Journal of Marital and Family Therapy

1984, Vol. 10, No. 3, 305-312

PREDICTING DIVORCE AT MARITAL THERAPY

INTAKE: WIVES DISTRESS AND THE MARITAL
STATUS INVENTORY*
D. Russell Crane
Brigham Young University

Neal Newfield Duane Armstrong

Texas Tech University Health Sciences Center
School of Medicine

The Marital Status Inventory (MSI), a measure of divorce potential was given
to couples in six independent samples. Scores from the five clinical sites and one
marital enrichment sample (N = 241 couples) were used to provide adequate
reliability, discriminant validity and predictive validity data. Compound prob-
ability for the five clinical sites supports the contention that, overall, wives are
more distressed than their husbands. The MSI was also able to identify couples
who later divorced. However, the Guttman properties of the MSI, previously
identified, were not replicated. Clinical implications of the higher wives scores
for prediction of divorce and marital therapy are discussed.

Marital and family therapists often find it of practical use to predict the divorce
potential of their clients. Knowledge that a client system is at risk for divorce may affect
the style, pacing and type of therapy used. For example, divorce adjustment may be the
therapy of choice as opposed to marital therapy. In response to the need for a means of
assessing thoughts and actions toward divorce, the Marital Status Inventory (MSI) was
developed (Weiss & Cerreto, 1980).
The MSI was thought to be a Guttman-type measure of progression along a contin-
uum, with no thoughts of divorce as one extreme and filing for or being divorced as the
other. Fourteen questions are presented in true-false format and inquire into specific
thoughts and behaviors believed to represent progressive steps toward divorce. Thus,
scores range from 0 to 14, with higher scores thought to be representative of greater
marital instability. Preliminary studies by Weiss and Cerreto (1980) and Crane and
Mead (1980) offer some initial validity, reliability, and normative data for the instru-
ment. Guttman analysis on the MSI was done with 134 married students and 24 couples
participating i n a marital therapy research project (Weiss & Cerreto, 1980). They found

*A preliminary version of this paper was presented at the American Association for Marriage
and Family Therapy Annual Meeting, Dallas, TX, October 1982.
D. Russell Crane, PhD, is a n Assistant Professor, Marital and Family Therapy, Department of
Family Sciences, Brigham Young University, Provo, UT 84602.
Neal Newfield, ACSW, CSW, is a n Adjunct Instructor, Marital and Family Therapy, Department
of Psychiatry, Texas Tech University Health Sciences Center School of Medicine, Lubbock, TX
79430.
Duane Armstrong, BA, is a Clinical Coordinator for the Department of Psychiatry, Texas Tech
Uiversity Health Sciences Center School of Medicine, Lubbock, TX 79430.

July 1984 JOURNAL OF MARITAL AND FAMILY THERAPY 305

a Coefficient of Scalability of .87 and a Coefficient of Reproducibility of .90. They,
therefore, argued that the scale was a unidimensional and cumulative measure.
The purpose of the present study was to evaluate the reliability, discriminant
validity, and predictive validity of the MSI using data from six independent data collec-
tion sites. Hopefully, the results of this study would support the use of the MSI for the
diagnosis of relationships, treatment selection, and research purposes. In addition, the
issue of differential distress between husbands and wives was addressed. The two earlier
studies found that marital therapy wives MSI scores were generally higher than their
husbands. However, this difference was statistically significant only for the Crane and
Mead (1980) sample.

METHOD

Subjects
Data from a total of six samples were available for analysis. Sample one was
composed of clients seen at the Marriage and Family Development Center (MFDC)
during 1980-1981. The MFDC is a practicum facility for students in Texas Techs
Marriage and Family Therapy training program. Couples in the MFDC sample ranged
in age from 20 to 41 (X = 31.6 husbands and X = 29.2 wives), had an average of 1.44
children (range 0-6)) and were primarily non-students.
The second sample included marital therapy cases from the Texas Tech University
Health Sciences Center, Department of Psychiatry (DOP) seen during 1980-1981. The
DOP sample was older, ranging in age from 21-50 (X = 34.6 for husbands and 32.7 for
wives). These couples had a n average of 2.36 children (range 0-6) and were also non-
students.
A third sample of couples participating in a church-sponsored marital enrichment
(ME) seminar in Dallas was available for study. Couples in the ME group ranged from
age 20 to 64 (X = 31, both husbands and wives), had an average of 2 children (range 0
to 5 ) , and were also non-students.
Data for the fourth sample was obtained from a group of Air Force (AF) couples
seeking family counseling from a Child and Adolescent Clinic in San Antonio, Texas.
Couples in the AF group ranged from age 22 to 50 (X = 30.1 husbands and TI = 28.5
wives). These couples had a n average of 2.2 children (range 1-5) and were active duty
military personnel or dependents. Data for these couples were partially reported in
Butler and Crane (1980).
Sample five was composed of couples requesting treatment from the Marriage and
Family Therapy Clinic at Brigham Young University (BYU) and sample six was com-
prised of couples seen a t the Psychology Clinic of the University of Oregon (U of 0).
Demographic information regarding the BYU and U of 0 samples may be found in
Crane and Mead (1980) and Weiss and Cerreto (19801, respectively.

Measures
The Marital Status Inventory discussed earlier was believed to be a Guttman-type
scale designed to evaluate a couples divorce potential.
The Locke-Wallace (1959) Marital Adjustment Test (MAT) is a widely used measure
of a couples marital adjustment. The MAT has been broadly used in research, especially
to distinguish distressed from non-distressed couples (e.g., Birchler & Webb, 1977;
Birchler, Weiss & Vincent, 1975; Williams, 1979).

Procedure
Practicum students at the MFDC and the staff of the DOP administered the MSI
and MAT as part of the couples intake process. The measures were completed in a

306 JOURNAL OF MARITAL AND FAMILY THERAPY July 1984

waiting room area prior to the intake interview; spouses were asked to complete the
measure independently and not to discuss their responses with one another.
In order to identify any couples who had divorced after terminating treatment, all
MFDC and DOP couples were followed-up by phone and by checking public records at
the local district court. The follow-up length was the same for both divorced and non-
divorced groups (% = 19 months, sd = 4.02 and X = 19.8months, sd = 4.92 respectively,
t = .32, df = 6 6 , p < .50).
Couples in the ME group were volunteers from a larger group who responded to a n
invitation to participate in a study of marital interaction. Couples who volunteered
were given a packet of measures for each individual and instructed to complete the
instruments in private. As a n inducement to participate, couples were offered a feed-
back interview to discuss the test results.
Couples in the AF group were tested as part of the intake in the USAF Medical
Centers Child and Adolescent Clinic. Couples in the BYU and U of 0 samples were also
tested during the intake interview.

RESULTS

Reliability
The reliability of the MSI was originally assessed in the Crane and Mead (1980)
study. They found a Spearman-Brown split-half reliability of .86 for the scale. In the
present study (combined MFDC and DOP samples), a Spearman-Brown split-half reli-
ability of .87 was obtained.

Discriminant Validity
The discriminant validity of the MSI can best be shown by examining its ability to
distinguish between types of therapy cases. Since the MSI is a measure of divorce
potential, couples requesting marital therapy should logically be more distressed (also
more likely to divorce) than couples requesting family counseling or marital enrichment.
This assumption was tested by comparing the available samples. Inspection of Table 1
shows that overall, the mean scores for marital therapy couples were higher than for
family therapy or marital enrichment couples, thereby supporting the discriminant
validity of the MSI.
An additional aspect of discriminant validity has to do with how similar the MSI
is to other measures of marital adjustment. In the present study, the MSI scores were

Table 1
MSI Scores for Marital Therapy, Family Therapy and
Marital Enrichment Couples
Type of Case Sample Husbands Wives
Site N Z sd N X sd t
Marital therapy BYU 30 2.70 (2.46) 30 4.30 (2.98) -2.86**
UofO 24 4.40 (2.80) 24 4.80 (2.10) 1.54
MFDC 41 4.51 (3.08) 44 6.64 (3.42) -3.00**
DOP 23 5.87 (3.92) 22 6.86 (3.63) - .88
Family therapy BYU 13 1.54 (2.30) 13 2.15 (2.64) 1.70*
UofO 32 1.80 (2.50) 32 2.20 (2.60) .32
AirForce 27 1.90 (2.41) 27 2.81 (3.43) 1.94*
MFDC 11 1.87 (1.64) 11 3.00 (2.65) 1.81*
Marital enrichment Dallas 40 1.38 (1.98) 40 1.38 (1.81) .02
*p < .05, one-tailed
**p < ,005, one-tailed

July 1984 JOURNAL OF MARITAL A N D FAMILY THERAPY 307

correlated with MAT scores for each spouse. Our r for husbands was - .49 ( p < .001)
and - .61 ( p < .001) for wives (combined MFDC and DOP samples). This result is in
contrast with the Weiss and Cerreto (1980) study which found non-significant rs of - .26
for both.

Predictive Validity
Also called empirical validity or criterion-related validity (Kerlinger, 1973), pre-
dictive validity refers to a scales ability to predict a likely outcome. Since the MSI is
designed as a divorce potential scale, its predictive validity may be evaluated by assess-
ing the degree to which MSI scores are predictive of a divorce or non-divorce outcome.
As a first step in estimating the MSIs predictive ability, couples were divided into
two groups: couples who were known to have divorced subsequent to therapy and couples
who had not. These groups (combined MFDC and DOP samples) were then compared
on their MAT and MSI scores. As shown in n b l e 2, the only difference between these
two groups was the wives MSI scores. Also, no difference was found between these
couples in age, education, or number of children.

Analysis by Sex
This analysis was conducted to determine if husbands and wives scores differed
on the MSI. Perusal of Table 1 reveals that for all couples seen in marital and family
therapy, wives MSI scores are consistently higher than husbands. The repetition of the
wives higher MSI scores across types of therapy and within each of the five independent
samples suggests that there was a n effect on the MSI scores due to sex. To test if wives
indeed scored higher in general than their husbands, a compound probability test was
conducted. This test is useful when one wishes to compare the results from several
independent samples each testing the same hypothesis. Compound probability for the
four marital therapy samples was calculated following the guidelines of Winer (1962,
pp. 43-44) for independent samples (using the formula for independent samples of N
less than 30). For marital cases a probability value ofp < .001 was obtained.
Compound probability for the four family therapy samples was also calculated. The
decision to conduct this test separate from the marital therapy samples was an attempt
to provide a conservative measure of their compound probability. It was thought that
the high probability values of the marital therapy samples might result in a statistically
significant result for the eight samples which would not occur if the family therapy
samples were examined alone. Analysis was done as described, resulting in a probability
of p < .01. The consistent results for both marital and family therapy samples argue
that wives MSI scores are consistently higher than their husbands.

Table 2
Marital Therapy Couples Who Did and Did Not Divorce
-
N X sd t df
MSI
Males Divorced 14 5.57 3.55
- .65 62
Nondivorced 53 4.87 3.59
Females Divorced 15 8.40 3.02
-2.11* 64
Nondivorced 55 6.27 3.57
MAT
Males Divorced 14 60.21 25.30
1.18 62
Nondivorced 54 69.87 27.64
Females Divorced 15 55.20 26.29
.96 64
Nondivorced 57 63.07 28.66
*p < .02, one-tailed

308 JOURNAL OF MARITAL AND FAMILY THERAPY July 1984

Analysis of MSI Items by Sex
In a n effort to clarify the cause of the wives' higher MSI scores, an item analysis
by sex was conducted (combined DOP and MFDC samples). It was expected that wives
would score higher than husbands on items emphasizing thinking and talking about
divorce since cultural norms would seem to support such actions. For example, Nye
(1976) found that married men were less likely than their wives to consult others on
intimate issues. Specifically,chi-square analysis for each question comparing husbands'
and wives' responses was conducted. Significant differences were found for MSI items
3 , 4 , 7 , 8 and 14 as presented in Table 3.
The differences between cells on item 3 was the higher number of wives having had
such thoughts. Item 4 was significant because of the higher number of wives having
made suggestions of divorce, separation, etc., to their husbands. Question 7 was signif-
icant because a higher number of wives reported talking to others about divorce con-
cerns. Items 8 and 14 were also significant indicating that thoughts of divorce were
more common or wives.

Scalogram Analysis
The ranking derived from the Guttman scalogram analysis is also presented in
Table 3. In addition, the scaling presented by Weiss and Cerreto (1980) is presented for
comparison. From Table 3 it can be seen that items 1, 3, 13 and 14 did not change
position. However, all other items changed from one to three positions. In examining
the Guttman-type characteristics of the MSI, our results were unable to replicate the
unidimensional and cumulative nature of the MSI (Weiss & Cerreto, 1980). Our Coef-
ficient of Scalability was .44, much lower than the .60 required of a true unidimensional

Table 3
MSI Items and Sealogram Rank Orders
Test Item Current Weiss &
Number Question Rank Cerreto Rank
1 I have not made any specific plans to discuss separation or 8 6
divorce with my spouse. I have not considered what I
would say, etc. (False)
2 I have set up an independent bank account in my name as 11 12
a measure of protecting my own interests. ("rue)
3" Thoughts of divorce occur to me very frequently, as often as 7 9
once a week or more. (True)
4" I have not suggested to my spouse that I wished to be 2 5
divorced, separated or rid of h i d h e r . (False)
5 I have thought specifically about divorce or separation; I 3 3
have considered who would get the kids, how things
would be divided, pros and cons of such actions, etc.
(True)
6 My spouse and I have separated. (True) 10 8
7" I have discussed the question of my divorce or separation 5 4
with someone other than my spouse (trusted friend,
psychologist, minister, etc.). (True)
8" I have occasionally thought of divorce or wished that we 1 1
were separated, usually after an argument or other
incident. (True)
9 I have not discussed the issue seriously or at length with 6 7
my spouse. (False)
10 I have filed for divorce or we are divorced. (True) 14 14

July 1984 J O U R N A L OF M A R I T A L AND FAMILY T H E R A P Y 309

 Table 3, continued

Test Item Current Weiss &

Number Question Rank Cerreto Rank
11 I have made no inquiries from nonprofessionals as to how 9 10
long it takes to get a divorce, grounds for divorce, costs
involved in such action, etc. (False)
12 I have not contacted a lawyer to make preliminary plans 13 13
for a divorce. (False)
13 I have not consulted a lawyer or other legal aid about the 12 11
matter. (False)
14" I have considered a divorce or separation a few times other 4 2
than during or shortly after a fight, although only in
vague terms.
"Items with male, female response differences
xz = 5.79, 5.78,5.32,13.39,7.07 respectively
-
df = l , p < .02
-

scale (Menzel, 1953). In addition, our Coefficient of Reproducibility was .83, somewhat
less than the minimum .90 required of a true Guttman scale (Edwards, 1957).

DISCUSSION

Overall, the results of this study suggest that the MSI is a reliable and valid measure
of divorce potential. In addition, the overall higher scores for wives were noted.
The reliability results for the MSI in this and the Crane and Mead (1980) study
argue for the internal consistency of the instrument. Both split-half reliability coeffi-
cients exceed the .70 minimum for internally consistent measurement (Bruning & Kintz,
1977).
The MSI also seems to be able to discriminate between different types of therapy
cases. The consistently higher MSI scores for marital therapy cases argue for the
usefulness of the MSI in identifying couples who would logically be more distressed and
probably more likely to divorce. This result is particularly convincing given the number
of independent data collection sites and their geographical diversity.
The discriminant validity of the MSI was also tested by correlating the MSI and
MAT scores for husbands and wives. The relatively weak correlation suggests that these
measures are not interchangeable and are probably measuring somewhat different
dimensions of marriage. It does seem, however, that the relationship between marital
dissatisfaction and steps towards divorce is stronger than originally believed. This may
be a function of higher levels of divorce potential in the present sample. However, the
degree of marital adjustment (MAT scores) was not different for couples who divorced
and those who did not. This result suggests that specific thoughts and actions taken
towards divorce are more important in eventual divorce than overall marital satisfac-
tion. One could easily be dissatisfied in marriage but not consider divorce as a n option.
In terms of predictive validity, the MSI appears to be able to identify couples with
a high risk for divorce. The couples who eventually divorced were characterized by high
wife MSI scores at the intake interview. The wives' mean for couples in the divorced
group was 8.4. This result supports Weiss and Cerreto's (1980) contention that a total
score of 8 points indicated that the couple had separated. In our sample, if the wife's
score reached 8 points, the couple was likely to divorce.

310 JOURNAL OF MARITAL AND FAMILY THERAPY July 1984

 These results may have implications for treatment selection. This is, if wives MSI
scores are 8 or more, marital therapy may not be the treatment of choice. Rather,
treatment related to divorce adjustment may be most appropriate. Also, treatment
procedures for couples with a high divorce potential need to be developed and tested
with these high risk cases.
Early work by Lewis, Beavers, Gossett and Phillips (1976) found that wives are
more distressed in relationships than husbands. Our results support this idea since
wives generally scored higher on the MSI than husbands. Consequently, early interven-
tion centered on changes in the husband may be most appropriate for such cases (iden-
tified by wife MSI scores of 8 or more). Such intervention may be necessary to keep the
couple in treatment if the wife is highly distressed.
The item analysis also adds some support to the idea that wives may be more
distressed than husbands. Results of the chi-square analysis yield the following general
conclusions: First, wives think about divorce more than their husbands (items 3, 8 and
14);second, wives are more active in making their dissatisfaction known (item 4);finally,
wives talk to others about their marriage problems (item 7). The overall higher scores
of wives on these cognitive and behavioral dimensions suggest that the divorce process
is more accurately reflected in the thoughts and actions of the wives.
Scalogram analysis with the present :ample failed to confirm the MSI as a unidi-
mensional and cumulative measure. Consequently, the total score is of greater value
than knowledge about a particular item. The failure to replicate the Guttman scaling
of the MSI may be due to a difference in samples. The Weiss and Cerreto (1980)sample
was married students and couples in a marital therapy research project. Our sample
was couples who requested marital or family therapy. These couples probably represent
a more highly distressed sample than those in the U of 0 study. Greater distress may
result in a less logical progression through the divorce process which was reflected in
our low Coefficient of Scalability.
Finally, the issue of the relative importance of individual rather than couple mean
scores was considered. The present samples r for husbands and wives (combined MFDC
and DOP samples) was .61, p < .01.Since only 37% of the variance in MSI scores is
shared between partners, the relationship component is not substantial. One person can
move further towards divorce more rapidly than the other. Given the importance of the
wives MSI score in divorce prediction, focusing on the highest score (especially the
wives) seems most helpful for clinical use. At present we are unable to speculate
regarding the importance of higher-than-average husband scores since the divorced and
nondivorced couples did not differ on this dimension. Further research is underway in
this regard.
The implications of the study related to couples who divorce after treatment should
be qualified by one important point. That is, that the therapists involved in the treatment
of these couples were primarily practicum students. Consequently, their proficiency as
therapists ranged from novice t o intermediate. Couples in treatment with more expe-
rienced therapists may have experienced a different outcome.

REFERENCES
Birchler, G. R. &Webb, L. T. Discriminating interaction behaviors in happy and unhappy marriages.
Journal of Consulting and Clinical Psychology, 1977,45,494-495.
Birchler, G . R., Weiss, R. L. & Vincent, R. P. Multimethod analysis of social reinforcement exchange
between maritally distressed and non-distressed spouse and stranger dyads. Journal of Per-
sonality and Social Psychology, 1975,31, 349-360.
Bruning, J. L. & Kintz, B. L. Computational handbook ofstatistics (2nd ed.). Glenview, IL: Scott,
Foresman and Company, 1977.

July 1984 JOURNAL. OF MARITAL AND FAMILY THERAPY 311

Butler, J. F. & Crane, D. R. Self-report schedules for use in assessing the marital adjustment of
abusive and nonabusive parents. The American Journal of Family Therapy, 1980,8,29-34.
Crane, D. R., & Mead, D. E. The marital status inventory: Some preliminary data on a n instrument
to measure marital dissolution potential. The American Journal of Family Therapy, 1980,8,
31-35.
Edwards, A. Techniques of attitude scale construction. Englewood Cliffs, NJ: Prentice-Hall, 1957.
Kerlinger, F. N. Foundations of behavioral research (2nd ed.). New York: Holt, Rinehart & Winston,
1973.
Lewis, J . M., Beavers, R. W., Gossett, J. T. & Phillips, V. A. N o single thread. New York: Brunneri
Mazel, 1976.
Locke, H. J. & Wallace, K. M. Short-term marital adjustment and prediction tests: Their reliability
and validity. Journal of Marriage and Family Living, 1959,21,251-255.
Menzel, H. A new coefficient for scalogram analysis. Public Opinion Quarterly, 1953,17,268-280.
Nye, F. I. Role structure and analysis ofthe family. Beverly Hills, CA: Sage, 1976.
Weiss, R. L. & Cerreto, M. The marital status inventory: Development of a measure of dissolution
potential. The American Journal of Family Therapy, 1980,8,80-86.
Williams, A. M. "he quantity and quality of marital interaction related to marital satisfaction: A
behavioral analysis. Journal of Applied Behavior Analysis, 1979,12, 665-678.
Winer, B. J. Statistical principles in experimental design. New York: McGraw-Hill, 1962.

American Association for

Marriage and Family Therapy
1984 Directory of Clinical Members
An excellent referral source for qualified marital and fam-
ily therapists throughout the United States and Canada.
The 1984 Directory of Clinical Members lists addresses
and telephone numbers in alphabetical and geographic
sections.
To order your copy of the AAMFT 1984 Directory of Clin-
ical Members send your check or money order for $20.00
to:

AAMFT, 1717 K St., N.W. #407, Washington, D.C. 20006

312 JOURNAL OF MARITAL A N D FAMILY THERAPY July 1984