You are on page 1of 5

Husband-Wife Influence in Purchase

DecisionsA Confirmation and Extension


ROBERT E. WILKES*

This article provides two kinds of information about


the measurement of husband-wife influence in family
decision making. First, confirmatory support is given
to certain conclusions by Davis [4] in his study of
alternative measures of relative spousal influence.
Second, the article extends the application of multitrait-multimethod procedures to the stages or phases
of the decision process to determine the commonality
of perceived influence across spouses and how these
phases interrelate. Limited information about ethnic
variations in family decision behavior is also presented
since only black families were included in the study.

by each of one or more methods. In the study reported


here there were four traits (the four measures of
relative influence in Table 1) and two methods (separate responses from husbands and wives).
RESULTS OF THE STUDY
Four Measures of Influence
Table 2 is the multitrait-multimethod matrix for the
four measures of influence used in the study. Interpretation of this matrix involves the consideration of
convergent and discriminant validity. Convergent validity concerns intrafamily agreement between husbands and wives as to perceived relative influence
in decisions and is indicated when the parenthesized
values that form the validity diagonal are significantly
different from zero. At the .05 level only the decision
process value is significant; this is also the only value
in the validity diagonal that is meaningfully large in
a nonstatistical sense. And although the correlation

METHODOLOGY
Independent interviews with 60 middle-income black
husbands and wives yielded responses to each of the
4 measures of influence in Table 1.' The two global
measures were administered approximately one year
later than the other two measures in a follow-up study
to the same families. More is said shortly about the
decision process as conceptualized in the study. Responses were obtained for one of the following categories: (1) husband alone, (2) husband more than wife,
(3) husband and wife the same, (4) wife more than
husband, and (5) wife alone. These were weighted
as shown in the parentheses for computation purposes.
The data were analyzed using a multitrait-multimethod approach which provides information about
the validity of alternative measures of relative spousal
influence and of the similarity of husband-wife perceptions regarding this influence. First adapted to
family research by Davis [4] from Campbell and Fiske
[2], this approach yields a matrix of intercorrelations
computed from measuring each of one or more traits

Table 1
FOUR MEASURES OF INFLUENCE
Global measures^
1. When there is a disagreement between you and your wife,
who usually wins?
2. Who is the real boss in your family?
Blood and Wolfe IndexWho decides the following:
What job the husband should take.
What car to get.
Whether or not to buy life insurance.
Where to go on a vacation.
What house or apartment to take.
Whether or not the wife should go to work or quit work.
What doctor to have when someone is sick.
How much the family can afford to spend per week on food.

'The use of small samples is typical in family research and seems


acceptable here since no generalizations are made to black families
at large nor to different products.

Decision process
Who was responsible for initial problem recognition.
Who was responsible for acquiring information about the
purchase alternatives.
Who made the final decision as to which alternative should
be purchased.
Who made the actual purchase of the product.
"Source: [9].

*Robert E. Wilkes is Assistant Professor of Marketing, Virginia


Polytechnic Institute and State University.
224

Journal of Marketing Research


Vol. XII (May 1975), 224-7

HUSBAND-WIFE INFLUENCE IN PURCHASE DECISIONS

Table 2
MATRIX OF INTERCORRELATIONS'' FOR FOUR MEASURES
OF HUSBAND-WIFE INFLUENCE

Husbands ' responses


4
1
3
2
Husbands' responses
1. Global measure 1
2. Global measure 2
3. Blood and
Wolfe
4. Decision
process

Wives' responses
1 2
3
4

\
\

of values without regard to the coefficients for the


four traits. Using as a measure of association a
computed coefficient of concordance [8], this criterion was satisfied by the data (W= .50, x^ = 10.0, d.f.
= 5, p < . 10) at a somewhat high level of significance.
This is contrary to the finding by Davis who found
the pattern of trait interrelationship to be sensitive
to the spouse determining relative influence. It is not
clear from this study whether this difference reflects
some unique but unidentified factor in black families
and their decision behavior.

.4l\
\
.35\

-.09

.Ol\

.19

.50

Wives' responses
1. Global measure 1
2. Global measure 2
3. Blood and
Wolfe
4. Decision
process

225

.17

K . ( . 1 6 ) ^v l 3

.10 \
\

.3ff\ (.34) \ 1 2
\

-.16

.35\

.22\ .(31) \ 0 5
\

.42

.00

.19

-.20

'iX

.10\ ,(.74)

.02

.02 - . I 3 \

"Pearson product-moment correlation coefficients.

between husbands' and wives' answers in this study


for the Blood and Wolfe index [1] was greater than
that obtained by Davis (.31 compared to .15), its
doubtful validity as a measure of purchase influence
is confirmed. The study presented here also confirms
Davis's finding of low convergent validity for global
measures of spousal influence.
Discriminant validity requires the satisfaction of
three criteria. First, each value in the validity diagonal
should exceed the values in its row and column; that
is, the correlation between husbands' and wives'
answers to the same measure of influence should
exceed the correlation for two different measures.
Only the first global measure fails this test. The second
criterion for discriminant validity requires that the
correlation between husbands' and wives' responses
to the same measure of influence exceed that of their
separate answers to two different measures of influence. This would require that for each trait the validity
diagonal entry be greater than the correlations in the
upper left and lower right heterotrait-monomethod
triangles. The data in Table 2 again confirm Davis's
studyneither the global measures nor Blood and
Wolfe's index satisfy this requirement.
A final criterion for discriminant validity is that
the four heterotrait triangles show the same rank order

Application to a Specific Purchase Decision


The multitrait-multimethod approach may be extended to the decision process if the latter can be
acceptably conceptualized. This approach will reveal
the similarity of husband-wife perceptions about their
relative influence in the decision activity and will also
suggest how different phases of this activity interrelate.
The problem of conceptualizing the decision process
was treated in a recent article by Davis and Rigaux
[5]. The relevant literature, while not yet having
produced a definitive model, is gravitating toward a
modified version of Dewey's [6, p. 72] classic problem-solving formulation. Davis and Rigaux [5], for
example, use three stages: problem recognition,
search, and final decision. Engel, Kollat, and Blackwell [7] have a five-stage model that adds purchase
and postpurchase outcomes and combines alternative
evaluation and search. This authoritative source also
emphasizes the shortcomings of any conceptualization
of consumer decision activity, namely that consumers
may not go through or may not be aware of going
through these stages or may do so in varying time
dimensions. In the study presented here the following
phases were employed: problem recognition, search
for information, final decision, and purchase. Respondents to the study were able to identify their
influence in these stages with minimum apparent
difficulty. The model thus appears plausible and would
seem to have face validity as well as logical consistency
[3, p. 51-2]. No doubt continued research will produce
a generally acceptable conceptualization of this
process.
Because the research so far [7, pp. 59-62] indicates
that extended rather than limited decision behavior
is associated more with relatively expensive, infrequently purchased products, major household goods
were selected for the study. Each family interviewed
had recently purchased such a product. The study
objective, to reiterate, was to examine the pattern
of husband-wife influence rather than to generalize
across types of products. Interestingly enough, even
for families that had purchased kitchen or laundry
appliances, both spouses recorded considerable participation over the phases of the decision process.
That is, only in rare instances did one partner dominate

226

JOURNAL OF MARKETING RESEARCH

the decision process; rather, both spouses were involved across the process. (For actual distribution,
see [10].)
Table 3 presents the matrix of intercorrelations for
the phases of the decision process. Convergent validity
is indicated as all five values in the validity diagonal
are significantly different from zero at the .001 level.
In addition, all of the coefficients are sufficiently large
in a nonstatistical sense. This would indicate that
husbands and wives within families held common
perceptions about their relative influence for a given
phase of the decision process. These values would
be biased upward, of course, if one spouse simply
exerted no influence but left the purchase decision
completely to the other partner. As stated, however,
the pattern of husband-wife influence varied substantially across the different phases of the decision
process.
This finding is supported by the discriminant validity
present in the data as all three criteria for discriminant
validity are satisfied. First, all of the values in the
validity diagonal exceed the values in their respective
rows and columns. Second, the coefficients in the
validity diagonal exceed their counterpart values in
the heterotrait-monomethod triangles for every trait.
Finally, the pattern of the correlation rankings is
similar across the four heterotrait triangles (W= .70,
Table 3
MATRIX OF INTERCORRELATIONS'' FOR PHASES OF THE
DECISION PROCESS

Husbands responses
1
2
3
4
Husbands' responses
1. Problem
recognition
2. Information
search
3. Final decision
4. Purchase

Wives' responses
1 2
3
4

.09\
\
.36

.35\

.20

.14

Wives' responses
1. Problem
recognition
2. Information
search
3. Final decision
4. Purchase

(.59)\r.24
\

- .01

.00 \

.17

This limited study confirms Davis's conclusion that


global measures and the Blood and Wolfe index are
poor indicators of relative spousal influence in purchase decisions. Neither measure produces meaningful
correlations between husbands' and wives' perceptions of relative influence. Most importantly, such
measures simply fail to examine the relevant prepurchase activity of interest to marketers. Quick adoption
of sociologically-based measures by marketing researchers will likely prove to be a costly convenience.
The study also demonstrated that marketers must
investigate the entire decision process since relative
influence may vary from one decision phase to the
next. While one pattern of interrelationships was
evident in this study, quite another could emerge for
different families and different products. The point
is that husband-wife influence must be investigated
for each stage rather than falling prey to the assumption
of stable influence across the process based upon
intuition or limited inquiry into a single phase of the
process. Finally, the value of a multitrait-multimethod
approach has been further verified in this study. This
promising approach should begin to find wider application in marketing research.
REFERENCES

-.28\
\

.60)'^
\

.21

.01

\
13\(

CONCLUSIONS

. 0 3 \ (.79)\ .26
.42

X^ = 14.00, d.f. = 5, p<.02). Consequently, the


study results would not have been significantly different had only one spouse been interviewed. There
was low sensitivity to the selection of which partner
was selected for questioning about the pattern of
husband-wife influence in the decision process.
Inspection of the two heterotrait-monomethod triangles in Table 3 shows the phases of purchase and
final decision to be virtually unrelated according to
husbands (.09), with wives' answers producing a
slightly inverse relationship (-.24). Thus, perceived
influence in one of these two activities does not
guarantee similar influence in the other. Neither would
it be safe to say from observing the partner responsible
for the act of purchase that the information program
should be directed toward the same spouse only since
neither of the spouses perceive these two phases of
the decision process to be strongly related (.14 for
husbands, .01 for wives). The strongest association
between phases of the decision process according to
both husbands and wives occurs for the final decision
and the act of purchase (.53 for husbands, .58 for
wives).

.34

1. Blood, Robert O. and Donald M. Wolfe. Husbands and

. 3 7 \ ,(.68)

Wives; The Dynamics

-.02\
\

.39

.01

"Pearson product-moment correlation coefficients.

.5^

of Married

Living.

G l e n c o e , III.:

The Free Press, 1960.


2. Campbell, Donald T. and Donald W. Fiske. "Convergent
and Discriminant Validation by the Multitrait-Multimethod Matrix," Psychological Bulletin, 56 (March
1959), 81-105.
3. Cox, Keith K. and Ben M. Enis. The Marketing Research

HUSBAND-WIFE INFLUENCE IN PURCHASE DECISIONS

4,
5,
6,
7,

Process. Pacific Palisades, Calif,: Goodyear Publishing,


1972.
Davis, Harry L, "Measurement of Husband-Wife Influence in Consumer Purchase Decisions," Journal of
Marketing Research, 7 (August 1971), 305-12,
and Benny P, Rigaux. "Perceptions of Marital
Roles in Decision Processes," Journal of Consumer
Research, 1 (June 1974), 51-62,
Dewey, John, How We Think. New York: Heath, 1910,
Engel, James F,, David T. Kollat, and Roger D. Blackwell, Consumer Behavior, second edition. New York:

227

Holt, Rinehart and Winston, 1973,


8, Siegel, Sidney, Nonparametric Statistics for Behavioral
Sciences. New York: McGraw-Hill, 1956,
9. Turk, James L, and Norman W, Bell, "Measuring Power
in Families," Journal of Marriage and the Family, 34
(May 1972), 215-22,
10. Wilkes, Robert E, "A Study of the Role of Husbands
and Wives in the Decision Process of Selected Negro
Families Purchasing Major Household Durable Goods,"
unpublished doctoral dissertation. The University of
Alabama, 1971,

You might also like