Citation
A psychometric study of the family adaptability and cohesion evaluation scales

Material Information

Title:
A psychometric study of the family adaptability and cohesion evaluation scales
Creator:
Alexander, Barrie Byrnes, 1953- ( Dissertant )
Johnson, Suzanne ( Thesis advisor )
Siegel, Lawrence ( Reviewer )
Johnson, James ( Reviewer )
Amatea, Ellen S. ( Reviewer )
Hall, Everette ( Reviewer )
Gutekunst, Richard ( Degree grantor )
Place of Publication:
Gainesville, Fla.
Publisher:
University of Florida
Publication Date:
Copyright Date:
1982
Language:
English
Physical Description:
vii, 95 leaves : ill. ; 28 cm

Subjects

Subjects / Keywords:
Child psychology ( jstor )
Factor analysis ( jstor )
Family members ( jstor )
Mothers ( jstor )
Parents ( jstor )
Philosophical psychology ( jstor )
Psychometrics ( jstor )
Questionnaires ( jstor )
Social desirability ( jstor )
Statistical discrepancies ( jstor )
Dissertations, Academic -- Clinical Psychology -- UF
Family psychotherapy -- Testing
Genre:
bibliography ( marcgt )
non-fiction ( marcgt )

Notes

Abstract:
The Family Adaptability and Cohesion Evaluation Scales, a measure of family functioning derived from the theory of structural family therapy, contains three scales: cohesion, adaptability and social desirability. These were slightly modified, based on data from a pilot study, in order to make them easier to read. The modified questionnaire (FACES-R) was then administered to members of 206 non-clinic families and 42 families being seen in a mental health clinic. The psychometric properties of the three scales were then examined. Sample characteristics, scale reliability and validity, and the normative properties of the scales were discussed. Comparisons with other studies were done where data were available. Support for the scales was found in high alpha coeffi- cients of reliability and the similarity of the present sample means and standard deviations to those of two other samples. Intercorrelations between family members' scores also moderately supported the reliability of the cohesion but not the adaptability scale. Factor analyses of the scales did not replicate that done by either Olson et al. or Portner but did fit with the Circumplex Model since each scale factored with one bipolar factor. The validity of the scales was not supported. The adaptability and cohesion scales correlated significantly with each other and cohesion correlated significantly with social desirability. Therapists' ratings also did not correlate with the scale scores. An analysis of covariance controlling for SES found no effect for SES and no differences between groups on cohesion or adaptability. There were group differences for social desirability. While chi'square analyses of group differences lend minimal support for the cohesion and social desirability scales' ability to discriminate between the clinic and non-clinic groups, factor analyses suggest that these differences are due only to social desirability. Attempts to improve the scales were made with results consistently indicating that the only dimension being tapped by the FACES-R scales is social desirability. Implications of these findings were discussed in light of the Circumplex Model.
Thesis:
Thesis (Ph.D.)--University of Florida, 1982.
Bibliography:
Includes bibliographic references (leaves 912-94).
General Note:
Vita.
General Note:
Typescript.

Record Information

Source Institution:
University of Florida
Holding Location:
University of Florida
Rights Management:
Copyright Barrie Byrnes Alexander. Permission granted to the University of Florida to digitize, archive and distribute this item for non-profit research and educational purposes. Any reuse of this item in excess of fair use or other copyright exemptions requires permission of the copyright holder.
Resource Identifier:
09291431 ( oclc )
ABU5932 ( ltuf )
0028556710 ( ALEPH )

Downloads

This item has the following downloads:

UF00102805_00001.pdf

psychometricstud00alex_0025.txt

psychometricstud00alex_0082.txt

psychometricstud00alex_0089.txt

psychometricstud00alex_0074.txt

psychometricstud00alex_0058.txt

psychometricstud00alex_0079.txt

psychometricstud00alex_0067.txt

psychometricstud00alex_0084.txt

psychometricstud00alex_0002.txt

psychometricstud00alex_0075.txt

psychometricstud00alex_0092.txt

psychometricstud00alex_0053.txt

psychometricstud00alex_0021.txt

psychometricstud00alex_0042.txt

psychometricstud00alex_0061.txt

psychometricstud00alex_0009.txt

psychometricstud00alex_0012.txt

psychometricstud00alex_0100.txt

psychometricstud00alex_0052.txt

psychometricstud00alex_0086.txt

psychometricstud00alex_0033.txt

psychometricstud00alex_0096.txt

psychometricstud00alex_0014.txt

psychometricstud00alex_0010.txt

psychometricstud00alex_0073.txt

psychometricstud00alex_0024.txt

psychometricstud00alex_0023.txt

psychometricstud00alex_0028.txt

psychometricstud00alex_0057.txt

psychometricstud00alex_0007.txt

psychometricstud00alex_0081.txt

psychometricstud00alex_0013.txt

psychometricstud00alex_0018.txt

psychometricstud00alex_0011.txt

psychometricstud00alex_0085.txt

psychometricstud00alex_0005.txt

psychometricstud00alex_0056.txt

psychometricstud00alex_0087.txt

psychometricstud00alex_0045.txt

psychometricstud00alex_0038.txt

psychometricstud00alex_0015.txt

psychometricstud00alex_0034.txt

psychometricstud00alex_0037.txt

psychometricstud00alex_0072.txt

psychometricstud00alex_0055.txt

psychometricstud00alex_0102.txt

psychometricstud00alex_0090.txt

psychometricstud00alex_0022.txt

psychometricstud00alex_0069.txt

psychometricstud00alex_0078.txt

psychometricstud00alex_0039.txt

psychometricstud00alex_0091.txt

psychometricstud00alex_0093.txt

psychometricstud00alex_0047.txt

psychometricstud00alex_0006.txt

psychometricstud00alex_0062.txt

psychometricstud00alex_0101.txt

psychometricstud00alex_0041.txt

psychometricstud00alex_0032.txt

psychometricstud00alex_0019.txt

psychometricstud00alex_0020.txt

psychometricstud00alex_0031.txt

psychometricstud00alex_0001.txt

psychometricstud00alex_0070.txt

psychometricstud00alex_0003.txt

psychometricstud00alex_0103.txt

psychometricstud00alex_0049.txt

psychometricstud00alex_0088.txt

psychometricstud00alex_0016.txt

psychometricstud00alex_0044.txt

psychometricstud00alex_0027.txt

psychometricstud00alex_0068.txt

psychometricstud00alex_0065.txt

psychometricstud00alex_0094.txt

psychometricstud00alex_0064.txt

psychometricstud00alex_0030.txt

psychometricstud00alex_0059.txt

psychometricstud00alex_0040.txt

psychometricstud00alex_0104.txt

psychometricstud00alex_0004.txt

psychometricstud00alex_0076.txt

psychometricstud00alex_0071.txt

psychometricstud00alex_0046.txt

psychometricstud00alex_0054.txt

psychometricstud00alex_0083.txt

psychometricstud00alex_0098.txt

psychometricstud00alex_0051.txt

psychometricstud00alex_0080.txt

psychometricstud00alex_0063.txt

psychometricstud00alex_0048.txt

psychometricstud00alex_0026.txt

psychometricstud00alex_0036.txt

psychometricstud00alex_0060.txt

psychometricstud00alex_0043.txt

psychometricstud00alex_0017.txt

psychometricstud00alex_0008.txt

psychometricstud00alex_0050.txt

psychometricstud00alex_0095.txt

psychometricstud00alex_0097.txt

psychometricstud00alex_0099.txt

psychometricstud00alex_0077.txt

psychometricstud00alex_0066.txt

psychometricstud00alex_0035.txt

psychometricstud00alex_0029.txt

psychometricstud00alex_0000.txt


Full Text














A PSYCHOMETRIC STUDY OF THE FAMILY ADAPTABILITY
AND COHESION EVALUATION SCALES














BY

BARRIE BYRNES ALEXANDER


A DISSERTATION PRESENTED TO THE GRADUATE COUNCIL
OF THE UNIVERSITY OF FLORIDA
IN PARTIAL FULFILLMENT OF THE REQUIREMENTS
FOR THE DEGREE OF DOCTOR OF PHILOSOPHY





UNIVERSITY OF FLORIDA


1982
















ACKNOWLEDGEMENTS


Ths list of individuals who helped to one degree or

another with this project numbers near 1,000. Each person who took the time to fill out a long questionnaire for me will never get a personal thank you so this must suffice.

Then there are the hard workers in the lab who know that the work and frustration they experienced cannot be adequately repaid with this simple thank you. Marika Spevack did an excellent job of helping to organize and motivate us all.

Similarly, Nan Hucker and her colleagues at the University Hospital's Community Mental Health Center in Jacksonville were wonderful about collecting data on the f amilies seen in therapy there and Abe Spevack helpfully transported it back to Gainesville.

The psychology interns at Shands Teaching Hospital also took time from their busy routines to ask their patients to help. The staff of P.K. Yonge Lab School also were invaluable.

Randy Carter spent many long hours helping me interpret factor analyses as well as initially helping me to conceptualize the project.











Ellen Amatea pointed me in the direction of helpful reading material.

My husband Scott kept his head when I was losing mine and provided me with lots of love and stability.

The American Association of University Women gave me a fellowship which made this year much easier and gave me confidence in the project too.

Last but most, Dr. Suzanne Johnson was with me every step of the way. I've learned most of what I know about research from her and I don't think I could have had a better teacher or friend.


iii
















TABLE OF CONTENTS


Page


ACKNOWLEDGEMENTS . ABSTRACT . . . . . . . . . . . . . . . . . . . . .


CHAPTER


I MODELS OF THE FAMILY.
Introduction. . . .
The Structural Model.
The Circumplex Model .

II FAMILY MEASUREMENT . .
Family Functioning Index.
Family Environment Scale.
The Family Adaptability and
Evaluation Scales . .

III PILOT STUDY .
Method. . . .
Results .

IV FACES-R STUDY .
Method. . . .
Procedure .
Measures. . .

V RESULTS OF FACES-R STUDY
Sample Characteristics.
Reliability of the scales
Independence of the Scales
Norms .
Validity. . .


Cohesion


VI ATTEMPTS TO IMPROVE THE FACES-R SCALES
Item Analysis and Further Factor
Analysis. . VII DISCUSSION. .












Page

APPENDIX

A PILOT STUDY REVISED ITEMS . . . . . . . . 66

B FACES-R QUESTIONNAIRE . . . . . . . . . . 73

C AUXILIARY FORMS . . . . . . . . . . . . . 83

D ITEM NUMBERS AND FACTOR LOADINGS . . . . 87


REFERENCE NOTES . . . . . . . . . . . . . . . . . 91

REFERENCES . . . . . . . . . . . . . . . . . . . 92

BIOGRAPHICAL SKETCH . . . . . . . . . . . . . . . 95















Abstract of Dissertation Presented to the Graduate Council of the University of Florida in Partial Fulfillment of the
Requirements for the Degree Doctor of Philosophy


A PSYCHOMETRIC STUDY OF THE FAMILY ADAPTABILITY
AND COHESION EVALUATION SCALES By

Barrie Byrnes Alexander

August, 1982


Chairwoman: Suzanne Johnson, Ph.D. Major Department: Clinical Psychology


The Family Adaptability and Cohesion Evaluation Scales, a measure of family functioning derived from the theory of structural family therapy, contains three scales: cohesion, adaptability and social desirability. These were slightly modified, based on data from a pilot study, in order to make them easier to read. The modified questionnaire (FACES-R) was then administered to members of 206 non-clinic families and 42 families being seen in a mental health clinic.

The psychometric properties of the three scales were then examined. Sample characteristics, scale reliability and validity, and the normative properties of the scales were discussed. Comparisons with other studies were done where data were available.

Support for the scales was found in high alpha coefficients of reliability and the similarity of the present











sample means and standard deviations to those of two other samples. Intercorrelations between family members' scores also moderately supported the reliability of the cohesion but not the adaptability scale. Factor analyses of the scales did not replicate that done by either Olson et al. or Portner but did fit with the Circumplex Model since each scale factored with one bipolar factor. The validity of the scales was not supported.

The adaptability and cohesion scales correlated significantly with each other and cohesion correlated significantly with social desirability. Therapists' ratings also did not correlate with the scale scores. An analysis of covariance controlling for SES found no effect for SES and no di4fferences between groups on cohesion or adaptability. There were group differences for social desirability. While chi'square analyses of group differences lend minimal support for the cohesion and social desirability scales' ability to discriminate between the clinic and non-clinic groups, factor analyses suggest that these differences are due only to social desirability.

Attempts to improve the scales were made with results

consistently indicating that the only dimension being tapped by the FACES-R scales is social desirability. Implications of these findings were discussed in light of the Circumplex Model.


vii















CHAPTER I
MODELS OF THE FAMILY



Introduction


In the past, the primary focus of investigators interested in assessing psychopathology was the individual and his personality "traits" which were thought of as invariant. The advent of behaviorism, however, made psychologists aware of the impact of a person's environment and the possibility of interaction effects between the individual and his environment. Thus, there has been a shift in attention from the individual to the larger interpersonal context in which the person functions. This has resulted in a large number of clinicians and researchers focusing on the family as being or primary importance. Most family investigators feel that there are family patterns and processes crucial to understanding the development of pathology within individual family members. It is the belief of a number of the theorists, researchers and therapists that the family is a basic unit of psychopathology. This new focus has resulted in a sizeable body of literature concerned with family functioning which seeks to identify the family "types" comprising healthy and unhealthy systems. Unfortunately, most of the research with couples and families has not tested hypotheses that are












considered therapeutically relevant (Olson, Russell, & Sprenkle, 1980). Additionally, most of the outcome studies of family therapy have not been theoretically based (Gurman & Kniskern, 1978). Integration of research, theory and practice has rarely been attempted. Reviewers of family interaction research (Jacob, 1975; Riskin & Faunce, 1972) and of family therapy outcome studies (DeWitt, 1978; Masten, 1979; Wells & Dezen, 1978; Wells, Dilkes, & Trivelli, 1972) agree, however, that methodological problems, including the lack of reliable and valid measurement, have prevented both researchers and family therapists from adequately assessing both their hypotheses and treatment efforts. This lack of adequate methodology has had a part in preventing the integration of theory and research to bring about advances in therapeutic techniques. Therapists do not engage in objective diagnostic evaluations because there is a scarcity of techniques that have been adequately tested or validated that also yield useful information (Cromwell, Olson, & Fournier, 1976).

The following is a brief discussion of two models of the family that have influenced the work of family therapists in recent years. The Structural Model (Minuchin, Rosman, & Baker, 1973), and the Circuniplex Model (Olson, Sprenkle, & Russell, 1979) have both evolved from a systems perspective influenced a great deal by the work of Von Bertalanffy (1968).












The Structural Model


This model of family intervention has been most completely developed by Salvador Minuchin and his colleagues at the Philadelphia Child Guidance Clinic. Here the focus of intervention is the structural patterns that are maladaptive for the system. Structure in healthy families is thought to be hierarchical yet flexible, with the parents allied at the top of the hierarchy. Analysis of the family structure consists of examining the patterns of alliances or coalitions between members and their "rules" governing boundaries. For example, there are four transactional patterns felt to be characteristic of a family process that encourages somatization (Minuchin et al., 1978). These are enmeshment, overprotectiveness, rigidity and lack of conflict resolution. Enmeshment refers to a condition where subsystem boundaries (e.g., the spouse relationship) are poorly differentiated, weak and easily crossed by other family members. Overprotectiveness is a high degree of concern for each other's welfare to the extent that nurturant acts are constantly elicited and provided. Rigid families are those in which very little change in transactional patterns occurs despite the ineffectiveness of existing ones. Lack of conflict resolution does not mean that conflict does not exist but rather that the family avoids or diffuses conflict, often at the expense of the symptomatic individual.










Minuchin et al. (1978) have done some work to test this model with diabetics and anorectics. They used two behavioral measures, a Diagnostic Interview and a Family Task, with various target behaviors operationally defined. Unfortunately, these measurement devices have not been published and it is unclear whether there is any support for their reliability or validity. It is thus impossible to evaluate objectively Minuchin's claim that the Structural Model is supported by research findings.



The Circumplex Model


This model, developed by Olson et al. (1979, 1980), hypothesizes that cohesion and adaptability are two independent dimensions underlying family functioning which emerge through an "inductive conceptual clustering" from a survey of the family literature across all disciplines (Olson et al., 1979). Family cohesion is defined by two components: the emotional bonding members have with one another and the degree of individual autonomy a person experiences in the family system. Enmeshment and disengagement (used as in Minuchin's model) are seen as two extremes of cohesion with a balance of cohesion being the most conducive to effective family functioning. Table 1 shows the large number of family therapists who have found cohesion or similar dimensions to be relevant to their work.







Table 1
Family Cohesion Dimension: Related Theoretical


Concepts


Extremely Low Cohesion Balanced Cohesion Extremely High Cohesion


Bowen


Emotional divorce


Hans and Handel Kantor and Lehr Litz
Minuchin

Olson






Reiss

Rosenblatt Scott and Askworth Steirlin


Vogel and Bell Wynne


Differentiated self


Separateness


Schism, Skew Rigid boundaries disengagement Low connectedness Independence Rigid boundaries Weak coalitions Interpersonal-distance sensitive Apartness



Expelling Centrifugal force

Scapegoating Pseudo-hostility


Clear boundaries

Separate connectedness
Interdependence Clear individual boundaries Marital coalition Environment-sensitive





Centripetal and Centrifugal balance


Mutuality


Undifferentiated family ego mass; emotional fusion
Connectedness Bounding


Diffuse boundaries Enmeshment High connectedness Dependence Diffuse boundaries Parent-child coalition Consensus-sensitive

Togetherness Extra-ordinary mutual involvement Binding
Centripetal force


Pseudo-mutuality Rubber fence


Source: Olson et al., 1979, p. 7.










Family adaptability is "the ability of a marital/family system to change its power structure, role relationships and relationship rules in response to situational and developmental stress" (Olson et al., Note 3, p. 2). Rigid families (again the concepts are similar to Minuchin's) are seen as one extreme while chaotic families comprise the other. Healthy families are thought to show a balance between change and stability.

Adaptability and cohesion have been conceptually linked in the Circumplex Model of Figure 1. From this, 16 possible family types emerge with the four center ones representing healthy families and the perimeter, unhealthy.

After this model was developed, it was discovered that

Angell (1936) had developed similar dimensions he called family integration and family adaptability. Dividing families into high, medium and low categories on these two dimensions, he formed nine types of families, or which he located eight in his work from case records (Angell, 1936).

The theoretical validity of the Circumplex Model is supported by a large number of studies that find cohesion and adaptability to be important concepts related to family functioning (Olson et al., 1980). Research testing the model directly has not been extensive, but there are a few studies cited by Olson et al. (1980) as supporting the model's validity (Russell, 1979; Sprenkle & Olson, 1978).

These can each be criticized on methodological grounds,

















Low COHESION High


Separated Connected Enmeshed


Chaoticall
s g
Disenqaq
Chaotically
Separated Sep rated ha t-cally






Flexibly lexibly
11 ar
. rrDisengaqedSeparated


cparated1

Structurally Disengaged



Rigidly Rigidly Separated isengaged


Modified from
Source: Olson et al., 1979, p. 17


tructur all




SEnmeshed Rigidbly Feil Connected i


F igure 1
Sixteen Possible Types of Marital and Family Systems
Derived from the Circumplex Model


Disengaged


L












particularly measurement and data analysis issues and must be viewed as preliminary attempts to verify the model.

In summary, these two systems models presently have much influence on the work of practicing family therapists. Students are being trained in family therapy based on these models, yet there has been little attention to developing measures adequate for evaluating either model. The tendency has been to attempt to "prove" the model through the use of woefully inadequate measurement.















CHAPTER II
FAMILY MEASUREMENT



Along with the shift in attention from the individual

to his larger interpersonal context came the problem of assessing him in that context. Attempts have ranged from using tests designed for individual diagnosis (e.g., Thematic Apperception Test, TAT) with each family member and synthesizing the information into a "family TAT," to tests, both behavioral and self-report, designed specifically for assessing family interactions. Cromwell, Olson and Fournier (1976) compiled a catalogue of the various instruments that have been used in diagnosing families. Within the self-report measures a number of relationship tests have been developed. These assess marital relationships, family and marital problems and a variety of parent-child interactions. The advantage of this type of test is its potential of yielding insight quickly and objectively into the interpersonal behavior between family members. Three tests have been selected for discussion here, because more data have accumulated concerning their reliability, validity and general utility in work with families than with most family instruments. The Family Functioning Index (FFI) and the Family Environment Scale (FES) are included to give the reader a flavor of the range of concepts that researchers have attempted to measure and










the limitations encountered. The third instrument, the Family Adaptability and Cohesion Evaluation Scales (FACES), is the one studied in the present report.



Family Functioning Index


Pless and Satterwhite (1973) developed the Family Functioning Index for use by medical doctors to index families requiring further attention. It is not meant to precisely measure the family's problems; by definition it is a global measure. The original 16 questions were chosen based on both theoretical and empirical considerations. Mothers of 300 school age children (part of a 1% random sample of all families in one county in New York) completed the questionnaire. Two-hundred nine of these children had chronic physical disorders; the remainder were healthy.

Factor analysis of the parents' answers yielded five

principal components: communication, togetherness, closeness, decision making and child orientation. These components were never used to modify the instrument into subscales to measure the individual concepts, probably due to the small number of items. Examining the factors, however, one can see that togetherness, closeness and decision making might be related to the models discussed earlier, particularly the dimension of "cohesion."

Reliability of the instrument is suggested by a .72 correlation between the independently obtained scores of











husbands and wives. Families seeking counseling rated as low runctioning by case workers showed more agreement between spouses' scores (r=.74) than those ranked high functioning (r=.41) (Pless & Satterwhite, 1973). Good five year test retest reliability has been reported for the total score (Satterwhite et al., 1976).

validity of the index was assessed by comparing scores

of new registrants at a community counseling agency with ratings of social workers. Mothers' (n=43) FF1 scores correlated .48 (p<.Ol) with the social workers' ratings, while fathers (n=39) correlated .35 (p<.Ol) (Pless & Satterwhite, 1973). The index has differentiated families seeking assistance from a counseling agency from those seeking no assistance (Pless & Satterwhite, 1973). Pless, Roghxnan and Haggerty (1972) found children from low functioning families were described as having more behavior problems and lower self-esteem than those from high functioning families. Korsch et al. (1978) reported that aspects of the FFE were predictive of non-compliant behavior among children with renal transplants.

These studies suggest that this 16 item questionnaire

may be useful in yielding a total score of family functioning. Its length makes it a useful screening device, but its global nature limits its utility for the therapist interested in specific areas of deficit.












Family Environment Scale


Rudolf Moos developed the Family Environment Scale to assess the social climates of all types of families (Moos, 1974). The items were constructed so that they were worded in a manner congruent with Moos' theory of Environmental Press. Each item had to identify characteristics of an environment which would exert a press toward cohesion, achievement or moral-religious emphasis. The focus of the measure is in three areas: interpersonal relationships among family members, the directions of personal growth emphasized in the family and the basic organizational structure of the family. Table 2 lists the dimension to which each of the ten subscales belongs along with their defining qualities. As can be seen from Table 2, the dimensions tapped by the FES, particularly scales four through ten, are descriptors rather than interaction measures. Moos has used the FES in preliminary attempts to obtain family profiles similar to MMPI profiles (Moos & Moos, 1976).

The 90 true-false statements which comprise the FES were selected and grouped into the ten subscales using the following criteria: (1) The overall item split should be as close to 50-50 as possible to avoid items characteristic of only extreme families. (2) Items should correlate more highly with their own than any other subscale. (3) Each of the subscales should have an approximately equal number of items scored true and false to control for acquiescence














Relationship Dimensions
The extent to which family members are concerned and committed to the family and the degree to which family members are helpful and supportive of each other. The extent to which family members are allowed and encouraged to act openly and to express their feelings directly. The extent to which the open expression of anger and aggression and generally conflictual interactions are characteristic of the family.

Personal Growth Dimensions
The extent to which family members are encouraged to be assertive, self-sufficient, to make their own decisions and to think things out for themselves. The extent to which different types of activities (i.e., school and work) are cast into an achievement oriented or competitive framework. The extent to which the family is concerned about political, social, intellectual and cultural activities. The extent to which the family participates actively in various kinds of recreational and sporting activities. The extent to which the family actively discusses and emphasizes ethical and religious issues and values.

System Maintenance Dimensions
measures how important order and organization is in the family in terms of structuring the family activities, financial planning, and explicitness and clarity in regard to family rules and responsibilities.


Table 2
Family Environment Scale Subscale Descriptions


1. Cohesion



2. Expressiveness


3. Conflict





4. Independence



5. Achievement
Orientation


6. IntellectualCultural
Orientation
7. Active
Recreational Orientation
8. MoralReligious
Emphasis


9. Organization


10. Control Assesses the extent to which the family
is organized in a hierarchical manner, the rigidity of family rules and procedures and the extent to which family members order each other around.
Source: Moos, R. and Moos, B., 1976, p. 360.










response set. (4) The final subscales should show only low to moderate intercorrelations. (5) Each item and subscale had to maximally discriminate among families (moos, 1976).

The FES was given to 285 families (1,000 individuals) of various sizes primarily from upper and middle income levels. Unfortunately, the data employed for item selection and subscale construction were obtained from this same sample so that the subscale dimensions were never independently confirmed. Despite the factor analytic rationale for item selection given above, Moos never factor analyzed his data (Robertson & Hyde, Note 1).

The ten subscales have been found to show adequate internal consistency (.64 to 179) and good eight week test-retest reliability (.68 to .86) and have an average intercorrelation of .20, indicating they measure distinct though somewhat related aspects of the family environment (Moos, n.d.).

A comparison between 42 clinic families and 42 "normal" families matched for size and composition showed the clinic families obtaining significantly lower scores on cohesion, intellectual-cultural orientation and active-recreational orientation. This provides some support for the construct validity of the FES (Moos, 1974). However, Janes and Hesselbrock (Note 2) found that four of the ten scales correlated significantly with the Personal Reaction Inventory measure of social desirability.










Robertson and Hyde (Note 1) performed an orthogonal factor analysis on responses from half their sample of 686 freshmen and sophomore high school students. Their data indicated that Moos' subscales did not adequately represent the dimensions of the FES. They found an eight factor solution the best but one factor was uninterpretable. Their reformulated scales are 1. Group Cohesion (13 items, overlaps Moos' cohesion scale but not exactly), 2. Conflict (7 items), 3. Activities (8 items), 4. Structure (7 items), 5. Protestant Ethic (13 items), 6. Religion (5 items), 7. Verbal-Intellectual orientation (10 items). Using these seven scales, they replicated using the rest of their sample and found that Group Cohesion, Conflict, Structure and Religion replicated very well while the other three replicated moderately well. These seven subscales were found to be internally consistent (.48-.67) and have relatively low intercorrelations, the highest being Conflict with Group Cohesion (r=-. 35).

Examining these factor analyzed subscales, one can again see the suggestion of a relationship between some of these subscales (e.g., cohesion, structure, conflict) and the constructs of the two models discussed earlier. If the factor analyses of these scales replicate with an adult population, a typology of families similar to that which Moos and Moos (1976) have attempted, could be pursued. Since the instrument is oriented toward describing the family environment











more than the interactions between family members, it is not of great utility to all family therapists. It is no surprise that it has not been extensively used by therapists.



The Family Adaptability and Cohesion Evaluation Scales


In response to the need for a method to measure the two dimensions, family cohesion and adaptability, postulated by their Circumplex Model, Olson et al. (Note 3) developed a self-report instrument, FACES. Unlike the FFI and the FES, FACES was designed as a tool for use by family therapists with the goal of diagnosing family target behaviors to be altered through therapy.

Initial items were composed to measure nine concepts posited by the model to be related to cohesion (emotional bonding, independence, family boundaries, coalitions, time, space, friends, decision making, interests, recreation) and seven thought to be relevant to adaptability (assertiveness, control, discipline, negotiation, roles, rules, system feedback). These 203 face valid items were rated by 35 marriage and family counselors for low to high cohesion and adaptability. Then 410 students in a college level family relationships course took the questionnaire. This sample, however, cannot be considered representative of the population for which the test is designed and the subjects' experience in their course indicates probably sophistication with respect to the subject matter.










These student responses were then factor analyzed but the two scales, cohesion and adaptability, were analyzed separately. Thus, the independence of these two dimensions has not been empirically tested. Partner (1982, p. 58) refers to the independence of the dimensions being confirmed by Candyce Russell (1979) in "two major efforts to test this assumption" but again, the factor analyses she refers to used 30 and 20 subjects respectively, an insufficient number for the large number of items.

From Olson and others' (Note 3) factor analysis of the

adaptability items, there emerge two factors accounting for 78.6% of the variance. Most of the chaotic and rigid items (the extremes of adaptability) loaded on factor one while the moderate items loaded on the second factor. Rather than supporting Olson and others' (Note 3) belief that "chaotic" and " rigid" families represent maladaptive extremes of an "adaptability" dimension, this two-factor solution implies a twodimensional structure for this variable. It also implies that the chaotic and rigid items which loaded on factor one are independent of the moderate items (factor two). This does not make sense in terms of the described model, but Olson et al. (Note 3) did not explain this discrepancy and indeed, described their results as if a discrepancy did not exist.

In factor analyzing the cohesion items the first three

factors accounted for 33.7%, 14.1% and 11.1% of the variance.












The next factor accounted for only 4.6% of the variance, yet Olson et al. (Note 3) opted for the four factor solution selecting their "disengaged" items from factor one, "enmeshed" from factor two and "moderate" from factors three and four. Again, their analysis did not support a unidimensional conceptualization of this variable's structure.

Additionally, Olson et al. (Note 3) did not use the results of their factor analyses to score their selected items using the factor scoring coefficients. They chose instead to use a weighted scoring system in which items at the chaotic extreme (for adaptability) are weighted or multiplied by three, the moderate items are weighted by two, and the other extreme (rigid items in this case) are weighted by one. This has the disadvantage of not incorporating the information yielded in their factor analyses. In selecting items to include in the final scales, their criteria necessitated six items for each subscale: two for each of the high, moderate and low levels of the concept. On the cohesion scale, in particular, this process resulted in items being retained that correlated as low as .11 with the factor. Thus, items were not selected so much with regard to their empirical derivation via factor loadings as by their face validity and theoretical fit with the Circumplex Model. It is entirely probable that some items were retained that loaded very low on a factor while others with higher loadings were discarded because they did not fit the rigid criteria mentioned above.










Included in the final 111 items were 15 items adapted for families from Edmonds' (1967) Marital Conventionalization Scale. This scale measures the "extent to which a person distorts his appraisal of a phenomenon in the direction of social desirability" (Edmonds, 1967, p. 682). Conventionalization, or social desirability, as it is more commonly called, is a form of self-deception and is not an intentional distortion of responses. Edmonds (1967) discusses support for his hypothesis that this tendency is a great problem in measuring marital interaction because social approval and consequent ego-involvement are greater in matters pertaining to one's marriage than in any other area. His results indicated a .63 correlation of his scale with the Locke-Wallace short scale of marital adjustment. Since matters of family functioning may be hypothesized to be equally sensitive to such tendencies, Olson et al. (note 3) are to be commended for including these items. Unfortunately, all items on the scale are answered on a four point scale from "4 = true all the time" to "l = true none of the time." This occasionally puts the respondent in the position of having to respond to a "never" question with a one to four answer, as in the following example of social desirability item number 38: 38. 1 have never regretted being with my family not even
for a moment.
1 2 3 4
Never True Sometimes True Usually True Always True










Joyce Portner, a graduate student working with Olson at the time, was the first to administer the questionnaire to several members of the same family (Portner, 1982). She gave FACES to 53 families in counseling at Youth Service Bureaus, 31 families with runaway adolescent and 117 nonproblem families. From each family, both parents and an adolescent child completed the questionnaire for a total of 603 individuals or 201 families. Despite the large number of problem families, this sample was used as a normative sample to establish cutting points for assigning families to categories based on their scores. Factor analyses (again of the two scales separately) were also done with these data but it is unclear whether she used an average family score for each item or used the total sample of 603 individuals as if they were independent. If the latter was the case, the assumptions of the factor analytic procedure were violated. No mention is made of how the social desirability items loaded in the analyses. Her factor analysis of the cohesion items resulted in three factors with factor one containing the "disengaged" items, factor two the "moderate" items, and factor three the "enmeshed" items. The alpha reliabilities for each factor were .52, .77 and .62, respectively.

For adaptability, the analysis also resulted in three factors with the "chaotic" items loading on factor one, the "moderate" items on factor two, and factor three containing










the "rigid" items. The alpha reliabilities of the factors were .81, .85 and .69 respectively.

The alpha internal consistency for the whole cohesion

scale was .83 with a Pearson's Split half of r=.49 (p .001). The Adaptability items had an alpha of .73 with a split half reliability of r=.42 (p<.001).

The subscales' (e.g., emotional bonding, roles, etc.) measures of internal consistency and split half reliability were low and Olson et al. (Note 3), therefore, recommended that they not be used for research purposes.

Partner (1982) also correlated the two dimensions with the social desirability items. While the correlation was insignificant for the adaptability items, it was moderately correlated with cohesion (r=.45, p<.001). Thus, it appears that individuals have a tendency to present their families in a socially favorable light when responding to the items on the cohesion scale. Partner (1982) suggests that this problem may be handled by employing a partial correlation technique but then does not do this in her subsequent data analysis.

Using the t-test statistic, the mean scores for fathers, mothers, adolescents and total family of the clinic and nonclinic groups were tested for group differences by Portner (1982). The non-clinic families scored significantly higher on cohesion than did the clinic families and this difference replicated for fathers and mothers but not for adolescents.










There were no significant differences for any group on adaptability. For social desirability the groups were significantly different with fathers, mothers and adolescents of the clinic group scoring lower than their non-clinic counterparts. Partner (1982) interpreted this as reflecting a willingness by clinic families to speak honestly and openly about their feelings.

Partner (1982) attempted to place her families in a 16 cell table similar to the Circumplex Model of Olson et al. (Note 3) in Figure 1 (see page 7). Dividing the families

into low, moderately low, moderately high, and high groups on each dimension, she used chi-square analyses. While the majority of both the clinic and non-clinic groups fell in the moderate area on cohesion, more clinic families fell in the low (disengaged) area and more non-clinic families fell in the enmeshed area (p<.0001). This was true for the average family score, and also for fathers' scores, mothers' scores and adolescents' scores.

On the adaptability scale, the majority of both groups again fell in the moderate range, but more clinic families were in the chaotic area than were non-clinic families (p<.04). This was true for average family score, fathers' scores, mothers' scores and adolescents' scores.

On both scales combined, the majority of all families

(45% clinic and 35% non-clinic) were in the moderate ranges. While a greater proportion of the clinic families fell outside the moderate ranges, the trend was non-significant.










Bilbro and Dreyer (1981) attempted a replication (minus

six items relating to coalitions) of Olson and others' (Note 3) factor analysis of the cohesion items. Their samples consisted of 48 husbands, their 48 wives, 86 female college students and a fourth sample made by standardizing scores within each subsample and combining them. The sample of husbands and wives were not independent samples, however, since they were from the same families. The size of each sample also was not adequate to factor analyze 48 items. They excluded the social desirability items from their analyses and concluded that the best solution for the cohesion scale was a one-factor solution. In computing alpha reliabilities of the cohesion scale, they found them to be low (wives 4=.58, husbands 4=.51, college females 4=.48, total 4=.65). They found the reliabilities of a scale made up only from their first factor to be much higher (r=.81).

Bilbro and Dreyer (1981) also criticized Olson and others' (Note 3) confusing construction of FACES in which it is not clear whether they view cohesion as a unidimensional or a multidimensional construct. The ratings of cohesion items by therapists on a scale from one to nine and the adding of scores of all items from each factor in final scoring suggest a unidimensional view. However, the listing of nine concepts related to cohesion and the identifying of four cohesion factors from their factor analysis suggests a multidimensional view. There are problems with a multidimensional











view of this construct in that it would allow families to be both enmeshed and disengaged at the same time and implies that the two are not related to each other.

It should be noted here that Olson and his colleagues are now working on a shortened version of FACES. They have made a 40 item version called FACES II and a 10 item version called FACES III which they are administering to 1,000 families.* These new versions will not include a social desirability scale.**

In summary, although of the three instruments discussed, FACES shows the most potential utility for family therapists interested in pinpointing areas of deficit in the relationships among family members, its method of construction and the manner in which the data analyses were conducted leave many unanswered questions about its psychometric properties: Are adaptability and cohesion really two independent dimensions? Is there support for the reliability of the instrument in agreement between the scores of various family members? Can the factors Olson et al. (Note 3) derived be replicated? Does the instrument successfully discriminate between problem and non-problem families? Are the cutting points presently established correct when checked with a normative family sample?



*Olson, D. Personal communication, February, 1982. "Portner, J. Personal communication, October, 1981.










The present research is a critical look at the Family Adaptability and Cohesion Evaluation Scales. Included in this examination are three aspects. First, a pilot study was undertaken to improve some of the items while leaving the questionnaire substantially intact. Second, an attempt

was made to replicate Olson cmd others' (Note 3) factor analyses and Portner's (1982) measures of reliability using families. Finally, the results of the factor analyses were used to compose empirically derived subscales from the original items and the validity issues relating to both the original and derived scales were explored.















CHAPTER III
PILOT STUDY



Method


A preliminary investigation of the 111-item FACES scales was done by administering the scale to 40 undergraduates in an introductory Psychology class. The same group was given a revised version of the scale with the items rewritten so that they would be easier to read. Using Fry's (1968) method for calculating the readability level, Olson's original FACES had a late sixth grade reading level while the revision had a late fourth grade reading level. The average completion time for each instrument was approximately 15-20 minutes.

If the three subscales (cohesion, adaptability and social desirability) are theoretically independent, each item thought to measure cohesion, for example, should correlate more with the cohesion scales than with either of the other two scales and the three scale scores should have low intercorrelations. The scales should also show good internal consistency. The extent to which this was true for the pilot sample was examined.










Results


The alpha coefficients of internal consistency, shown in Table 3, were adequate for the three original scales. However, the items did not consistently show a higher correlation with their own scale than with the other two scales. Seventy-nine percent of the cohesion items correlated more with either the adaptability scale or the social desirability scale than with cohesion. Sixty-seven percent of the adaptability items correlated more with one of the other scales than with their own. Only on the social desirability scale were all the items most highly correlated with their own scale.



Table 3
Alpha Coefficients of Internal Consistency


Original FACES Final Revised Version Cohesion .88 .85

Adaptability .95 .93

Social Desirability .71 .67




If the scales are independent, one would not expect the scale scores for cohesion and adaptability to be correlated with each other. Table 4 shows the intercorrelations for this preliminary sample. Olson and others' (Note 3) original










Table 4
Pilot Study Scale Correlations


Original Original Original Revised Revised Revised
Cohesion Adaptability Social Cohesion Adaptability Social
Desirability Desirability

Original .90*
Cohesion p<.01

Original .43* .88*
Adaptability p<.01 p<.01

Original Social .27 .28 .97*
Desirability p<.09 p<.08 p<.01

Revised
Cohesion

Revised .22
Adaptability p<.17

Revised Social .27 .24
Desirability p<.10 p<.14


*Statistically significant











cohesion scales correlated significantly with their original adaptability scale. The original social desirability scale did not correlate with the other two scales. The same analyses were undertaken for the rewritten items. Items for the final version of FACES to be used with a large sample of families were chosen from both the original and rewritten versions of FACES with the following criteria:

(1) Each item was to be more highly correlated with

its own than any other scale. When neither alternative met this criterion, as was true in many cases, the item that most closely fit the criterion was chosen.

(2) Scale intercorrelations between the final revised version and the original version were to be high positive correlations, so that the original FACES and the final revision are substantially the same.

(3) Alpha internal consistency was to remain high.

(4) The reading level of the final version was to be lower (easier to read) than that of the original.

When these criteria were followed, 14 of the cohesion items, 16 of the adaptability items and 3 of the social desirability items were altered for the final version of the questionnaire. These items, their item correlation coefficients and the subscales to which they belong are listed in Appendix A.

In addition to the above changes, some format changes

were also made. To make administration to a large number of










people easier and to minimize problems with transcription of answers to a separate answer sheet, the questions were written with the answer immediately below. This enabled the respondent to simply circle the best choice.

The final revised version (referred to from this point on as FACES-R) is a questionnaire that is highly similar to Olson's original FACES, but it is easier to read (low sixth grade level). More of the FACES-R items correlated with their own scale rather than the other two scales, and each scale had good internal consistency. A copy of FACES-R is in Appendix B. The correlations between the scales of Olson

and others' (Note 3) original FACES and that of FACES-R are available in Table 4. The correlations between the three original and revised scales were highly significant. Thus, the revision is substantially the same in meaning, while being improved psychometrically, making it more likely that

Olson and others' (Note 3) theory of the independence of the cohesion and adaptability dimensions would be supported.













CHAPTER IV
FACES-R STUDY



Method


Subjects were 248 families who had at least one child

between the ages of 2 and 18 living in the home. of these, 42 families were "clinic" families being evaluated or with at least one member in treatment for a problem that could be seen as appropriate for family intervention. These clinic families were seen at either a community mental health center or in the psychiatry department of a teaching hospital. The remaining 206 families were non-clinic families from a medium size metropolitan, university city. About half were solicited through the elementary and secondary school run by the university, while the other half were solicited through girl scout troops and church groups.



Procedure


In the case of the clinic families, therapists were

asked to administer the measures to their patients at evaluation or early in treatment. They were also asked to provide descriptive data on the referring problem as well as rate the families on cohesion and adaptability as defined












by Olson et al. (Note 3). All materials were then given to the researcher to be scored anonymously.

The participation of non-clinic families was solicited by letter through the school and a packet with all necessary forms was sent home with the child if the parents agreed to participate. For the other groups (e.g., girl scouts), the researcher went to a group meeting, explained the project to the children and gave them a packet to take home with a letter of explanation enclosed for their parents. In both cases, follow-up phone calls were made to answer questions, encourage response and arrange for the pick up of the completed questionnaires. Anonymity was insured by use of only first names on the questionnaire. Family members were instructed to complete the questionnaires by themselves and not to read those of other members. All members of the family age 11 or older were asked to participate and informed consent (see Appendix C) was obtained from all families.



Measures


Forms. FACES-R (see Appendix B) was the primary measure used in the study. In addition, a family demographic form (in Appendix C) was completed by a parent of each family. For the clinic families, the therapists were also asked to complete the therapist form (also in Appendix C).










Scoring. FACES-R was scored by Olson and others'

(Note 3) directions. This procedure involved weighting the enmeshed and chaotic items by multiplying the response by three. The moderate items were multiplied by two and the disengaged and rigid items were multiplied by one. One modification was made to handle the problem of missing answers since Olson and others (Note 3) did not suggest a method. If a question was left blank, its value was estimated based on that subject's responses to the other five items on the subscale. In no case were more than two items in a subscale estimated.















CHAPTER V
RESULTS OF FACES-R STUDY



Sample Characteristics


While an attempt was made to obtain a cross-section of families of many types, there were nonetheless differences. Socioeconomic status estimates (SES) were obtained by using Hollingshead and Redlich's (1958) two-factor scale which estimates SES from occupation and education of the providing parent. In cases where both parents work, the highest level obtained by computing the scores for each parent was used. Table 5 summarizes the characteristics of this group as well as the clinic group. Where the percentages do not add to 100, the others were not able to be classified from the data. As can be seen from the table, the clinic group was skewed in the direction of lower SES and had proportionally more blacks. The non-clinic group was skewed in the direction of higher SES and white families.

The clinic families also were more likely to be single parent families, and the non-clinic families more often had both parents working. Proportionally more of the clinic families had young children living in the home.





Race


Table 5
Percentages of Clinic and Non-Clinic Groups
Sample Characteristics


Clinic


Non-Clinic


Black White


39% 61%


6% 92%


32%
21% 31%
12% 4%

16% 56%


21% 56% 51%
12%


SES
1 (high)
2 3
4
5 (low)

Single Parent Families Families with Two Parents Working


24% 26% 39%

54% 34%


37% 68% 53% 10%


Children Living in Home
Under age 6
Age 6-12 Age 13-17
Over age 18









Reliability of the Scales


Internal consistency assessed using Cronbach's alpha (Cronbach, 1951) was high. For the average family scores on cohesion, it was .95, while for the average adaptability scale scores, it was .92. Social desirability had an alpha coefficient of .68. Even when only the youngest respondents (age 11) were considered, reliability calculated in this manner remained high for cohesion and adaptability (.92 and .90, respectively), but lost some on social desirability (.57).

Correlations between family members' scores can be considered another form of interrater reliability. These are shown for the cohesion and adaptability scales in Table 6. For the cohesion scale, fathers' scores correlated significantly with those of mothers and oldest siblings, while mothers' scores did so with those of both siblings. The siblings' scores also followed this pattern. On the adaptability scale, only the mothers' scores and the second oldest siblings' scores correlated significantly. In all cases, even the significant correlations are low.



Independence of the Scales


In Olson and others' (Note 3) model, the dimensions of adaptability and cohesion are assumed to be independent. The two FACES scales were also assumed to be independent, but this has never been adequately tested. To assess the

















Table 6
Correlations between Family Members


Cohesion Adaptability
Mother Oldest Second Mother Oldest Second Sibling Oldest Sibling Oldest
Sibling Sibling

Father .38* .19* .07 -.08 -.05 -.01
p<.03 p<.03 p<.62 p<.31 p<.55 p<.92
N=.28 N=128 N=58 N=146 N=124 N=58

Mother .25* .38* .03 .24*
p<.0007 p<.0006 p<.73 p<.03
N=179 N=79 N=176 N=79

Oldest .35* .09
Sibling p<.001 p<.43
N=83 N=83


*Statistically significant










extent to which the cohesion and adaptability are independent, the scale scores were intercorrelated for the FACES-R scales. Tables 7 and 8 show scale intercorrelations for the entire sample (all fathers, mothers and children) and for average family scores.

For the entire sample, the correlation between adaptability and cohesion was low but highly significant. The same was true when the statistic was computed with average family scores.

In addition, each of the two scales was correlated

with the scale score on social desirability. For the cohesion scale this correlation was moderate and highly significant. This was true both for the entire sample and for the average family scores.

The adaptability scale showed a low and barely significant correlation with social desirability for average family scores. For the entire sample, this was not replicated.



Norms


Table 9 contains the scale means and standard deviations (where available) for the groups in the present study,

Olson and others' (Note 3) study and Portner's (1982) study. In the case of families, an average family score was used. While this is not ideal since it tends to average the extreme members of the family to a perhaps deceptive moderate












Table 7
Scale Intercorrelations entire sample (N=625)


Adaptability Social Desirability

Cohesion .24* .51*
p<.0001 p<.0001

Adaptability .06
p<.ll


*Statistically significant






Table 8
Scale Intercorrelations
average family scores (N=252)


Adaptability Social Desirability

Cohesion .30* .59*
p<.0001 p<.0001

Adaptability .12*
p<.05

*Statistically significant





Table 9
Normative Comparison of Three Studies


Cohesion Adaptability Social Desirability
non-clinic clinic non-clinic clinic non-clinic clinic

Olson et al. x=251 x=183 x=35
(Note 3) S.D.=19 S.D.=15 S.D.=5
Students N=410 N=410 N=410

Portner (1981) x=255 x=245* x=182 x=184
Families N=117 N=53 N=117 N=53
Fathers x=257 x=245* x=179 x=180 x=38 x=34*
N=117 N=53 N=117 N=53 N=117 N=53
Mothers x=253 x=243* x=179 x=182 x=37 x=33*
N=117 N=53 N=117 N=53 N=117 N=53
Adolescents X=254 x=248 x=186 X=190 x=38 x=33*
N=117 N=53 N=117 N=53 N=117 N=53

Present Study x=261 x=254 R=181 x=182 R=38 x=36
Families S.D.=14.7 S.D.=24.5 S.D.=10.4 S.D.=13.5 S.D.=3.6 S.D.=4.4
N=206 N=42 N=206 N=42 N=206 N=42
Fathers x=266 R=246 x=181 x=177 x=39 x=35
S.D.=19.5 S.D.=19.5 S.D.=14.7 S.D.=19.8 S.D.=4.5 S.D.=3.3
N=146 N=ll N=146 N=12 N=146 N=12
Mothers x=261 x=255 x=177 X=180 x=38 x=36.3
S.D.=19 S.D.=24 S.D.=12.7 S.D.=14.5 S.D.=4.3 S.D.=14.5
N=193 N=39 N=194 N=39 N=194 N=39
Adolescents x=258 x=249 x=185 x=184 x=37 x=36
S.D.=20.7 S.D.=26.9 S.D.=16 S.D.=15.5 S.D.=5.0 S.D.=15.5
N=134 N=17 N=131 N=17 N=134 N=18


*Statistically significant group differences in Portner's (1981)


study.











score, the alternative also has disadvantages. This would involve categorizing families into groups such as consensus, non-consensus but cannot take into account the strength of the variable. Partner (1982) discussed all the alternatives and settled on the use of average scores.

On each scale, the families (both clinic and non-clinic) in the present study fell within one standard deviation of

Olson and others' (Note 3) sample means. A comparison is not as easy for Portner's (1982) study since she did not report standard deviations.



Validity


There are several ways to approach the question of the validity of FACES-R. If the questionnaire is measuring cohesion and adaptability as defined by Olson et al. (Note 3), any correlation shown with ratings of the same constructs by therapists would be evidence of its validity. If the scales show significant differences between the clinic and nonclinic families, this would further support the scales' validity. Factor analyses in which the cohesion items, for example, load in the expected direction on the same factor, would also be evidence of validity. Ratings

Therapist ratings of cohesion and adaptability were obtained for the clinic families. These were correlated with












the average family scores, fathers' scores, and mothers' scores. All of the ratings for cohesnion or adaptability showed very low correlations, and these were nonsignificant. Table 10 summarizes these findings.



Table 10
Correlations of Therapists' Ratings with Scale Scores


Rated Cohesion Rated Adaptability

Average family scores -.02 .08
p<.91 p<.64

Fathers' scores 0.23 .24


mothers' scores -.08 .12
p<.65p<4




Group Differences

Determining whether the scales differentiated between

clinic and non-clinic groups was done in two ways. An analysis of covariance controlling for the effect of socioeconomic status (SES) was done to assess whether the means of the two groups were significantly different. The main effects are summarized in Table 11. In no case did SES significantly affect the results. The social desirability scale was the only one that differentated between the clinic and non-clinic group. This between groups effect was highly significant with F = 7.73, p<.006.

















Table 11
Main Effect Results Analyses of Covariance Controlling for SES




Clinic Non-Clinic


Cohesion x = 254 x = 261
SD = 24.5 SD = 14.7

Adaptability x = 182 x = 181
SD = 13.5 SD = 10.4

Social Desirability x = 36* x = 38*
SD = 4.4 SD = 3.6


*F = 3.87, p<.02


Chi-square analyses were also used to assess any group differences. Each group was divided into low, medium and high scoring families. This was done by determining the means and standard deviations of the non-clinic group for the cohesion and adaptability scales. Low scores were defined as those lower than one standard deviation from the non-clinic group mean. Moderate scores were within one standard deviation and high scores were higher than one standard deviation of this mean. Low scores on the cohesion scale would correspond to what Olson et al. (Note 3) called "disengaged families while high scorers would be the










"enmeshed" families. For the adaptability scale, low scorers would be the "rigid" families and high scorers the "chaotic" families. Moderate scorers, in both cases, would be the moderate or "healthy" families.

Tables 12 and 13 summarize the results of the chi-square analyses for each scale. The cell chi-square is the contribution of each cell to the total chi-square. The overall chi-square was significant only for the cohesion scale and this can primarily be attributed to the disproportionate number of clinic families scoring low on this scale.

An attempt was made to combine the two dimensions in a nine-cell chi-square but more than 20% of the cells had expected counts less than five, thus invalidating the test. The frequency counts thus obtained are displayed in Table 14. However, according to Olson et al. (Note 3), the low-low, high-high, low-high and high-low cells should have a larger proportion of clinic families than non-clinic families if these scales are able to distinguish the two groups, while the other cells should have proportionately more non-clinic families. As can be seen, this was not consistently true. Factor Analyses

The third support for the validity of the scales would be factor analyses consistent with the composition of the scales. The cohesion items, for example, should factor analyze with one factor having "enmeshed" item loading high and












Table 12
Chi-Square Analysis--Cohesion Scale
Average Family Score


Frequency
Cell Chi Square
Percent
Row Percent
Column Percent Clinic Non-Clinic Total

14 28 42
6.7 1.4
Low 5.65 11.29 16.94
(disengaged) 33.33 66.67
33.33 13.59

22 152 174
1.9 0.4
Moderate 8.87 61.29 70.16
12.64 87.36
52.38 73.79

6 26 32
0.1 0.0
High 2.42 10.48 12.90
(enmeshed) 18.75 81.25
14.29 12.62

Total 42 206 248
16.94 83.06 100.00


Chi Square =10.381* p<.006

*Statistically significant.












Table 13
Chi-Square Analysis--Adaptability Scale
Average Family Score


Frequency
Cell Chi-SquareToa
PercentToa
Row Percent Frequency
Column Percent Clinic Non-Clinic Percent

8 27 35
0.7 0.1
Low 3.23 10.89 14.11
(rigid) 22.86 77.14
19.05 13.11

23 145 168
1.0 0.2
moderate 9.27 58.47 67.74
13.69 86.31
54.76 70.39

11 34 45
1.5 0.3
High 4.44 13.71 18.15
(chaotic) 24.44 75.56
26.19 16.50

Total frequency 42 206 248
Percent 16.94 83.08 100.00


Chi-Square = 3.93, p<.14









Table 14
Frequencies and Percent of Families in each Category for Two Scales Combined



LOW aerate Cohesion - High
(disengaged) (enmeshed) Totals
Adaptability F % F % F % F %

Low C 4 9 C 4 9.5 C 0 0.0 C 8 19.0
N 9 4.4 N 17 8.3 N 1 0.5 N 27 13.1


Moderate C 8 19.0 C 13 31.0 C 2 4.8 C 23 54.8
N 16 7.8 N 107 52.0 N 22 10.7 N 145 70.4


High C 2 4.8 C 5 11.9 C 4 9.5 C 11 26.2
N 3 1.5 N 28 13.6 N 3 1.5 N 34 16.5



TOTAL
Clinic 14 33.4 22 52.4 6 14.3
Non-Clinic 28 13.6 152 73.8 26 12.6












positive and "disengaged" items loading high and negative on that factor. The moderate items should show high loadings, also on the same factor. The same should be true for the factor analysis of the adaptability items with the "chaotic" items being positive and the "rigid" items negative. In addition, the factor scale scores obtained using the weights from these factor analyses should not correlate with each other since the two scales should be independent if they are measuring two distinct constructs.

- An unrotated principal axis orthogonal factor analysis was carried out on the items of each scale. Since there was a large drop in both cases from factor one to factor two in eigen values and percentage of variance accounted for, a one factor solution seemed most appropriate using Cattell's scree test (1966). However, the first five factors are included in Tables 15 and 16 for comparison. Olson and others' (Note 3) data are also included. See Appendix D for the factor loadings of each item on the first factor.

The items tended to load high and in the appropriate direction (e.g., negatively for disengaged items) with few exceptions. The patterns of the factor analyses were the same for average family score, fathers' scores and mothers' scores. Unfortunately, in each case the amount of the total variance accounted for was not high.

A factor analysis of the social desirability items also yielded a one-factor solution with the first factor having an eigen value of 6.3 and accounting for 42% of the variance.











Table 15
Eigen Values and Variance Accounted for by First Five Cohesion Factors


Factor
Sample 1 2 3 4 5


Present Study
Total Sample
(av. family score)


Fathers


Mothers


Olson et al.
(Note 3)


Eigen Value % of variance Cum. variance

Eigen value % of variance Cum. variance

Eigen Value % of variance Cum. variance

Eigen Value % of variance Cum. variance


12. 7 23.5 23.5

8.2 17. 2 17.2

8.3
17.4 17. 4

17.39 33.7 33.7


3.2 5.8 29. 3

3.3 7.0
24.1

3.1
6.4 23.8

7.3
14. 1 47.8


2.8 5.1
34.5

2.7 5.5 29.7

2.4 5.1 28.9

5.7 11. 1 58.9


1.9 3.6 38.0

2.1 4.3 3.4


3.8 32.7

2.4 4.6 63.5


1.8
3.4 41.4

1.8 3.8 37.8

1.7 3.5 36.2

1.8 3.6 67.0










Table 16
Eigen Values and Variance Accounted for by First Five Adaptability


Factors


Factor
Sample 1 2 3 4 5

Total Sample Eigen Value 10.2 3.9 2.6 1.5 1.5
% of variance 24.2 9.2 6.2 3.6 3.6
Gum. variance 24.2 33.4 39.7 43.3 46.8

Fathers Elgen Value 6.9 2.7 2.4 1.7 1.5
% of variance 19.7 7.7 7.0 4.8 4.2
Gum. variance 19.7 27.4 34.4 39.2 43.4

Mothers Eigen Value 6.2 2.8 2.3 1.6 1.5
% of variance 17.7 7.9 6.6 4.4 4.1
Gum. variance 17.7 25.6 32.2 36.7 40.8

Olson et al. Bigen Value 36.7 12.6 3.5 1.7 1.6
(Note 3) % of variance 58.5 20.1 5.6 2.7 2.5
Gum. variance 58.5 78.6 84.2 86.9 89.4












The scoring coefficient matrix obtained from each factor analysis is an empirically derived weighting of each item's contribution to that factor. As such, it is the most suitable weighting for that item and can be used to score the factor to obtain a factor score for each of the three scales. This was done and the three obtained factor scores were then intercorrelated to assess the extent to which they are independent and measure distinct concepts. Table 17 displays the results of these correlations. All three factor scores were highly correlated at a highly significant level.



Table 17
Intercorrelations of Factor Scores for FACES-R Scales


Adaptability Social Desirability

Cohesion -8*.86*
p<.0001 p<.0001

Adaptability - .76'*
p<. 0001


*Statistically significant















CHAPTER VI
ATTEMPTS TO IMPROVE THE SCALES



Item Analyses and Further Factor Analyses


Because the factor scores were so highly intercorrelated, implying that they were measuring highly similar constructs, an attempt was made to improve the scales by selecting the best items from each. It was hoped that by doing this scales could be formed that would still account for a reasonable amount of variance yet be more independent of each other.

An item analysis was done using the correlations of each item with its own scale. Cohesion and adaptability items were retained if they correlated significantly (p<.05) with their own scale and had factor loadings, in the initial factor analyses, greater than .60. This led to the retention of 15 cohesion items (#20, 39, 58, 60, 24, 62, 98, 47, 66, 84, 102, 35, 91, 18, 37) and 12 adaptability items (#2,

59, 95, 27, 46, 65, 101, 15, 53, 108, 55, 74). All 15 social desirability items were retained. They all correlated significantly with their own scale but some loaded low on their first factor.

Factor analyses were carried out separately on the retained cohesion and adaptability items and again the cohesion items yielded a one-factor solution. This first factor had












an eigen value of 7.5, accounting for 47% of the variance. The adaptability items also yielded a one factor solution with the first factor having an eigen value of 5.6, accounting for 49% of the variance. Appendix D has the loadings of each item on the first factor.

Using the results of the factor analyses on the retained items, factor scores were again calculated and intercorrelated. Table 18 contains the results of these intercorrelations. Again, all three factor scores were still highly significantly correlated.

Table 18
Intercorrelations of Factor Scores for Retained Items


Adaptability-Retained Social Desirability

Cohesion- -.85* -.81*
Retained p<.0001 p<.0001

Adaptability- .77*
Retained p .0001

*Statistically significant

Since there were now less items, it was possible to combine these in one factor analysis to see if they yielded the three factors which would be expected if they measure three distinct dimensions. Thus, the 15 social desirability items, 15 retained cohesion items and 12 retained adaptability items were combined in a factor analysis which yielded a one-factor solution. This factor had an eigen value of 16.7 accounting for 40% of the variance. An examination of the items loading










on this factor revealed that most of the social desirability, cohesion and adaptability items loaded moderate to high on the factor. The conservative way to describe the factor is as a social desirability factor. Table 19 contains the items along with their factor loadings on the first factor.

Since the initial attempt to choose items that would yield a distinct factor for each of the three constructs failed, another attempt was made to select a group of items that met even more stringent criteria and would show evidence for some construct other than just social desirability.

Again using the correlation of each item with its own and other scales, items were selected that correlated more with their own than with either of the other two scales. This resulted in 13 cohesion items (#76, 3, 5, 24, 80, 45, 64, 11, 104, 51, 75, 93, 111) and 15 adaptability items (#77, 4, 42, 61, 44, 63, 81, 99, 31, 87, 105, 72, 108, 17, 92) being retained.

A factor analysis of these retained cohesion items

yeilded another one-factor solution with the first factor having an eigen value of 2.9 accounting for 22.5% of the variance. The same was true for the retained adaptability items. In this case, the one factor accounted for 23% of the variance and had an eigen value of 3.4 Appendix D contains the factor loadings of each item on the first factor.

Intercorrelations of the factor scores obtained were

again highly significant but the correlations were somewhat reduced. These are shown in Table 20.








55


Table 19
Items and Loadings on Factor One All Retained Items


Factor
Scale Item # Item Loading

SD 107 1 don't think any family could live .79
together with greater harmony than
my family.

A 65 Our family does not discuss its prob- .77
lems.

A 27 We talk about our problems and feel .79
good about the answers.

SD 48 my family has all the qualities I've .75
always wanted in a family.

A 46 We feel good about how we solve prob- .75
lems in our family.

SD 71 1 don't think anyone could possibly .73
be happier than my family and I when
we are together.

C 91 We don't talk much to each other at -.73
home.

C 47 Although family members have individ- .73
ual interests, they still participate
in family activities.

C 102 Members of our family share many in- .71
terests.

C 66 Our family doesn't do things together. -.69

84 We have difficulty thinking of things -.68
to do as a family.

SD 103 Our family is as well adjusted as any .68
family in this world can be.

A 59 In our family, it's important for every- .67
one to express their opinion.

A 15 In our family we make the rules to- .66
gether





Item

If I could be a part of any family in the world, I could not have a better match.

We try to plan some things during the week so we can all be together.

Family ties are more important to us than any friendship could possibly be.

Family members find it easier to discuss things with persons outside the family.

If one way doesn't work in our family, we try another.

We like to spend some of our free time with each other.

Family members do not turn to each other when they need help.

Family members feel free to say what's on their mind.

It seems as if family members can never find time to be together.

Family members do not check with each other when making decisions.

Family members feel they have no say in solving problems.

We help each other find new ways to do things.

We respect each other's privacy.

Home is one of the loneliest places to be.

Family members understand each other completely.

In our family we are on our own when there is a problem to solve.


Factor
Loading

.66 .65 .65 .65



.64 .63


-.62


.62


-.62


-.62


.62 .62 .61

-.61


.60


-.60


Scale Item

SD 57


74 24 39


2


98 18 101 55 35 58 85 37













Factor
Scale Item # Item Loading

SD 19 my family always understands me no .60
matter how I feel.

SD 89 My family could be happier than it is. -.59

SD 33 Every new thing I've learned about my -.58
family has pleased me.

A 95 Family members rarely say what they -.56
want.

SD 52 There are times when I don't like my -.55
family a lot.

SD 38 1 have never regretted being with my .54
family, not even for a moment.

A 108 It is hard to know what the rules are -.52
in our family because they always
change.

A 53 When rules are broken, family members .52
are treated fairly.

SD 14 There are some things I need that 1 .50
don't get from my family.

SD 10 There are times when other family mem- .44
bers do things that make me happy.

SD 29 Our family is not a perfect success. .39

SD 67 If my family has any faults, I am not -.03
aware of them.












Table
Intercorrelations of Factor


Scores for Retained Items


Adaptability-Retained Social Desirability

Cohesion- -.44* .62*
Retained p<.0001 p<.0001

Adaptability- -.50*
Retained p<.0001


*Statistically significant















CHAPTER VII
DISCUSSION



Since the FACES-R questionnaire was designed as a selfreport instrument for use in assessing the functioning of families, the present sample was appropriate to study the scales' psychometric properties since it contained clinic and non-clinic families. Portner (1982) was the only other study with this instrument where responses were obtained

from parents and children. Olson and others' (Note 3) standardization sample was undergraduates who were asked to respond retrospectively about their families of origin. While the present study's clinic and non-clinic groups differed somewhat, primarily in SES and race, the results should be comparable to Portner's (1982) study.

Before the appropriateness of the Circumplex Model can be assessed, there must be evidence supporting the ability of the FACES-R scales to reliably and validly measure the two dimensions postulated in the model.

Support for the reliability of the scales can be seen in the high alpha coefficients of internal consistency obtained in the present study. These exceeded any reported in previous studies and were contrary to Bilbro and Dreyer's (1981) findings that the cohesion scale reliabilities were low. A possible explanation was reflected in the spontaneous












comments of many respondents in the present study who said the questionnaire was too long and kept asking the same thing over and over. It may be that respondents recognized that if they answered a question in one direction, they should thereafter answer in a manner consistent with their first response.

Some support for the reliability of the cohesion scale was the extent to which family members' scores intercorrelated even though the correlations were low. This pattern was not apparent for the adaptability scale, however, There is also a question as to whether all family members' scores should be expected to correlate with each other. Theoretically, one might expect the responses of a healthy adolescent, for example, to deviate a moderate amount from those of his parents. One would expect parents' responses to agree, however. Since the latter was true for the cohesion scale, there is some evidence for the reliability of this scale.

The similarity between the present study's sample means on the scales and those of Olson et al. (Note 3) and Portner (1982) also supports the consistency of the scales. In every case, the means for the present sample were within

one standard deviation of Olson and others' (Note 3) normative sample.

It is in examining support for the validity of the

scales that problems arose. Partner (1982) attempted to address the validity question by categorizing her clinic and










non-clinic families into low-moderate and high groups. While the majority of both groups consistently fell in the moderate ranges, she had some limited success in differentiating the two groups.

The present study approached the issue of validity in several ways. Therapists were asked to rate the cohesiveness and adaptability of the clinic families. Since the definitions they were given of these constructs were those of Olson et al. (Note 3), the complete lack of any correlations between ratings and scale scores raised validity questions. Therapists may not have perceived families in a manner congruent with the families' self-perceptions. Since there was no data to support the reliability of the therapists' ratings and these were global in nature, they may not have been appropriate. The ratings were also obtained early in therapy whereas, if they had been obtained later the therapists may have had a better data base from which to rate the families. Or, there is the possibility that what was measured by the FACES-R scales was not cohesion and adaptability as defined by Olson et al (Note 3).

Examination of the scale intercorrelations lent subsistance to the latter conclusion. The cohesion and adaptability scales were clearly not independent since they had a highly significant low correlation with each other. In addition, in both Portner's (1982) study and the present study, the cohesion scale consistently correlated with social desirability.

While neither scale showed any consistent differences between the means for the clinic and non-clinic groups, the







62

cohesion scale was significantly different for the groups in a chi-square analysis. Partner (1982) found differences between her groups on this scale using both T-tests and chisquare analyses. She did not find differences using the Ttest for the adaptability scale, although she did find a difference with the chi-square statistic. The present study used an analysis of covariance to control for SES and still found no difference between the groups on these two scales.

The social desirability scale appeared to be the only

scale which consistently revealed highly significant differences between the clinic and non-clinic groups. Partner (1982) found this, as did the present study. In both studies the clinic families' scores were lower on this scale than the non-clinic families. This perception by clinic families could be interpreted as accurately reflecting reality since it is still considered somewhat socially undesirable to require psychotherapy, and it would be difficult to perceive one's family as perfect yet still seek help. The social desirability scale measured the tendency to perceive one's family in a socially favorable light, rather than a tendency to bias the presentation of the family in that direction. In other words, it is not a "faking" scale.

The factor analyses were another way to approach the question of validity. while the findings of the present study's factor analyses did not replicate either those of Oldon et al. (Note 3) or Portner (1982), they did support a bipolar cohesion dimension and a bipolar adaptability












dimension. That is, the cohesion items had one factor on which items, for the most part, loaded in the expected direction depending upon whether they were "disengaged" (loaded negatively) or "enmeshed" (loaded positively). The same was true for the adaptability items. These findings were consistent with the Circumplex Model.

It is unclear why the factor analytic solutions of the two earlier studies were different from those of the present study. The present findings for the cohesion scale were consistent with those of Bilbro and Dreyer (1981) who also found a one-factor solution with those items.

Problems with validity again become evident when the

factor scores obtained from the factor analyses of the three scales were intercorrelated. If there existed three separate dimensions these empirically derived factor scores should not correlate with each other. To the contrary, they showed very high and significant intercorrelations. This implied that all three factors obtained from the three separate factor analyses were tapping highly similar dimensions.

Taking the items from each factor which loaded highest on their own factor and correlated the most with their own scale in an item analysis, did not alter the results of a subsequent factor analysis and factor scores intercorrelation. The scales remained highly similar.

In addition, a factor analysis done with all these retained items yielded only one factor rather than the three












separate factors that would substantiate three separate dimensions. This one factor was best described as a social desirability factor. A third attempt to select a different set of items to retain still did not yield factor scores that were uncorrelated.

Thus, the factor analytic results implied that even

though there was some evidence that there were differences between the groups on cohesion and less clearly on adaptability, these distinctions may not be due to the theoretical dimensions the scale purports to measure. Instead, it is more

likely that these differences were due solely to differences in the groups' perceptions of the extent to which their family is a "perfect" family.

The lack of support in the present study for the validity of the Family Cohesion and Adaptability Evaluation Scales must be taken in context. The entire field of family measurement is in its infancy. Very few attempts have been made to develop a measure of constructs relevant to family functioning and useful to family therapists. The dimensions that FACES-R sought to measure have strong theoretical support among family therapists. It is possible these exist only in the minds of therapists. It is more likely, however, that the problems in measuring by self-report, behaviors of which family members are unawares well as the inherent tendency of members to perceive their families in the best possible light, have handicapped this first attempt at measuring








6 5



family cohesion and adaptability. In future attempts at such a self-report measure, the more concrete and easily interpretible the item, the more likely it will be to elicit an objective response. While the findings reported here are discouraging, it is by no means time to give up. Olson and his colleagues are to be commended for making this first attempt. However, these scales are not appropriate, in their present form, for use by therapists in assessing families.















APPENDIX A
PILOT STUDY REVISED ITEMS



Changed Cohesion Items


Item No.

1. Original:

Revised: 11. Original:

Revised:


16. Original:

Revised:


24. Original:

Revised: 28. Original:

Revised:


43. Original:

Revised: 49. Original:

Revised: 54. Original:

Revised:


Family members are concerned with each other's welfare.
People in my family care about each other.

In our family we know where all family members are at all times. In our family we always know where everyone is.

It seems like there is never any place to be alone in our house. It seems like there is no place to be alone in our house.

Family members like to spend some of their free time with each other. We like to spend some of our free time with each other.

Family members share almost all interests and hobbies with each other. In our family we like almost all the same things.

Even when everyone is home, family members spend their time separately. Even when we are all home, we all do different things.

Family members are totally on their own in developing their ideas.
Family members are on their own in doing a project.

Family members don't enter each other's areas or activities. Family members don't get into each others things or business.















































Subscales of Changed Cohesion Items


wt. Enmeshed Mod. Disengaged

1 Emotional Bonding #1
1 Family Boundaries #78
3 Time #80 #24 #43
0 Friends
1 Interests & Rec. #28
#86
3 Independence #11
0 Coalition #91, 49
3 Space #16 #54
2 Decision Making #93
#56
TOTAL 5 5 4


Item No.


56. Original:


Revised: 78. Original:

Revised: 80. Original:

Revised: 86. Original:
Revised: 91. Original:

Revised: 93. Original:

Revised:


Family members discuss important decisions with each other, but usually make their own choices.
We talk about big decisions with each other but make our own choices.

Family members feel comfortable inviting their friends along on family activities. When my family does something, friends are welcome.

Family members feel pressured to spend most free time together.
Family members feel we should spend most of our free time together.

It seems as if we agree on everything. It seems as if we agree on everything in my family.

Family members seem to avoid contact with each other when at home. We don't talk much to each other at home.

We decide together on family matters and separately on personal matters. We decide family matters together and personal matters by ourselves.















Item Correlation Coefficients
Changed Cohesion Items


Cohesion Adaptability Social Desirability


1 Original Revised


11 0
R


16 0
R


24 0
R


28 0
R


43 0
R


49 0
R


54 0
R


56 0
R


78 0
R


80 0
R


86 0
R


91 0
R


93 0
R


r =.11 r = .08


.27 .36


. 19 .37


.34 .42


.21 .21


-.12
-.36


-.06
-.10


-.02
-. 11


.02 .10


.08
-.28


.07 .30


.32
.41


-. 10
-.25


-.05
.05


r = .26 r = .18


-.41
-.24


-.06
-.06


.12 .10


.12
.04


-.19
-.04


-.09
-.09


.13
-.08


.31 .08


.02 .18


.30 .03


.18 .17


.08 .01


- .21
.02


r = .33 r = .35


. 15
.02


-.35
-.35


.24 .37


.46 .51


-.23
-.20


-.15
-.20


.45 .36


.10
.48


.18
-.01


-.08
-.00


.71 . 58


-.44
-.66


-.06
-.01












Changed Adaptability Items


6. Original: Revised: 8. Original: Revised: 12. Original:

Revised:


15. Original:
Revised: 17. Original:

Revised: 21. Original:

Revised: 27. Original:

Revised: 31. Original:

Revised: 34. Original:

Revised:


40. original:

Revised:


Family members are afraid to tell the truth because of how harsh the punishment will be. Family members are afraid to tell the truth because they will get in trouble.

Family members talk a lot but nothing ever gets done.
In our family we talk about doing a lot of things but we never do them.

Family members have some say in what is required of them.
Family members get some say in what they have to do.

Family members make the rules together. In our family we make the rules together.

It is difficult to keep track of what other family members are doing. It is hard to know what other family members are doing.

When our family has an argument, family members just keep to themselves. When our family disagrees, family members don't talk.

Family members discuss problems and usually feel good about the solutions. We talk about our problems and feel good about the answers.

No one in our family seems to be able to keep track of what their duties are. No one in our family knows what their choices are.

our family has a rule for almost every possible situation.
Our family has a rule for almost all situations.

It is hard to know what other family members are thinking.
It is hard to know what other family members think.











































Subscales of Changed Adaptability Items


wt. Chaotic Mod. Rigid

2 Assertiveness #40 #21
1 Control #61
2 Discipline #99
# 6
4 Negotiation #8, 83 #27, 46

4 Roles #31 #12, 69 #50
2 Rules #15 #34
2 System Feedback #17 #55
TOTAL 6 6 5


46. Original:

Revised: 50. Original:

Revised: 61. Original:
Revised: 69. original:

Revised: 83. Original:


Revised:



99. Original:

Revised:


We feel good about our ability to solve problems.
We feel good about how we solve problems in our family.

Once a task is assigned to a family member, there is no chance of changing it. If a family member is given a job, they can't change it.

There is no leadership in our family. There is no leader in our family.

In our family, everyone shares responsibilities.
In our family we share duties.

When trying to solve problems, family members jump from one attempted solution to another without giving any of them time to work. When there is a problem, we jump from one answer to another without giving any of them a chance to work.

Family members are severely punished for anything they do wrong. People in our family get in lots of trouble if they do anything wrong.








71



Item Correlation Coefficients
Changed Adaptability Items


Adaptability Cohesion Social Desirability

6 Original -.16 .14 -.36
Revised -.22 .18 -.36

8 0 .17 .03 -.39
R .42 .12 -.00

12 0 .24 .15 .27
R .15 -.06 .15

15 0 .35 .11 .46
R .34 -.15 .32

17 0 .20 .05 -.12
R .21 .04 .08

21 0 .02 .17 -.18
R -.34 .01 -.59

27 0 .24 .50 .57
R .18 .15 .55

31 0 .18 .25 -.09
R .36 .22 -.15

34 0 -.20 .31 .09
R -.24 .19 -.13

40 0 .02 -.05 -.35
R -.15 -.03 -.14

46 0 .21 .17 .63
R .37 .05 .72

50 0 .10 .35 -.20
R .10 .11 -.11

61 0 .64 -.01 .34
R .70 -.13 .24

69 0 .27 .26 .38
R -.16 -.09 .21

83 0 .42 -.10 -.25
R .50 .03 -.15

99 0 .17 .01 .15
R -.22 .01 -.19












Changed Social Desirability Items


Changed Social Desirability Items


Social Desirability Adaptability Cohesion

14 Original .30 -.02 .17
Revised -.35 .06 -.06

19 0 .57 .42 .48
R .61 .33 .26

52 0 -.45 -.01 -.14
R -.68 -.17 -.08


14. Original:

Revised:


19. Original:

Revised:


52. Original:

Revised:


I have some needs that are not being met by family members.
There are some things I need that I don't get from my family.

My family completely understands and sympathizes with my every mood. My family always understands me no matter how I feel.

There are times when I do not feel a great deal of love and affection for my family. There are times when I don't like my family a lot.



Item Correlation Coefficients
















APPENDIX B
FACES-R QUESTIONNAIRE



First Name Age Sex
IDENTIFICATION:

FACES-R

INSTRUCTIONS: Please answer each question by circling the
number that best describes your family.


1. People in my family care about each other.
1 2 3 4
Never true Sometimes true Usually true Always true

2. Family members feel free to say what's on their mind.
1 2 3 4
Never true Sometimes true Usually true Always true

3. We don't have spur of the moment guests at mealtime.
1 2 3 4
Never true Sometimes true Usually true Always true

4. It is hard to know who the leader is in our family.
1 2 3 4
Never true Sometimes true Usually true Always true

5. It's difficult for family members to take time away
from the family.
1 2 3 4
Never true Sometimes true Usually true Always true

6. Family members are afraid to tell the truth because they will get in trouble.
1 2 3 4
Never true Sometimes true Usually true Always true

7. Most personal friends are not family friends.
1 2 3 4
Never true Sometimes true Usually true Always true

8. In our family we talk about doing a lot of things but
we never do them.
1 2 3 4
Never true Sometimes true Usually true Always true












9. Family members feel guilty if they want to spend some time alone.
1 2 3 4
Never true Sometimes true Usually true Always true

10. There are times when other family members do things
that make me unhappy.
1 2 3 4
Never true Sometimes true Usually true Always true

11. In our family we always know where everyone is.
1 2 3 4
Never true Sometimes true Usually true Always true

12. Family members get some say in what they have to do.
1 2 3 4
Never true Sometimes true Usually true Always true

13. The parents in our family stick together.
1 2 3 4
Never true Sometimes true Usually true Always true

14. There are some things I need that I don't get from my
family.
1 2 3 4
Never true Sometimes true Usually true Always true

15. In our family we make the rules together.
1 2 3 4
Never true Sometimes true Usually true Always true

16. It seems like there is no place to be alone in our house.
1 2 3 4
Never true Sometimes true Usually true Always true

17. It is hard to know what other family members are doing.
1 2 3 4
Never true Sometimes true Usually true Always true

18. Family members do not check with each other when making
decisions.
1 2 3 4
Never true Sometimes true Usually true Always true

19. My family always understands me no matter how I feel.
1 2 3 4
Never true Sometimes true Usually true Always true

20. Family ties are more important to us than any friendship could possibly be.
1 2 3 4
Never true Sometimes true Usually true Always true












21. When our family disagrees, family members don't talk.
1 2 3 4
Never true Sometimes true Usually true Always true

22. Family members often answer questions that were addressed to another person.
1 2 3 4
Never true Sometimes true Usually true Always true

23. The parents check with the children before making important decisions in our family.
1 2 3 4
Never true Sometimes true Usually true Always true

24. We like to spend some of our free time with each other.
1 2 3 4
Never true Sometimes true Usually true Always true

25. Punishment is usually pretty fair in our family.
1 2 3 4
Never true Sometimes true Usually true Always true

26. Family members are encouraged to have friends of their
own as well as family friends.
1 2 3 4
Never true Sometimes true Usually true Always true

27. We talk about our problems and feel good about the answers.
1 2 3 4
Never true Sometimes true Usually true Always true

28. In our family we like almost all the same things.
1 2 3 4
Never true Sometimes true Usually true Always true

29. Our family is not a perfect success.
1 2 3 4
Never true Sometimes true Usually true Always true

30. Family members are extremely independent.
1 2 3 4
Never true Sometimes true Usually true Always true

31. No one in our family knows what their chores are.
1 2 3 4
Never true Sometimes true Usually true Always true

32. Family members feel it's "everyone for themselves."
1 2 3 4
Never true Sometimes true Usually true Always true












33. Every new thing I've learned about my family has pleased
me.
1 2 3 4
Never true Sometimes true Usually true Always true

34. Our family has a rule for almost all situations.
1 2 3 4
Never true Sometimes true Usually true Always true

35. We respect each other's privacy.
1 2 3 4
Never true Sometimes true Usually true Always true

36. Once our family has planned to do something, it's difficult to change it.
1 2 3 4
Never true Sometimes true Usually true Always true

37. In our family we are on our own when there is a problem
to solve.
1 2 3 4
Never true Sometimes true Usually true Always true

38. 1 have never regretted being with my family, not even
for a moment.
1 2 3 4
Never true Sometimes true Usually true Always true

39. Family members do not turn to each other when they need
help.
1 2 3 4
Never true Sometimes true Usually true Always true

40. It is hard to know what other family members think.
1 2 3 4
Never true Sometimes true Usually true Always true

41. Family members make visitors feel at home.
1 2 3 4
Never true Sometimes true Usually true Always true

42. Parents make all of the important decisions in our family.
1 2 3 4
Never true Sometimes true Usually true Always true

43. Even when we are all home, we all do different things.
1 2 3 4
Never true Sometimes true Usually true Always true












44. Parents and children in our family discuss together the
method of punishment.
1 2 3 4
Never true Sometimes true Usually true Always true

45. Family members have little need for friends because the
family is so close.
1 2 3 4
Never true Sometimes true Usually true Always true

46. We feel good about how we solve problems in our family.
1 2 3 4
Never true Sometimes true Usually true Always true

47. Although family members have individual interests, they
still participate in family activities.
1 2 3 4
Never true Sometimes true Usually true Always true

48. My family has all the qualities I've always wanted in a
family.
1 2 3 4
Never true Sometimes true Usually true Always true

49. Family members are on their own in doing a project.
1 2 3 4
Never true Sometimes true Usually true Always true

50. If a family member is given a job, they can't change it.
1 2 3 4
Never true Sometimes true Usually true Always true

51. Family members seldom take sides against other members.
1 2 3 4
Never true Sometimes true Usually true Always true

52. There are times when I don't like my family a lot.
1 2 3 4
Never true Sometimes true Usually true Always true

53. When rules are broken, family members are treated
fairly.
1 2 3 4
Never true Sometimes true Usually true Always true

54. Family members don't get into others things or business.
1 2 3 4
Never true Sometimes true Usually true Always true.

55. We help each other find new ways to do things.
1 2 3 4
Never true Sometimes true Usually true Always true.












56. We talk about big decisions with each other but make our
own choices.
1 2 3 4
Never true Sometimes true Usually true Always true

57. If I could be a part of any family in the world, I could
not have a better match.
1 2 3 4
Never true Sometimes true Usually true Always true

58. Home is one of the loneliest places to be.
1 2 3 4
Never true Sometimes true Usually true Always true

59. In our family, it's important for everyone to express
their opinion.
1 2 3 4
Never true Sometimes true Usually true Always true

60. Family members find it easier to discuss things with
persons outside the family.
1 2 3 4
Never true Sometimes true Usually true Always true

61. There is no leader in our family.
1 2 3 4
Never true Sometimes true Usually true Always true

62. We try to plan some things during the week so we can
all be together.
1 2 3 4
Never true Sometimes true Usually true Always true

63. Family members are not punished or reprimanded when
they do something wrong.
1 2 3 4
Never true Sometimes true Usually true Always true

64. In our family we know each other's close friends.
1 2 3 4
Never true Sometimes true Usually true Always true

65. Our family dies not discuss its problems.
1 2 3 4
Never true Sometimes true Usually true Always true

66. our family doesn't do things together.
1 2 3 4
Never true Sometimes true Usually true Always true












67. If my family has any faults, I am not aware of them.
1 2 3 4
Never true Sometimes true Usually true Always true

68. Family members enjoy doing things alone as well as together.
1 2 3 4
Never true Sometimes true Usually true Always true.

69. In our family we share duties.
1 2 3 4
Never true Sometimes true Usually true Always true

70. Parents agree on how to handle the children.
1 2 3 4
Never true Sometimes true Usually true Always true

71. 1 don't think anyone could possibly be happier than my
family and I when we are together.
1 2 3 4
Never true Sometimes true Usually true Always true

72. It is unclear what will happen when rules are broken in
our family.
1 2 3 4
Never true Sometimes true Usually true Always true

73. When a bedroom door is shut, family members will knock
before entering.
1 2 3 4
Never true Sometimes true Usually true Always true

74. If one way doesn't work in our family, we try another.
1 2 3 4
Never true Sometimes true Usually true Always true

75. Family members are expected to have the approval of others before making decisions.
1 2 3 4
Never true Sometimes true Usually true Always true

76. Family members are totally involved in each other's
lives.
1 2 3 4
Never true Sometimes true Usually true Always true

77. Family members speak their mind without considering how
it will affect others.
1 2 3 4
Never true Sometimes true Usually true Always true












78. When my family does something, friends are welcome.
1 2 3 4
Never true Sometimes true Usually true Always true

79. Each family member has at least some say in major family decisions.
1 2 3 4
Never true Sometimes true Usually true Always true

80. Family members feel we should spend most of our free
time together.
1 2 3 4
Never true Sometimes true Usually true Always true

81. Members of our family can get away with almost anything.
1 2 3 4
Never true Sometimes true Usually true Always true

82. Family members share the same friends.
1 2 3 4
Never true Sometimes true Usually true Always true

83. When there is a problem, we jump from one answer to another without giving any of them a chance to work.
1 2 3 4
Never true Sometimes true Usually true Always true

84. We have difficulty thinking of things to do as a family.
1 2 3 4
Never true Sometimes true Usually true Always true

85. Family members understand each other completely.
1 2 3 4
Never true Sometimes true Usually true Always true

86. It seems as if we agree on everything in my family.
1 2 3 4
Never true Sometimes true Usually true Always true

87. It seems as if males and females never do the same
chores in our family.
1 2 3 4
Never true Sometimes true Usually true Always true

88. Family members know who will agree and who will disagree with them on most family matters.
1 2 3 4
Never true Sometimes true Usually true Always true

89. My family could be happier than it is.
1 2 3 4
Never true Sometimes true Usually true Always true












90. There is strict punishment for breaking rules in our
family.
1 2 3 4
Never true Sometimes true Usually true Always true

91. We don't talk much to each other at home.
1 2 3 4
Never true Sometimes true Usually true Always true

92. For no apparent reason, family members seem to change
their minds.
1 2 3 4
Never true Sometimes true Usually true Always true

93. We decide family matters together and personal matters
by ourselves.
1 2 3 4
Never true Sometimes true Usually true Always true

94. our family has a balance of closeness and separateness.
1 2 3 4
Never true Sometimes true Usually true Always true

95. Family members rarely say what they want.
1 2 3 4
Never true Sometimes true Usually true Always true

96. It seems there are always people around home who are
not members of the family.
1 2 3 4
Never true Sometimes true Usually true Always true

97. Certain family members order everyone else around.
1 2 3 4
Never true Sometimes true Usually true Always true

98. It seems as if family members can never find time to
be together.
1 2 3 4
Never true Sometimes true Usually true Always true

99. People in our family get in lots of trouble if they do
anything wrong.
1 2 3 4
Never true Sometimes true Usually true Always true

100. We know very little about the friends of other family
members.
1 2 3 4
Never true Sometimes true Usually true Always true













101. Family members feel they have no say in solving problems.
1 2 3 4
Never true Sometimes true Usually true Always true

102. Members of our family share many interests.
1 2 3 4
Never true Sometimes true Usually true Always true

103. Our family is as well adjusted as any family in this
world can be.
1 2 3 4
Never true Sometimes true Usually true Always true

104. Family members are encouraged to do their own thing.
1 2 3 4
Never true Sometimes true Usually true Always true

105. Family members never know how others are going to act.
1 2 3 4
Never true Sometimes true Usually true Always true

106. Certain individuals seem to cause most of our family
problems.
1 2 3 4
Never true Sometimes true Usually true Always true

107. 1 don't think any family could live together with
greater harmony than my family.
1 2 3 4
Never true Sometimes true Usually true Always true

108. It is hard to know what the rules are in our family
because they always change.
1 2 3 4
Never true Sometimes true Usually true Always true

109. Family members find it hard to get away from each other.
1 2 3 4
Never true Sometimes true Usually true Always true

110. Family members feel that the family will never change.
1 2 3 4
Never true Sometimes true Usually true Always true

111. Family members feel they have to go along with what the
family decides to do.
1 2 3 4
Never true Sometimes true Usually true Always true















APPENDIX C
AUXILIARY FORMS



Informed Consent


I understand that this is a study of how the members of my family see our family functioning. I understand that the members of my family will be asked to complete a questionnaire about what occurs in our family on a daily basis. I also understand that we will be asked to give information about the occupations and incomes of family members as well as whether they have been in therapy. This information will be confidential to the extent provided by law and no names will be used so that it will be anonymous.

I understand that there are no risks associated with this study and no direct benefits to my family for participating. It is agreed that the information may be used for educational purposes, which may include publication.

I have read the above description and my questions have been answered. I give my permission for the members of my family to participate.

I understand that we may withdraw from the study at any time if we no longer want to participate.




Signature of parent or guardian

If you would like
a copy of this
study, please put
your mailing address
here: Signatures of others in family (option ill)

Date


Signature of person obtaining Signature of person obtaining consent (witness) consent (witness)












Instructions and Questions
For Therapists Administering FACES


Identification of family:
Therapist:


All family members who are 12 years or older should be asked to complete the FACES questionnaire. After these are done, one parent (or both) should complete the Demographic Information sheet. Please make sure that each person who completes the questionnaire is identified on that questionnaire.

1. Who was the identified patient when this family presented? If there was no one person singled out by
the family, put none.

2. What was the presenting problem?

3. Which stage was this family in when they completed the
questionnaire? (Check one)
referred but not yet evaluated
evaluation in process or completed
being seen in psychotherapy
therapy successfully completed
therapy terminated without therapist's recommendation
referred elsewhere after evaluation

4. If you did an initial evaluation, what was your recommendation?

5. Did the family follow through?

6. Had any member been in therapy before? If yes, state
who and why.

7. Family cohesion is defined as: "The emotional bonding
which members have toward one another and the individual autonomy that a person has in the family system."
At the extreme of high family cohesion, there is an
over-identification with the family which results in extreme bonding and limited individual autonomy. The low extreme is characterized by low bonding and high
autonomy from the family. A balanced degree of family cohesion is thought to be the most conducive to effective family functioning and to facilitating individual
development.












On this five-point scale, please rate this family's
level of cohesion at the time they completed FACES.

1 2 3 4 5
too little cohesion balanced too much cohesion

8. Family adaptability is defined as: "The ability of a
marital/family system to change its power structure,
role relationships, and relationship rules in response to situational and developmental stress." The assumption is that an adaptive system requires a balance between change and stability.
On this five-point scale, please rate this family's
level of adaptability at the time they completed FACES.

1 2 3 4 5
too little change balanced too much change

9. Does any member of this family have a chronic illness?
Who and what?




















IS THIS PERSON HIGHEST GRADE
FIRST NAME AGE SEX RACE LIVING AT HOME? OCCUPATION COMPLETED


others




Has your family or any member ever psychiatrist? yes no


Family Demographic Information


This form is to be completed by


a parent. Please use first names only.


(self) (spouse)

your children


yes yes


yes yes yes yes yes yes yes yes yes by a


no no


no no no no no no no no no


counselor,


been seen


therapist, psychologist or


Does any member of your family have a chronic illness for which they are under a doctor's care? If yes, which member and what illness? yes -no





First Factor Loadings


APPENDIX D
RESULTS OF FACTOR ANALYSES


of All Adaptability Items


Item #


2
21 40 59 77 95
4
23
42 61 79 97
6
25
44 63 81 99
8
27
46


Factor Loading


-.61
.57
.54
-.71
.25
.63 .13
-.43
.08 .09
-.54
.50
.52
-.59
-.38
.33
.43 .26 .35
-.74
-.71


Item #


65 83 101
12 31 50 69 87 105 15
34 53 72 90 108 17 36 55
74 92 110


Factor Loading


.78 .60 .66
-.49
.48 .18
-.56
.32 .38
-.69
-.08
-.61
.38
.02 .61 .55
-.01
-.61
-.67
.49 .19

















(All Cohesion Items)


First Factor Loadings


Item #


Factor Loading


.56 .63
-.60
-.62
.47 .56 .01
-.40
.51
-.63
.33
-.00
-.09
.64
-.34
.65
.44
-.64
-.38
.48 .16 .57 .39
-.49
-.24
.54 .75


Item #

66
84 102 11 30
49 68 86
104 13 32 51 70 88 106 16 35
54 73 91 109 18 37 56 75 93 ill


Factor Loading


-.69
-.67
.72
.47
-.09
-.30
.21 .56 .35 .59
-.72
.35
.60
-.19
-.54
-.37
.63 .39 .31
-.73
-.33
-.63
-.60
.20 .32
.14
-.08


1
20 39 58 76
94
3
22 41 60 78 96
5
24 43 62 80 98
7
26
45 64 82 100
9
28
47





First Factor Loadings of


First Factor Loading of 13 Retained Cohesion Items


15 Retained Cohesion Items


I tem #


Factor Loading


-.64

.65

.64 .65

-.66

-.69

.68

-.76


Item #


66

84 102 35 91 18 37


Factor Loading


-.76

.73

-.74

-.57

.76

.65 .65


First Factor Loadings of 12 Retained Adaptability Items


I tem. #


Factor Loading


.68 .76

-.62

.79 .76

-.79


Item #


101 15 53 108 55

74


Factor Loading



-.68

.72 .59

-.58

.65 .69


Item #


Factor Loading


.65 .18 .23

.59 .66

.41 .54


Item #


Factor Loading


11

104 51 75 93 ill


.61 .39 .38 .55 .31 .35










90




First Factor Loading 15 Retained Adaptability Items


Item #


Factor Loading


.24


Item #


Factor Loading


99 31 87 105 72 108 17 92


.08 .65 .29
.43 .53 .71 .65
.54


-.20
.45
-.14
.58 .63
















REFERENCE NOTES


1. Robertson, D., and Hyde, J. The factors of the family
environment scale. Unpublished and undated manuscript,
available from the first author P.O. Box 249, Warren
State Hospital, Warren, PA.

2. Janes, C.L. and Hesselbrock, V.M. Perceived family environment and school adjustment of children of schizophrenics. Paper presented at the American Psychological Association Eighth Annual Convention. Washington, D.C.,
September, 1976.

3. Olson, D., Bell, R., and Portner, J. Family adaptability and cohesion evaluation scales. Family Social Science Department, University of Minnesota, Unpublished
manuscript, 1978.















REFERENCES


Angell, R. The family encounters the depression. New York:
Charles Scribner's Sons, 1936-.

Bilbro, T.L., & Dreyer, A.S. A methodological study of a measure of family cohesion. Family Process, 1981, 20, 419427.

Cattell, R.B. The scree test for the number of factors. Multivariate Behavioral Research, 1966, 1(2), 245-276.Cromwell, R., Olson, D., & Fournier, D. Tools and techniques
for diagnosis and evaluation in marital and family therapy. Family Process, 1976, 14(1), 1-49.

Cronbach, L.J. Coefficient alpha and the internal structure
of tests. Psychometrika, 1961, 16(3), 297-334.

DeWitt, K.N. The effectiveness of family therapy. Archives
of General Psychiatry, 1978, 35, 549-561.

Edmonds, V. Marital conventionalization: Definition and
measurement. Journal of Marriage and the Family, 1967,
29, 681-688.

Fry, E. A readability formula that saves time. Journal of
Reading, 1968, 11, 514-577.

Gurman, A., & Kniskern, D. Research in marital and family
therapy: Empirical, clinical and conceptual issues.
In S.L. Garfield and A.E. Bergin (Eds.) Handbook of psychotherapy and behavior change (Revised Edition)Y.
New York: Wiley, 1978.

Hollingshead, A., & Redlich, F. Social class and mental illness. New York: John Wiley and Sons, Inc., 1958.

Jacob, T. Family interaction in disturbed and normal families:
A methodological and substantive review. Psychological
Bulletin, 1975, 82(l), 33-54.

Korsch, B., Fine, R., & Negrete, V. Noncompliance in children
with renal transplants. Pediatrics, 1978, 61(6), 872-876.

Masten, A.S. Family therapy as a treatment for children: A
critical review of outcome research. Family Process,
1979, 18(3), 323-335.











Minuchin, S., Rosman, B., & Baker, L. Psychosomatic families: Anorexia nervosa in context. Cambridge: Harvard University Press, 1978.

Moos, R. The family environment scale manual. Palo Alto,
California: Consulting Psychologists Press, n.d.

Moos, R., & Moos, B. A typology of family social environments. Family Process, 1976, 15(4), 357-371.

Olson, D., Russell, C., & Sprenkle, D. Circumplex model of
marital and family systems II: Empirical studies and
clinical intervention. Advances in family intervention,
assessment and theory. N.P.: JAI Press, 1980, 1, 129176.

Olson, D., Sprenkle, D., & Russell, C. Circumplex model of
marital and family systems I: Cohesion and adaptability dimensions, family types and clinical applications.
Family Process, 1979, 18(1), 3028.

Pless, I., & Satterwhite, B. A measure of family functioning and its application. Social Science and Medicine,
1973, 7, 613-621.

Pless, I., Roghman, K., & Haggerty, R. Chronic illness,
family functioning, and psychological adjustment: A model for the allocation of preventive mental health
services. International Journal of Epidemiology, 1972,
1, 271-277.

Portner, J. Parent/adolescent relationships: Interaction
types and the circumplex model (University of Minnesota). Dissertation Abstracts International, 1982,
42(1), 4609-A.

Riskin, J., & Faunce, E.E. An evaluative review of family
interaction research. Family Process, 1972, 11, 365456,

Russell, C.S. Circumplex model of marital and family systems III: Empirical evaluation of families. Family
Process, 1979, 18, 29-45.

Satterwhite, B., Sqeig, S., Iker, H., & Pless, I. The family functioning index--five year test-retest reliability and implications for use. Journal of Comparative
Family Studies, 1976, 7, 111-116.

Sprenkle, D., & Olson, D. Circumplex model of marital systems III: Empirical study of clinic and non-clinic couples. Journal of Marriage and Family Counseling,
1978, 4, 59-74.