Journal ofFamih Therapy ( 1990) 12: 45-5 7
The GRIMS. A psychometric instrument for the
assessment of marital discord
John Rust," Ian Bennun,? Michael Crowe: and
Research in marital therapy has been disadvantaged
short and recent psychometric questionnaire to objectively
a marriage for research, demographic and
Rust Inventory ofMarital State (GRIMS)
the Golombok Rust Inventory ofsexual Satisfaction (GRISS), and concen-
trates on aspects other than the sexual in
a dyadic relationship between two
dults living together. It isa 28 item psychometrically constructed inventory
designed to produce a single scale along which changes in a marriage may
develop as marital therapy progresses. It has been shown to be valid for this
purpose, and to have a good reliability.
by the lack of a good,
assess the state of
clinical purposes. The Golombok
is a companion questionnaire to
Many existing marital or relationship questionnaires tend to be rather
long and depend on particular theoretical perspectives. However within
marriage guidance as a wider community there is a mine of practical
experience which has generated its own values and expectations about
treatment and outcome, to some extent independently of any specific
theoretical stance. The Golombok Rust Inventory
(GRIMS) has been constructed to give a short and easy to administer
questionnaire to assess the state of a marriage and to identify the extent
of any discord. It may be used for research purposes in any marital or
relationship clinic. It is hoped that it will enable even small establish-
ments to develop research on the basis of their clinical activity.
Of the existing marital and relationship questionnaires, the
Locke-Wallace Marital Adjustment Test (Locke and Wallace,
was one ofthe earliest instruments developed and
in conceptualization. The Spouse Observation Check List (Weiss and
Margolin, 1977) comprises four hundred items and is therefore tedious
to complete. Other scales used in the assessment of marital satisfaction
_ _ _ _ __
Accepted version received December 1988.
* Department of Psychology, University
of London, Institute
Woburn Square, London WClH OAA, U.K.
t Department of Psychology, Exeter University, U.K.
Department of Psychiatry, Maudley Hospital, London, U.K.
of Marital State
is now somewhat dated
~ ~~ ~
of Education, 25
8 Department of Psychology, City University, London, U.K.
0163-4445/90/010045+ 13 $03.00/0
46 3. Rust et al.
include the Marital Communication Inventory (Bienvenu, 1978), The
Dyadic Adjustment Scale (Spanier, 1976), the Marital Satisfaction
Inventory (Snyder, 1982), the Areas of Change Questionnaire (Weiss,
Hops and Patterson, 1983) and the Marital Pre-Counselling Inventory
(Stuart and Stuart, 1972). Some
of these instruments were developed to
be consistent with a theoretical perspective ofmarital satisfaction and as
a restricted definition
of marital adjustment/satisfaction.
These scales, while receiving some subsequent validity and reliability
estimations, have not generally been psychometrically constructed and
consequently tend to contain a large number ofredundant items. A
used in the U.K. and Holland has been the Maudsley Marital Question-
naire (Crowe, 1978). Yet while this scale has been
and 20item versions at demonstrating reliability and
1983; Arrindell and Schaap, 1985) it still raises doubts concerning the
comprehensiveness and specificity of its original development.
The GRIMS is a questionnaire for the assessment ofthe overall quality
of a couple’s relationship. There are many differrnt ideas and theories
about the ideal marriage. While in the construction
deliberate attempt has been made to avoid any
preconceptions, ideas about
a good or bad relationship cannot exist in an
ideological vacuum. However, in spite
procedures appear again and again under different labels and theories,
and represent a folk psychological underpinning of marital treatment.
‘The test specification ofthe GRIMS is based on the common aspects of
the expertise of practitioners, counsellors and clients in the field. Many
of these people will, of course, have their own ideas
taken as a whole they will also have common observations and priorities
based on their often considerable experience.
bt’idesprrad consultations among a range
and clients have provided the test specification, which
practical situations encountrred by them. The test specification of the
GRIlMS is therefore a functional one. The domain of itcms in the
GRIMS particularly addresses those areas in which a marital therapist
would hope to see change during therapy.
this specification have been reduced, first conceptually and then psycho-
metrically, to yield a short efficient scale.
successful in both nine
of the GRIMS a
of these thcoretical
of this many ideas and treatment
and theories, but
of practising counsellors,
A large number ofitems under
To generate the test specification, both marital therapists and their
clients werr used in clarifying spccific areas ofdistrcss. Fifteen therapists
The GRIMS: the assessment of marital discord
were asked to identify (1) areas which they believed to be important in
marital harmony and
(2) the areas they would
interviews. Information from clients was obtained in the context of a
marital therapy study (Bennun, 1985) where 57 couples were specifically
requested to identify their targets for change. Targets reported included
the request that their partner alter specific behaviours or habits that were
felt to be irritating or unpleasant ( 19% :I. These included issues related to
punctuality and reliability, cleanliness, nagging,
the other’s interests), tidiness and the like. Three separate areas often
raised included communication problems
(1 1 .go/,) and household/domestic problems (1 1 .g%). Communication
problems included not just a lack of communication but also what was
said and how it was expressed. Couples who described communication
difficulties often experienced problems in communicating about
topics (e.g. sex, finance). Household/domestic problems generally con-
cerned the family home, responsibility for financial matters and sharing
tasks. The three next most frequent problems included difficulties caring
for each other and showing affection in a non-sexual way
problems related to child care (7.7%) and relationships with extended
family members (6.1 Yo). These two latter areas of conflict show that
often marital/relationship problems arise when
the stability of the dyad. Another problem area that
especially among young couples, was related to dependence-independ-
ence conflicts (5.0%). Moving from a life as a single person into a two-
person and later three person-relationship requires that each individual
maintain a boundary to some degree that enables them to have some
autonomy and a continuing capacity for individuation. Other problems
that were raised included sexual problems, jealousy, financial and work
related problems, violence, and one partner presenting with
psychological disorder that adversely affected the marriage. The views
of the clients and therapists were collated, reviewed and structured to
produce a two dimensional test specification. Axis one specified areas of
(i) interests shared (work, politics, friends etc.) and degree ofdependence
and independence, (ii) communication, verbal and non-verbal, (iii)
warmth, love and hostility, (iv) trust and respect, (v) roles expectations
and goals, (vi) decision-making, and (vii) coping with problems and
crises. Axis two specified areas in which the content
become manifest; (i) beliefs about, insight into, and understanding ofthe
nature of dyadic relationships, (ii) behaviour within the actual
relationship, (iii) attitudes and feelings about relationships, (iv)
motivation for change, understanding the
assess during initial
feeling excluded (by
(1 1.2%), decision-making
a third person influences
of axis one may
possibility of change, and
48 3. Rust et al.
commitment to a future together, (v) extent
of agreement within the
The pilot version
Between three and six items were constructed for each cell of the test
specification, generating 183 items which were reduced to
overlap was eliminated. Items were sought which could meaningfully
answered by either sex. Six lie items were also included. Items were
prepared as statements to which the respondent was asked to strongly
agree, agree, disagree or strongly disagree. The items were thus forced
choice (there was no ‘don’t know’ category), but
strength of feeling to affect scores where views were strongly held.
The pilot version of the test was administered to both partners in
120 individuals) from marital therapy and
marriage guidance clinics in the United Kingdom.
also allowed for
Preliminary item screening of the data eliminated all items which had a
difficulty value of greater than 0.80 or less than 0.20 for either men or
women. That is, items to which less than 20% either agreed or disagreed
(whether strongly or otherwise) were excluded. Further items
excluded where the number of non-responses exceeded
clients were strongly urged to answer all the items, there are invariably
some items within a pilot version which they find impossible to answer
due to non-applicability or other special reasons. The remaining 58 item
were factor analysed using principal factoring.
techniques on relatively small samples was justified on the grounds
the difficulty of obtaining subjects of this type, (b) the replication of the
analysis on different samples, (c) the use of parallel selection techniques,
items only being selected when they behaved in the same way in two
separate samples (this aspect in fact making the technique much more
powerful than that which would be obtained through merely increasing
the sample size itself), (d) confirmatory replication on subsequent
samples, giving a total number of 358 and (e) the superiority
nique over other item analysis procedures.
The pilot factor analysis was carried out for the men (n= 60) and
women (n=60) separately. Apart from some differentiating items the
overall results were similar. It had been hoped that, as with the
it would be possible to extract a main scale and a series of sub-scales
The use offactor analytic
of the tech-
The GRIMS: the assessment o f marital discord
which could provide
suggested a main factor which accounted
For men, four subsequent factors accounted for 7.2%, 4.6%, 3.1 YO and
2.6% of the variance respectively, with a similar pattern for women.
Factor identification by inspection of both the male and female items
identified factor one as a general marital discord factor with some
relation to a second factor which had
and lie items. The low communality of subsequent factors at this stage
was an indication of a unidimensional scale, however steps were taken to
at least attempt to identify a sub-scale structure as this can be informative
in certain circumstances. Re-analyses of the first four, five, six and seven
factor solutions with oblique rotations were carried out to identify signs
of stability. These analyses looked at men and women, combined and
separately. Factors other than the first proved to be unstable under
conditions, although factors which could be named e.g. jealousy, the
traditional marriage, outside relationships, sex problems, violence, etc.
did occasionally emerge. It seemed that, although these various elements
are important to the quality of a marriage, the interactions across and
between possible sub-categorizations are so strong that they cannot on
their own generate unidimensional scales. Directions of causality can
point in either direction, and non-linear relationships may
In spite of this, the large general factor indicates that from this pool of
interacting situations, feelings, motivations and coping strategies, there
does arise a strong and measureable single scale of marital quality. On
this basis a unidimensional scale for marital discord was targeted and no
attempt was made to build any sub-sc. <L 1 e structure.
Elimination of items with low communalities on first a five and then
a three factor solution reduced the 58 items to 42. These were further
reduced to six major criteria (a) balance
specification, (b) balance of positive and negative items to avoid
acquiescence, (c) balance ofitems to which more than
items to which less than 50% agreed, to reduce social desirability bias,
(d) similarity of item behaviour in men and women, (e) size of loading
on the general factor, and (9 low correlation with ‘lie’ items. Criteria (a)
ensured content validity and eliminated duplication, criteria (b), (c)
and (Q reduced bias, while criterion (e) achieved the aim ofclassical item
analysis in selecting items with good discrimination. Criteria
the development of an identical scale for men and women which
simplifies administration. The resultant scale has
in such a way that a high score represents a problematic relationship.
The factor structure of this scale was replicated on subsequent samples
a profile analysis. The factor structure clearly
for about 20% of the variance.
an increased emphasis on sexuality
across the original test
50% agreed, with
28 items and it is scored
50 J. Rust et al.
and found to be stable. For the male scale the
33% of the variance, and for the female scale 29%.
first factor accounted for
The 28-item GRIMS was standardized using two groups: (i) a sample
attenders at a General Practitioners clinic in central London [30 men
and 48 women (Rust, Golombok and Pickard, 1987)], and (ii)
presented as clients at marriage guidance clinics, sexual and marital
counselling clinics throughout England. The general practitioner group
was used as an approximation to general population data (Golombok,
Rust and Pickard, 1984).
The item scale correlations (with the scale adjusted for each item) for
both the pilot and the standardization studies are given in Table 1.
For the 60 couples (120 subjects) in the pilot sample the mean GRIMS
scoreformenwas37.12 (s.d.= 10.99),andforwomen41.54[s.d.= 11.26
(Rust et al., 1986)l. For the standardization clinical sample the male
GRIMS mean was 30.76 (s.d.= 13.03), and the female GRIMS mean
35.51 (s.d. = 10.37). These are both significantly lower than
pilot clinical sample. However there were some differences in sampling
between the two populations. In the pilot study the majority
subjects came from marriage guidance council and marital therapy
referrals, while in the standardization group there
number of couples referred for general relationship therapy, including
sexual problems. For the general practitioner group
GRIMS male score was 28.37 (s.d.
= 9.03) and the mean female GRIMS
score was 27.21 (s.d.= 10.02).
A closer investigation of particular items containing substantive
information showed that 19 (3 1 %) ofthe clinical couples considered they
were on the verge of separation, while only two (7%) of the GP group,
did so. Further, 42 of the men in the clinical group (51
dissatisfied with their relationship, while only two of the GP men were
dissatisfied (7%). The apparently anomalous group of 27 clinical men
who were very satisfied with their relationship (33%) accounted for the
relatively low GRIMS mean of the clinical group. It may be that
of these men were presenting primarily with sexual problems and were
making the point that the relationship
However only eight of the clinical women considered that they
satisfied with their relationship.
A further explanation
discrepancy may be that these men considered that it
partner who had a problem, and that they were attending the clinic to
those for the
was a much larger
(GP) the mean
itself (sex apart) was satisfactory.
was only their
The GRIMS: the assessment of marital discord 51
total GRIMS scale adjusted for that item for men and women in the pilot and
standardization clinical groups and in the General Practitioner
Item analysis o f the GRIMS. The correlations between each item and the
Pilot group Standardization group
52 J. Rust et al.
provide support and to aid the therapist. This interpretation
support from the literature (Lamb, 1986).
While these factors have mitigated against
difference between GP and clinical groups for the men, and reduced it
women, the raw GRIMS scores show much larger proportions ofpersons
with high scores in the clinical group. Thus,
(2 1 %) have a GRIMS score ofgreater than 40, compared with one (3%)
in the GP group. For women, the comparative figures were 26 (23%) in
the clinical group, and one (4%) in the GP group.
A combination of norm referencing and criterion referencing yielded
a transformed GRILMS scale which was able to give a good indication of
the existence and severity of any relationship problem. Transformation
was to pseudo-stanine scores (from 1 to 9 with higher scores indicating a
worse relationship) and the transformation table is given in the GRIMS
Handbook (Rust et al., 1988).
Differences between male and female raw scores were not sufficiently
large to justify separate transformations. These are therefore the same
men and women. A transformed score of5 (average) represents a normal
relationship just slightly worse than average in the population at large.
Scores of 9 (very severe problems) and 8 (severe problems) will only be
found where a relationship is in serious difficulty. Scale points 7 (bad)
and 6 (poor) represent mild and moderate relationship problems
respectively. Scores of 4 (above average), through 3 (good) to 2 (very
good) represent varying degrees of above average relationships.
Transformed GRIMS scores of 1 should be treated cautiously. The
respondents are either being untruthful or are at such a tender stage of
the relationship that prediction of its future course would be invalid on
the basis of their responses alone.
a significant mean
23 men in the clinical group
Split-halfreliabilities and Cronbach alpha coefficients were obtained for
men and women separately for the construction group and
standardization groups (see Table 2). The figures for the construction
group will however be upwardly biased as they are not independent of
In both standardization groups taken together the split-halfreliability
was 0.91 (n= 132) for men, and 0.87 (n= 128) for women.
The GRIMS: the assessment o f marital discord
Split-half reliabilities and Cronbach alpha for the subject groups
M F M
reliability 0.94 0.86
0.92 0.89 0.89 0.91 0.86 0.85
Content validity of the scale is high with respect to its specification, and
high face validity has been incorporated into the item selection.
Initial evidence of diagnostic validity was obtained within the pilot
group, where use was made of the fact that many couples
clinics were present primarily with sexual problems and with marriages
which were otherwise satisfactory. Therapists were asked to make this
diagnosis for the sample on the basis of their clinical interviews. Of the
60 couples, nine were diagnosed
as having a sexual rather than a marital
problem, and another
15 as having a strong sexual element
marital problem. For men these three groups had GRIMS means
40.87 for the marital problems group, 32.54
group, and 27.89 for the sexual problems only group.
variance gave a significance of 0.0028 for the difference between these
means. For women the
GRIMS means were 45.37, 39.23 and
respectively (P< 0.0003). As expec’ted, those with problems which were
predominantly sexual in nature had significantly lower
In the standardization study there
was no significant difference in ages
between the GP group (mean age
= 34.97 years, s.d. = 13.01, n = 29) and
the clinical group (mean age= 36.33 years, s.d. = 12.13, n = 24; with
some missing age data), the t-value being 0.40. For women in the
standardization groups, the mean GRIMS
(s.d. = 10.02, n = 48) in the GP group, and 35.5 1 (s.d. = 10.37, n = 80) in
the clinical group. The t-test for the difference between womens’
for the sexual complications
scores on the
scores were 27.21
54 J. Rust et al.
scores in these groups was significant at the 0.00001 level (t = 4.44). For
the men in the standardization study, group means
28.37 (s.d.=9.03, n=30) in the GP group, and
n = 84) in the clinical group. The difference between these means was not
(t= 0.93). However
qualification. Cochran’s C test for non-homogeneity of variance was
significant at the 0.01 level, and a Multivariate t-test between the 28
GRIMS items in the two groups was significant at the 0.0001 level. The
explanation almost certainly lies in the diverse nature
presenting for marital therapy. It must be remembered that
couples are not a random sample ofbad marriages, but a group ofcouples
who are sufficiently concerned about specific factors in their relation-
ships to seek help. This is demonstrated quite clearly when the men in the
marital therapy group are subdivided on the
for item 26: ‘I suspect we are on the brink of separation’, 19 men in the
clinical group (32%) agree to this item, while in the GP group only two
(7%) of the men agree.
A rating of the validity of the GRIMS as an estimator of change was
obtained from 24 clinical couples who
completed the GRIMS before therapy began and again after the fifth
session (or the final session if earlier). The therapists, who were
the GRIMS results, were asked to rate the couple on a five-point scale
ranging from ‘O-improved a great deal’ through ‘l-improved
moderately’, ‘2-slightly improved’, ‘Shot improved at all’
For the men, the GRIMS score changed from 50.29 (s.d.
before therapy to 36.67 (s.d. = 12.60) after therapy. The improvement
was statistically significant at the 0.001 level (t= 4.64). For the women
the initial GRIMS score was 52.52 (s.d. = 14.24) and following therapy
it was 39.93 (s.d.
= 10.09). This improvement
significant at the 0.001 level (t= 5.10). The average overall change
resulting from therapy for men and women together was 13.23 GRIMS
raw scale points.
Evidence of the validity of the GRIMS in assessing change was
obtained. The GRIMS scores for the male and female partner were
averaged for each couple. This average GRIMS
therapy was subtracted from the average GRIMS
therapy, or at the fifth session, to give a GRIMS score for change during
therapy (a large negative score representing a large improvement). ‘Ihis
GRIMS change score was correlated with the therapist’s ratings
change, giving a correlation coefficient of 0.77 (n = 24, P< 0.000 1 ). This
for the GRIMS were
30.76 (s.d.= 13.03,
this result needs considerable
basis of critical items. Thus
received marital therapy and
was also statistically
score from before
The GRIMS: the assessment o f marital discord
provides firm evidence for the validity of change in the GRIMS
an estimate of change in the quality
effectiveness of therapy.
While the scores of both the male and female partners are indicators
of the state of the relationship itself, differences between them would be
expected, particularly when one of the partners is doubtful of the
viability of relationship while the other is keen on its continuation. The
scores of both partners can be seen as representing some common and
some specific variance. The extent ofthe communality
correlation between the partners scores which, for the pilot study was
0.73 (representing a common variance of 50%). Within the standard-
ization study the correlation between male and female partner’s scores
was 0.53 (P< 0.000 1, n = 73). The size ofthis correlation is a good indica-
tion ofthe power ofthe test, and is ofsuch a size as to give some confidence
in predicting the state of a relationship from one of the partners alone.
of the of the relationship, or
is indicated by the
questionnaires is its simplicity ofadministration. The client has to answer
28 questions on one side of paper within a standardized format. This
makes it quick and uncomplicated for the client. The carbonized self-
scoring sheet enables the therapist to obtain the result within
The marital relationship itself is, of course, always particularly complex
and each couple’s problems will have a unique aspect, so that the
therapist when first seeing the clients will need to probe particular issues
and underlying dilemmas. The GRIMS can provide
for this venture by giving an objective and standardized
severity of the couple’s problem, while a quick preview
of the client’s respone to the individual GRIMS questions can identify
useful avenues to explore and save valuable clinical time.
The GRIMS is a companion test to the Golombok Rust Inventory of
Sexual Satisfaction [GRISS (Rust and Golombok, 1985a and 6, 1986)]
which is in use in sex therapy and research. For this reason the GRIMS
asks no questions directly about the sexual side
although it does include expression ofwarmth and affection. This lack
overlap means that with the GRISS and GRIMS together it
to pinpoint a relationship problem as either marital or sexual and to
adjust therapy accordingly (Rust
et al., 1988; Golombok and Rust,
The GRIMS is primarily applicable to married or unmarried hetero-
of the GRIMS over other marital or relationship
a useful framework
view of the
by the therapist
of the relationship,
56 J. Rust et al.
sexual couples who live together. It is further applicable to such couples
who are separated for work or similar reasons on a temporary
to couples who are temporarily separated
either one identifies the other as a primary partner. It may be used to
some degree within multiple relationships or homosexual relationships,
although no standardization data is available yet for these groups.
for other reasons so long as
properties of the Maudsley Marital Questionnaire
ratings in distressed and 'normal' volunteer couples based on the Dutch version.
Personality and Indiridual Uzffences, 4: 293-306.
Questionnaire (MMQ): a further step towards its validation.
lndi~idual Lhfferences, 4: 457--464.
C. (1985) l h e Maudsley Marital Questionnaire
( M M Q : an extension of its construct validity. British Journal oJP.ychiatv,
BENNUN, I. (1985) Prediction and responsiveness in behavioural marital therapy.
Behaoiourial Pvchotherapy, 13: 186-20 1.
Saluda. Family Life.
CKO\\.E, M. (1978) Conjoint marital therapy: a controlled outcome study. P,ychological
Medicine, 8 623-636.
GOl.OllBOK, S.. RUST, J. and PI(:KARD, C. (1984) Sexual problems encountered
general practice. British Journal qf'Sexua1 Medicine, 11: 17 1-1 75.
S. and RUST, J. (1988) Diagnosis of sexual dysfunction: Relationships
and the GRISS. Sexual andMarita1 Therapy,3( l): 25-31.
LAMB, M. ( 1986) The Father's Rok: ilpplied Per-spertiuu, New York. \Viley.
LOCKE, H. J. and \VAI.I.AC:E, K. M . (1957) Short marital adjustment and prediction
their reliability and validity. Marriage and Family Lioing, 21: 251 255.
RUST, J. and ~ O I . O M B O K , S. (1985,) The validation of the Colombok Rust Inventory
of Sexual Satisfaction, British Journal OfClinical I'.ycholoLU, 24: 63-64.
RuSI, J. and GOLOMBOK,
S. (19856) l h e GRISS: A psychometric instrument for the
assessment ofsexual dysfunction. Archioec o f Sexual BeharGmr, 15(2): 153-161.
RusI, J. and GOLOhlBUti, S. (1986) The Go[ombok Ru.tt Znr'entorq' ofSexual Sati.~faclion,
RL'ST, J., BENNLIN,
validation of the Golombok Rust Inventory of Marital State, Sexual and Marilal
Rr-Sl, J., GOLOXIBOK,
S. and PICKARI),
C. (1 987) Llarital problems in General Practice,
Sexual and Marital Therapy, 2(2): 127-1 30.
RUST, J., BENNUN,
M . and GOLOMBOK,
o f Marital State (CRIMS) (Te.tl and Handbook), NFER-Nelson. t2:indsor.
Rusl', J., GOLOMBOK,
S. and COLLIER,
.J. (1988) Xlarital problems and sexual
dysfunction: how are they related?
British Journal ofPychintv, 152: 629-63 1.
M'. A., BOEIXNS, LV. and LmwRr, H. (1983) On the psychometric
evaluations of self
W . A., EhluEI.ti.AXII', P. 31. G . and BAsr, S. (1983) The Maudsley Marital
W . A . and S(:HAAIJ,
J. (1978) A Councellor's Guide to Accompany a Marital Communication Inzrenlory,
and GOLOXIR~E;, S. (1986) The construction and
S. (1988) The Golornbok Rust Inoentorq'
The GRIMS: the assessment of marital discord 57 Download full-text
and psychometric approaches. In: C. D. Speilberger and J. N. Butcher (Eds),
Advances in Personality Assessment. Volume I , Hillsdale, NJ. Lawrence Erlbaum.
G. B. (1976) The measurement of marital quality. Journal o f Sex and Marital
Therapy, 5 : 288-300.
SILTART, R. and STU.~RT,
F. [ 1972) The Marital Pre-Counselling Inventroy, Champaign, Ill.
W~rss, R. and MARCOLIN, G. (1977) Marital conflict and accord. In: A. Ciminero, K.
Calhoun and H. Adams (Eds), Handbook o f Behavioural Assessment, lVew York. Wiley.
WEISS, R., Hors, H. and PATTERSON, G. (1983) A framework for conceptualizing
marital conflict: A technology for altering it, some
Hammerlynrk, L. Hardy and E. March (Eds), Behaoiour Change: ConceplsandPractice,
Champaign, Ill. Research Press.
D. K. (1982) Advances in marital assessment: behavioural, communications
data for evaluating it. In: L.