influence of one spouse relative to the other for any given decision. The size of this percentage seems to vary with the specificity of questions and, not surprisingly, with the number of response categories.





in which husbands perceived recognition to be syncratic while wives perceived recognition to be autonomic; and in the cases of phase 2 and phase 3 for primary family car, husbands perceived both search and final decision to be husband dominant while wives perceived these activities to be autonomic. Similarly, in only 4 cases among the traditional couples was there not aggregate agreement between husbands and wives. Among traditionalists, husbands perceived syncratic recognition of need for TV for living or family room while wives perceived recognition as autonomic (Syncratic agreement among all couples), husbands perceived autonomic recognition for children's toys for birthdays and holidays while wives perceived wife dominance (wife dominance agreement among all couples), husbands perceived wife dominant search for information on adhesive bandages while wives viewed this activity as autonomic (autonomic agreement among all couples), and husbands viewed the decision about children's toys as autonomic while wives viewed the final decision about children's toys as autonomic while wives viewed the final decision as wife dominant (autonomic agreement among all couples).



Table 2 provides information about intrafamily agreement in role perceptions. While inspecting average consensus, Davis and Rigaux said their results supported the previous research because, in averaging each decision item across the 3 decision process phases, only 3 of the 25 decisions were classified according to different influence patterns when comparing aggregate husband and wife responses. Without averaging over the 3 phases, similar results were found in the present study. In only 3 of the 60 cases in the present study was there not aggregate husband-wife agreement. Nonagreement was found in the case of family vacation in the first phase or agreement, among couples would not normally be meaningful since such summary statistics may be biased by the selection of decision items, averages were useful in this case because they provided a means of obtaining implications about item specifications and sample types.



The results paralleled those of Davis and Rigaux in that the percentage of all couples who agreed about their perceived role--consensus--was relatively stable from phase to phase (72, 73, and 68%). However, the average proportion of agreement was slightly, but consistently, higher in each phase than found by Davis and Rigaux (69, 69, and 66%), thus suggesting very slightly greater consensus for more specifically defined consumer decision items. In the aggregate sense, no real difference existed between the 28 traditional couples and the total sample of 60 couples, although the relatively sharper decline in average consensus found at the final decision phase (65% for traditionalists) may be noteworthy. In general, the previous finding that high consensus decisions were those in which relative influence was very skewed (e.g., children's shoes, household cleaning products) was supported.

In order to determine whether systematic bias in perception of roles existed among couples who did not agree, Davis and Rigaux defined two types of nonconsensus, resulting from modesty or vanity (p. 58):

modesty--either or both spouses overestimating the other's influence in a decision or understanding their own influence.

vanity--either or both spouses overestimating their own influence in a decision or understanding the other's influence.

Davis and Rigaux utilized a 5 percentage point differential between aggregate modesty and vanity as an operational criterion for systematic bias in nonconsensus. They found a majority of cases indicating vanity. Among the 19 decision items common to both studies, 6 of 9 nonconsensus items in the problem recognition phase were vanity biased, all 10 in the search for information phase, and 7 of 9 in the final decision phase.

The vanity bias results were not supported in the present study. At the problem recognition phase, there was no particular bias for either the total sample (5 modesty and 6 vanity) or the traditional couples (5 each). While the total sample was in line with the earlier results at the search phase (8 of 9 vanity biased), traditional couples were more evenly divided (4 modesty and 5 vanity). Finally, at the final decision phase, the total sample was marked by more systematic bias cases (13 to 9) than the previous study as well as having much greater modesty bias (11 of the 13 cases) than was previously found, Since no consistent bias difference pattern emerged through inspection of the 3 decision items described the same in both studies (household cleaning products, children's toys for birthdays and holidays, and family vacation), it can be suggested these between-study differences in systematic nonconsensus bias were a result of differences in item specification.

There were also some differences between traditional couples and all couples with respect to nonconsensus bias. At the recognition phase, systematic bias in the modesty direction among traditional couples was found for washing machine and primary family car when there was no systematic bias among all couples for these items. Again, while there was no systematic bias among all couples for children's toys for birthdays and holidays and, to a lesser extent, household cleaning products, there was a vanity bias among traditional couples. Among all couples, a modesty bias was found for movies while a vanity bias was found among traditional couples.

At the second, search for information, phase, traditional couples were modesty biased with respect to primary family car and neck tie for the husband, while all couples were vanity biased in the former case and not systematically biased in the latter. Traditional couples, on the other hand, were vanity biased with respect to washing machine and children's toys for birthdays and holidays, while there was no systematic non-consensus bias for the total sample for those decision items at this phase.

At the decision phase, the all couples group was found to be modesty biased with respect to sofa for living room or family room and replacement or additional pots and pans for the kitchen while the traditional couples exhibited no systematic nonconsensus bias. While there was no systematic nonconsensus bias for, the total sample with respect to lawnmower, traditional couples exhibited a modesty bias.



Product Classes and Systematic Nonconsensus Perceptual Bias of Roles

It may be possible to generalize about perceptual bias in marital roles in terms of product classes. The classes used were convenience goods, shopping goods, specialty goods, and preference goods; a scheme suggested by Holbrook and Howard (1977) as cited by Frankel (undated). The standard definitions were utilized for the first three goods classifications, with preference goods defined as "those for which there is a strong brand preference, but Low shopping effort is make (Frankel, p. 18)." It was recognized, of course, that the classification of any product is open to question since it would be a result of perceptions about how the consumer purchasing agents in question shop for those goods,

The convenience goods appeared to be beef roast and, somewhat surprisingly perhaps, adhesive bandages. These products were characterized by modesty bias at all phases. The preference goods appeared to be tooth paste, lawn-mower, children's shoes, and replacement tires for the primary family car. Preference goods appear to be characterized by vanity bias, either in all 3 phases or a number of combinations of 2 phases, but not modesty biased in the recognition or final decision phase. The shopping goods appeared to be sofa for living room or family room, replacement of additional pots and pans for the kitchen, primary family car, insurance on husband's life, homeowner's or renter's insurance, movies, and family vacation. All these items were characterized by a modesty bias in the decision phase. Children's toys for birthdays and holidays appeared to be specialty goods with no systematic bias in the first 2 phases and a vanity bias in the final decision phase. Not all items were classifiable in this manner. Household cleaning products, neck tie for the husband, slacks for the wife, and washing machine were not characterized by systematic bias at any stage. These items could well have been inadequately specified as it was expected that individual household cleaning products would be preference goods while the remaining 3 would be shopping or specialty goods.

Among the traditional couples, children's toys for birthdays and holidays appeared to be a preference rather than a specialty item, while washing machine appeared to be a shopping good, and there was no clear pattern for replacement or additional pots and pans for the kitchen.


As in the Davis and Rigaux (1974) study, these results are based on rather intensive analysis of a small data base. These results are useful, however, in that a non-convenience sample was utilized permitting generalization to a specifiable population.

The general movement toward role specialization in search for information about alternatives found by Davis and Rigaux was not supported, as such, in this study. A major reason was due to the fact that problem recognition specialization was more prominent in this study than was true previously. Specialization at the problem recognition phase, in turn, was likely to be a result of the specific nature of the decision items rather than a difference in sample populations or methods.

Role specialization obviously does exist in household decision processes and the implications discussed by Davis and Rigaux (pp. 59, 60) have clearly not been mitigated by these results. These results strongly support, however, clearly specifying decision items in research. In fact, item attributes (e.g., color of car) should be specified, although this information may be impossible to obtain for all decision process phases. Previous research (e.g., Davis, 1970; Hempel, 1972) has specified item attribute specialization, but not across decision process phases.

Davis and Rigaux held that one implication of high non-consensus in role perceptions was a need to word questions more specifically (p. 60) as was done in the present study. Interestingly enough, however, high item specificity appeared to have little effect on total consensus, 68% for the items common to both studies in the present study at the decision phase compared to 66% in the Davis and Rigaux study. Evidently communicators must take note of the degree and direction of nonconsensus rather than hoping to eliminate it through question wording. Nonetheless, the position taken throughout this analysis has been in favor of high specificity of items.

Decision makers should want to be knowledgeable about whether or not there is a vanity bias in perception of marital roles during the household decision process. It stands to reason that individual spouses with a vanity bias would get involved in the decision process regardless of role specialization, possibly in the phase not covered in these studies: evaluation of the final decision. These cases would thus require communication to both spouses.

Finally, while some differences were apparent between traditional couples and all couples in terms of role specialization at different phases of the decision process and level of role consensus, the number of these differences, while large enough to be significant, was also small enough to be, at least potentially, a result of random error. Thus, it would be difficult to make any generalizations with confidence. Nonetheless, the potential for differences appears established enough to be further studied in future research.




S. L. Bem, "The Measurement of Psychological Androgyny," Journal of Consulting and Clinical Psychology, 42 (April 1974), 155-62.

S. L. Bem, "Sex Role Adaptability: One Consequence of Psychological Androgyny," Journal of Personality and Social Psychology, 31 (April 1975), 634-43.

H. L. Davis, "Dimensions of Marital Roles in Consumer Decision-Making," Journal of Marketing Research, 7 (May 1970), 168-77.

H. L. Davis, "Measurement of Husband-Wife Influence in Consumer Purchase Decisions," Journal of Marketing Research, 8 (August 1971), 305-12.

H. L. Davis and B. P. Rigaux, "Perception of Marital Roles in Decision Processes," Journal of Consumer Research, 1 (June 1974), 51-62.

M. Frankel, A Summary Report: What Do We Know About Consumer Behavior?, Washington, D. C.: National Science Foundation, undated.

D. H. Granbois, and R. P. Willett, "Equivalence of Family Role Measures Based on Husband and Wife Data," Journal of Marriage and the Family, 32 (February 1970), 68-72.

D. J. Hempel, "A Cross-Cultural Analysis of Husband-Wife Roles in House Purchase Decisions," in M. Venkatesan, Proceedings, Association for Consumer Research, Chicago: ACR, 1972, 816-29.

M. L. Holbrook and J. A. Howard, "Frequently Purchased Goods and Services, "in R. Ferber, A Synthesis of Selected Aspects of Consumer Behavior, Washington, D. C.: National Science Foundation, 1977.

S. Sudman, Applied Sampling, (New York: Academic Press, 1976).

J. Scanzoni, "A Note on the Sufficiency of Wife Responses in Family Research," Pacific Sociological Review, 8 (Fall 1965), 109-15.

E. A. Wilkening and D. E. Morrison, "A Comparison of Husband and Wife Responses Concerning Who Makes Farm and Home Decisions," Marriage and Family Living, 25 (August 1963), 349-51.

D. M. Wolfe, "Power and Authority in the Family," in D. Cartwright (ed.), Studies in Social Power, (Ann Arbor: University of Michigan Press, 1959).