The Circumplex Model and head injury family types: a test of the balanced versus extreme hypotheses.

The long-term physical, cognitive, and psychosocial deficits that result from head injury affect an individual's interpersonal, emotional, and vocational functioning (Kay & Lezak, 1990). Consequently, the needs for care and assistance of persons who sustain head injuries and the costs of meeting such needs are substantial. Given that the responsibility for the lifelong well-being of persons with head injuries rests primarily on the individual's family (DeJong, Batavia, & Williams, 1990), head injury affects not only the person who sustains the injury, but the individual's entire family system.

Unfortunately, an extensive body of literature provides firm evidence that head injury hurts families socially, emotionally, and financially. Kozloff (1987) found that one of the most profound changes following a member's head injury was the social isolation of families. Jacobs (1988) reported that a recurring family worry was the chronic financial strain due to the person with the head injury not working and loss of income if a family member gives up work to supervise the person with the injury. Shock, disbelief, denial, and anger are common in relatives immediately following injury (Mathis, 1984). Problems with role functioning, marital relationships, and the physical health of individual family members become evident during the first year after injury (Livingston, 1990). As time goes by (i.e., from two to six years after injury), depression, poor communication, and drug and alcohol abuse are observed among family members. Finally, Thomsen (1984) concluded that family functioning is negatively affected by personality changes in the person with the injury and the day-to-day strain of meeting his or her needs for up to 15 years after injury.

Given the fact that head injury undoubtedly exerts a large and lasting toll on families, rehabilitation professionals working with families of persons with head injuries can benefit from information useful for planning family assessment and intervention and the development of long-term family support services (Miller, 1993; Rosen & Reynolds, 1994). As recommended by Kosciulek (1995), a useful starting point for designing services for families of people with head injuries is to understand underlying patterns of family functioning, or family types. Family types, defined as a "set of basic attributes about the family system which characterizes and explains how a family system typically, appraises, operates, and/or behaves" (McCubbin & McCubbin, 1991, p. 6), can play a vital role in head injury family assessment, intervention, and research. They may provide clues to rehabilitation professionals about how to best help families following a member's injury. Olson's (1993) Circumplex Model of Marital and Family Systems, a resource for assessment-based treatment planning with families experiencing stress, provides a clinical and theoretical framework for studying head injury family types.

Brief Overview of the Circumplex Model

The Circumplex Model of Marital and Family Systems bridges family theory, research, and practice. The central dimensions of family behavior integrated in the model are Cohesion and Adaptability (Olson, 1993). Family Cohesion is defined as "the emotional bonding that family members have toward one another" (Olson, Russell, & Sprenkle, 1983, p. 70). As shown in Figure 1, within the Circumplex Model there are four levels of Cohesion ranging from extremely low (disengaged), to moderate (separated, connected), to extremely high (enmeshed).

Family Adaptability is defined as "the ability of a marital or family system to change its power structure, role relationships, and relationship rules in response to situational and developmental stress" (Olson et al., 1983, p. 70). There are four levels of adaptability ranging from extremely low (rigid), to moderate (structured, flexible), to extremely high (chaotic). The Family Adaptability and Cohesion Evaluation Scale II (FACES II) (Olson, Bell, & Portner, 1990) is a family self-report instrument based on the Circumplex Model designed to assess family type. FACES II Cohesion and Adaptability scale scores enable clinicians and researchers to place families within the Balanced or Extreme family types of the model (McCubbin & McCubbin, 1991).

A central hypothesis derived from the Circumplex Model is that Balanced families will function more adequately than Extreme families. A number of studies provide empirical evidence to support this hypothesis. Portner (1981) compared 53 clinical families with 117 non-clinical families and found that the non-clinical families were moderate on cohesion and adaptability and that the clinical families scored significantly more often in the Extreme ranges of these dimensions. Garbarino, Sebes, and Schellenbach (1985) compared 27 high-risk families with 35 low-risk families on FACES II. As hypothesized, these researchers found that most low-risk families were Balanced, and that most of the high-risk families were Extreme types. Carnes (1987) studied the family of origin and the current family of sex offenders. Almost half (49%) of the families of origin and 66% of the current families were classified as Extreme types, but only 19% of families in a comparison control group were so classified. In contrast, 57% of the control group families were Balanced, and only 11% of the families of origin and 19% of the current families of the sex offenders were Balanced types.

Research is also available that supports the second primary Circumplex Model hypothesis: that Balanced families will have more positive communication skills than Extreme families. For example, investigating parent-adolescent communication patterns, Barnes and Olson (1985) found that Balanced families had significantly more effective communication skills than Extreme families. Similarly, in their study of mother-adolescent interaction comparing 29 mother-son dyads in which the son was an adolescent juvenile offender with 29 control dyads, Rodick, Henggler, and Hanson (1986) reported that mothers in the Balanced types exhibited significantly more supportive communication, explicit information, and positive affect than did mothers typed in the Extreme area. Olson (1993) concluded that, together, these studies provide strong empirical support for the hypotheses that Balanced types of families are more functional and have more effective communication than Extreme family types.

Purpose of the Study

A major task for rehabilitation professionals providing services to families following a member's head injury is to determine which elements of intervention are most appropriate and effective with which presenting family problems (Miller, 1993; Olson, 1993). While hypotheses have been tested relating family types described in the Circumplex Model to a variety of families (Olson, Russell, & Sprenkle, 1989), no research has been reported using the Circumplex Model that would help rehabilitation professionals understand the changes that take place in family functioning related to stress and crises following a member's head injury. Given its clinical and research utility, the Circumplex Model may prove valuable for guiding head injury family assessment, intervention, and research.

The purpose of this study was thus to assess the utility of the Circumplex Model of Marital and Family Systems for guiding family assessment and intervention following a member's head injury. It was anticipated that results would provide information useful for guiding assessment-based treatment planning and future head injury family research. To this end, the following two research hypotheses were tested: (a) Balanced head injury families will function more adequately than Extreme head injury families, and (b) Balanced head injury families will have more positive communication skills than Extreme head injury families.

Method

Participants

The population of interest in this study was families with a member with a head injury. The sample was drawn from the family support groups of the Georgia Head Injury Association (GHIA), the Georgia Chapter of the National Head Injury Foundation (NHIF). The sampling procedure required that the family member who was the primary caregiver for the person with the head injury complete study materials, that families were English-speaking, and that the family member with the head injury was at least one year post-injury and 18 years of age. A total of 82 families met the criteria for inclusion in the final sample.

The participants (i.e., primary caregivers) were predominantly mothers (61%) and wives (22%) whose ages ranged from 20 to 71 with a mean of 46.7 years and standard deviation of 13.5 years. Ninety percent of the participants were White and 10% were Black. The majority of participants were married (80%), Protestant (74%), and employed full-time in professional or semi-professional occupations. Annual family income ranged from $8,000 to $100,000 with a median of $30,747. Participants represented urban, suburban, rural, and semi-rural regions of Georgia.

Instruments

The first instrument used in this study was a Family Information Sheet (FIS) developed to collect information regarding a variety of family demographic variables (e.g., age, gender, family income). Data from the FIS was used to describe participant characteristics.

The second instrument was the Family Adaptability and Cohesion Evaluation Scales II (FACES II) (Olson, Bell, & Portner, 1990). FACES II was designed to enable clinicians and researchers to place families within the Balanced or Extreme family types of the Circumplex Model (McCubbin & McCubbin, 1991). Respondents are asked to read 30 statements with response categories of "Almost Never," "Once in Awhile," "Sometimes," "Frequently," and "Almost Always," and decide for each one how frequently the described behavior occurs in his/her family. The categories are given values of 1 through 5, respectively. Two scales, Cohesion (16 items) and Adaptability (14 items), are scored using procedures developed by Olson and Tiesel (1991). Based on these scores a family is then classified as a Balanced or Extreme type. In this study, FACES II provided a measure of head injury family type.

Adequate FACES II internal consistency reliability has been demonstrated. Olson (1989) reported Cronbach's alphas (Cronbach, 1951) of .91 and .80, respectively, for the Cohesion and Adaptability scales. Four-week test-retest reliability of FACES II was .84 for Cohesion and .80 for Adaptability (McCubbin & McCubbin, 1991). The internal consistency reliability coefficients for the Cohesion and Adaptability scales in the present study were .86 and .78, respectively.

The third instrument used in this study was the Family Assessment Device (FAD) (Epstein, Baldwin, & Bishop, 1983), a 60-item questionnaire designed to assess the following six dimensions of family functioning: Problem Solving, Communication, Roles, Affective Involvement, Affective Responsiveness, and Behavior Control. Additionally, an overall assessment of family functioning (General Functioning) is also included.

The FAD produces scores ranging from 1 to 4 for all items. Item scores are then summed and divided by the number of items completed to produce scale scores that range from 1 (healthy functioning) to 4 (unhealthy functioning). In this study, the FAD General Functioning scale provided a measure of adequacy of family functioning (i.e., overall family health/pathology) and the Communication scale provided a measure of family communication skills (i.e., effectiveness, clarity, and directness of information exchange in the family). The FAD scales have good reported internal consistency reliabilities ranging from .72 to .92 (Epstein et al., 1983). In addition, FAD test-retest reliability has been found to be adequate (Miller, Bishop, Epstein, & Keitner, 1985).

Research also supports the validity of the FAD. Using discriminant analysis, Epstein et al. (1983) accurately classified 67 percent of nonclinical and 64 percent of clinical families into their appropriate groups based on FAD scores. In a test of predictive evidence for validity, these same researchers found that the FAD was a more powerful predictor of morale and health in older couples than the Locke-Wallace Marital Satisfaction Scale (Locke & Wallace, 1959).

Procedure

Participants were recruited from the head injury support groups affiliated with the GHIA as part of a comprehensive study of family adaptation to head injury. Of these 26 groups, five were excluded because they focused on serving individuals with head injuries rather than families. Thus, the 21 remaining GHIA family support groups were involved in the study. Each support group facilitator was contacted by telephone to obtain assistance in enlisting family participation. To guarantee confidentiality of family names and addresses, the facilitators were provided materials which they mailed to families on their group membership lists as of February 1994.

Families were mailed a FIS along with a cover letter emphasizing the significance of the study and assuring confidentiality. Families who expressed an initial interest in the study, as indicated by completion and return of the FIS, were then mailed the FACES II and FAD. To maximize the total response rate, a follow-up reminder postcard was sent to those participants who had not returned the FACES II and FAD within two weeks. Data were collected from March 1994 through May 1994.

A total of 338 families were mailed introductory study materials. One-hundred twenty-five families completed and returned the FIS, yielding a 37% response rate. Of the questionnaires then sent, 97, or 78%, were completed and returned to the investigator. Fifteen data sets were not included in the final data analysis because they did not meet sample requirements. Five were from families whose member with the head injury was under 18 years of age, three were from families whose member with the injury was less than one year post-injury, and questionnaires from seven families were missing a significant amount of data and were thus deemed not usable. Therefore, a total of 82 questionnaires from families meeting sample criteria were sufficiently complete to be included in the study.

Research Design and Data Analysis

The purpose of this study was to assess the utility of the Circumplex Model of Marital and Family Systems for guiding family assessment and intervention following a member's head injury. To this end, the following statistical hypotheses were tested: (a) [H.sub.0]: There is no significant difference between FAD General Functioning scores of Balanced and Extreme head injury families ([u.sub.1] = [u.sub.2]) and (b) [H.sub.0]: There is no significant difference between FAD Communication scores of Balanced and Extreme head injury families ([u.sub.1] = [u.sub.2]). An ex post facto research design was required for assessing the relationship between Balanced and Extreme head injury family types (Cozby, 1993). In accordance with the purpose of the investigation and the nature of the ex post facto research design, t-tests for a difference between two independent groups were used to test the two primary Circumplex Model hypotheses (Saxon, Alston, & Holbert, 1994).

As recommended by Cohen (1988) and Lipsey (1990), a statistical power estimate using Borenstein and Cohen's (1988) computer program was conducted before the t-tests to determine the appropriate alpha level for hypothesis testing under conditions of a fixed sample size of 82. Given an alpha level of .05 and assuming a small effect size, the pre-analysis power estimate yielded a power of .77 for testing the research hypotheses using independent sample t-tests. A statistical power of approximately .80 would be considered adequate for rejecting a null hypothesis if it were false (Cohen, 1988). The t-tests were therefore planned at the .05 alpha level.

Results

As shown in Table 1, the t-tests for a difference between two independent groups (i.e., Balanced and Extreme families) both produced significant results. The null hypotheses that there were no significant differences between FAD General Functioning and Communication scores of Balanced and Extreme head injury families were both rejected. Balanced head injury families scored significantly lower than Extreme families on both the General Functioning and Communication scales, indicating that they were functioning more effectively on these two dimensions (Epstein, Baldwin, & Bishop, 1983). These results thus lend support to the alternate hypotheses that Balanced head injury families function more adequately and have more positive communication skills than Extreme head injury families.

Discussion

The central purpose of the current study was to assess the utility of the Circumplex Model of Marital and Family Systems (Olson, 1993) for guiding family assessment and intervention following a member's head injury. It was anticipated that results would provide information useful for guiding assessment-based treatment planning and future head injury family research. Results provide support for the two following research hypotheses: (a) Balanced head injury families will function more adequately than Extreme head injury families, and (b) Balanced head injury families will have more positive communication skills than Extreme head injury families. These findings thus provide initial evidence to support the validity of the Circumplex Model as a framework useful for assessment and intervention with families of people with head injuries.

Prior to discussing implications for head injury family intervention and research, several caveats regarding the limitations of this study should be noted. The first relates to the level of measurement. While the level of assessment was at the individual level (i.e., primary caregivers), the level of construct formation was at the family level (i.e., family types). Copeland and White (1991) recommend that researchers acknowledge such mixed conceptual levels when reporting family study results.

The second caveat relates to external validity, or the degree to which the findings of this investigation can be generalized across persons, times, and settings (Parker, 1993). The sampling plan in this study was non-probability in nature. Generalizing results to all families in Georgia who have a member with a head injury, or even to all families who participate in support groups affiliated with the GHIA, would thus be inappropriate. In addition, the cross-sectional nature of data collection procedures allow only for interpretation of results concerning sample families at the time of the study.

The final limitation concerns research design. Due to the nature of the ex post facto research design, no causal attributions can be made regarding the relationship between family types and the adequacy of general family functioning and family communication skills. An experimental research design is necessary to begin to establish cause-effect relationships.

Table 1

Mean FAD General Functioning and Communication Scores of Balanced
and Extreme Head Injury Families

                          Balanced      Extreme

                          (n = 30)      (n = 52)

FAD Scale                M       SD     M      SD       t

General Functioning     1.95    .31    3.15    .44    2.03(*)

Communication           1.81    .28    2.94    .39    2.11(*)

Note. Degrees of freedom for the t tests were 80.

* p [less than] .05.

Head Injury Family Intervention Using the Circumplex Model

The literature clearly documents that families experience extensive behavioral and psychological burden as a result of traumatic head injuries (Miller, 1993). The fact that the burden on families persists for years has also become increasingly clear (Livingston, 1990; Thomsen, 1984). Furthermore, the family must often assume the major responsibility for the injured person's long-term care. As a result, parents, spouses, children, and siblings of persons with head injuries may present difficulties to rehabilitation professionals related to behavioral problems in their member with the injury, depletion of family finances, marital discord, and family social isolation (Jacobs, 1988; Williams, 1991).

Given this data, a major challenge to head injury rehabilitation practitioners is to determine which elements of intervention are most appropriate and effective with which presenting problems and with which elements of family functioning. The Circumplex Model, a resource for assessment-based treatment planning with families experiencing major stressors such as head injury, provides a guide for accomplishing this task (Thomas & Olson, 1993).

The model provides a conceptual framework and empirically-based instrument (i.e., FACES II) for assessing head injury family system functioning on two fundamental dimensions of family organization, Cohesion and Adaptability. This descriptive typology of transactional patterns can be used to determine a family's current level and style of functioning on each dimension and to guide treatment planning to strengthen particular components of functioning toward clearly specified and realistic objectives (Thomas & Olson, 1993). Thus, as described below, the most efficacious family intervention following a member's head injury is not limited to the reduction or interruption of extreme dysfunctional patterns, but should also be directed systematically toward the promotion of more functional family interaction patterns.

Moving families from rigid to structured systems (Adaptability dimension). In families with extremely rigid inter-actional patterns, leadership tends to be authoritarian, with one or both parents highly controlling. Discipline is typically autocratic, based on a simplistic principle of "law and order," and consequences are strict, even harsh, without leniency. As illustrated in Figure 1, Olson (1993) proposed that counseling with rigid families target interventions to shift their organization to the structured range on the family Adaptability dimension. The chief therapeutic task with rigid head injury families is to promote the interaction flexibility that is lacking.

Moving families from chaotic to flexible systems (Adaptability dimension). Families at the chaotic extreme on the Adaptability dimension are characterized by a lack of structure, order, and predictability in their interactions. Family counseling with chaotic head injury families should be aimed to shift their organizational pattern toward the flexible range on the family adaptability dimension. The chief therapeutic task with chaotic families is building structure (Olson, 1993).

Moving families from disengaged to separated systems (Cohesion dimension). A disengaged relationship is characterized by extreme emotional and physical separateness. The paradigmatic rule in disengaged families tends to be everyone for him or herself. There is little interest, involvement, or sense of mutual caring between or among family members (Olson, 1993). Intervention with disengaged head injury family systems should aim toward the separated level of family Cohesion. A chief therapeutic goal with disengaged families is to promote connectedness through shared problem solving.

Moving families from enmeshed to connected systems (Cohesion dimension). The fourth area depicted in Figure 1 focuses on moving families from the enmeshed to the connected level of the Cohesion dimension. Enmeshed relationships are characterized by an extreme amount of emotional closeness and demands for loyalty and consensus. An "All for one and one for all" philosophy permeates enmeshed family structures. Unlike disengaged families, enmeshed head injury families are likely to come together for counseling, even when individual sessions are scheduled. Further, they may experience difficulty separating from the rehabilitation program and individual practitioners at the termination of their injured member's program.

Clinical head injury family intervention using the Circumplex Model can help rehabilitation professionals move beyond a limited focus on the present problems to the clinically relevant dynamics of family Cohesion and Adaptability. Treatment goals based on the model should focus on moving families extreme on one or both dimensions to more balanced levels of functioning. Given the long-term effect of head injury on families, a metagoal of therapy is to help families better learn how to negotiate system change over time (Miller, 1993; Olson, 1993; Thomas & Olson, 1993). Families need to learn how to balance their separateness versus togetherness on Cohesion, and their stability versus change on Adaptability, to remain functional as a system in the years following a member's head injury.

Future Head Injury Family Research Using the Circumplex Model

The Circumplex Model can also be used to guide head injury family research. Future studies should focus on identifying critical family variables that differentiate Balanced and Extreme family types. If such factors can be identified in larger, more rigorously designed research, intervention strategies to support families of persons with head injuries could then be more effectively designed and targeted. It may also be instructive to examine the influence of characteristics of the member with the head injury and family life cycle stage on family functioning patterns (Miller, 1993; Sachs, 1991). In addition, in the current study, adequate FACES II internal consistency reliability was established in a sample of families of people with head injuries. Following development of additional reliability and validity data with head injury families, the FACES II may prove to be a useful screening tool for assessing family type and vulnerability to maladaptation following head injury. In this manner, information gained from empirically-derived head injury family types based on the Circumplex Model may add to our understanding of the impact of head injury on families and help to explain how different family types are linked to different family outcomes.

Acknowledgement

The author expresses gratitude to Luz Hamilton for her assistance with the preparation of this article.

References

Barnes, H., & Olson, D. H. (1985). Parent adolescent communication and the Circumplex model. Child Development, 56, 438-447.

Borenstein, M., & Cohen, J. (1988). Statistical power analysis: A computer program. Hillsdale, NJ: Erlbaum.

Carnes, P. (1987). Counseling sexual abusers. Minneapolis, MN: CompCare.

Cohen, J. (1988). Statistical power analysis for the behavioral sciences (2nd ed.). Hillsdale, NJ: Erlbaum.

Copeland, A. P., & White, K. M. (1991). Studying families. Newbury Park, CA: Sage.

Cozby, P. C. (1993). Methods in behavioral research (5th ed.). Mountain View, CA: Mayfield.

Cronbach, L. J. (1951). Coefficient alpha and the internal structure of tests. Psychometrika, 16, 297-334.

DeJong, G., Batavia, A. I., & Williams, J. M. (1990). Who is responsible for the lifelong well-being of a person with a head injury? Journal of Head Trauma Rehabilitation, 5, 9-22.

Epstein, N. B., Baldwin, L., & Bishop, D. (1983). The McMaster Family assessment device. Journal of Marital and Family Therapy, 9, 171-180.

Garbarino, J., Sebes, J., & Schellenbach, C. (1985). Families at risk for destructive parent-child relations in adolescence. Child Development, 55, 174-183.

Jacobs, H. E. (1988). The Los Angeles head injury survey: Procedures and initial findings. Archives of Physical Medicine and Rehabilitation, 69, 425-431.

Kay, T., & Lezak, M. (1990). The nature of head injury. In D. W. Corthell (Ed.), Traumatic brain injury and vocational rehabilitation (pp. 21-65). Menomonie, WI: University of Wisconsin-Stout, Research and Training Center.

Kosciulek, J. F. (1995). On identifying head injury family types. Journal of Applied Rehabilitation Counseling, 26(3), 13-18.

Kozloff, R. (1987). Networks of social support and the outcome from severe head injury. Journal of Head Trauma Rehabilitation, 2 (3), 14-23.

Lipsey, M. W. (1990). Design sensitivity: Statistical power for experimental research. Newbury Park, CA: Sage.

Livingston, M. G. (1990). Effects on the family system. In M. Rosenthal, E. R. Griffith, M. R. Bond, & J. D. Miller (Eds.), Rehabilitation of the adult and child with traumatic brain injury (2nd ed., pp. 225-235). Philadelphia: Davis.

Locke, H. J., & Wallace, K. M. (1959). Short marital adjustment and prediction tests: Their reliability and validity. Marriage and Family Living, 21, 251-255.

Mathis, M. (1984). Personal needs of family members of critically ill patients with and without acute brain injury. Journal of Neurosurgical Nursing, 16, 36-44.

McCubbin, M. A., & McCubbin, H. I. (1991). Family stress theory and assessment: The Resiliency Model of family stress, adjustment, and adaptation. In H. I. McCubbin & A. I. Thompson (Eds.), Family assessment inventories for research and practice (pp. 3-32). Madison, WI: University of Wisconsin-Madison, Family Stress, Coping, and Health Project.

Miller, I. W., Bishop, D. S., Epstein, N. B., & Keitner, G. I. (1985). The McMaster family assessment device: reliability and validity. Journal of Marital and Family Therapy, 11, 345-356.

Miller, L. (1993). Family therapy of brain injury: Syndromes, strategies, and solutions. The American Journal of Family Therapy, 21, 111-121.

Olson, D. H. (1989). Circumplex model and family systems VII: Family assessment and intervention. In D. H. Olson, C. S. Russell, & D. H. Sprenkle (Eds.) Circumplex model: Systemic assessment and treatment of families (3rd ed., pp. 7-49). New York: Haworth.

Olson, D. H. (1993). Circumplex model of marital and family systems. In F. Walsh (Ed.), Normal family processes (3rd ed., pp. 104-137). New York: Guilford.

Olson, D. H., Bell, R., & Portner, J. (1990). Family adaptability and cohesion evaluation scales II (FACES II). St. Paul, MN: University of Minnesota, Family Social Science.

Olson, D. H., Russell, C., & Sprenkle, D. H. (Eds.). (1983). Circumplex model: Systemic assessment and treatment of families (2nd ed.). New York: Haworth.

Olson, D. H., Russell, C., & Sprenkle, D. H. (Eds.). (1989). Circumplex model: Systemic assessment and treatment of families (3rd ed.). New York: Haworth.

Olson, D. H., & Tiesel, J. (1991). Family adaptability and cohesion evaluation scales II (FACES II): Linear scoring & interpretation. St. Paul, MN: University of Minnesota, Family Social Science.

Parker, R. M. (1993). Threats to the validity of research. [editorial]. Rehabilitation Counseling Bulletin, 36, 130-138.

Portner, J. (1981). Parent-adolescent interactions of families in treatment. Unpublished doctoral dissertation, University of Minnesota, St. Paul.

Rodick, J. D., Henggler, S. W., & Hanson, C. L. (1986). An evaluation of the Family Adaptability and Cohesion Evaluation Scales (FACES) and the Circumplex model. Journal of Abnormal Child Psychology, 14, 77-87.

Rosen, B., & Reynolds, W. E. (1994). The impact of public policy on persons with traumatic brain injury and their families. Journal of Head Trauma Rehabilitation, 9 (2), 1-11.

Saxon, J. P., Alston, P. P., & Tolbert, D. (1994). Principles for research in rehabilitation. Athens, GA: Elliott & Fitzpatrick.

Thomas, V., & Olson, D. H. (1993). Problem families and the Circumplex model: Observational assessment using the Clinical Rating Scale (CRS). Journal of Marital and Family Therapy, 19, 159-175.

Thomsen, I. V. (1984). Late outcome of very severe head trauma: A 10-15 year second follow-up. Journal of Neurology, Neurosurgery, and Psychiatry, 47, 260-268.

Williams, J. M. (1991). Family reaction to head injury. In J. M. Williams & T. Kay (Eds.), Head injury: A family matter (pp. 81-99). Baltimore: Paul H. Brookes.