Abstract
The purpose of the current study was to develop and test the psychometric properties of the Family Genogram Interview (FGI), a standardized genogram format theoretically grounded in Bowen Family Systems Theory. The sample consisted of 50 heterosexual, married women who had at least one child between 4 and 10 years of age, who completed the FGI with a trained interviewer. The results indicate the FGI has good interinterviewer reliability while construct validity assessments yielded mixed results. The results indicate the FGI is a positive step toward standardization and advancement of genogram interviewing in research and clinical applications. Limitations and directions for future studies are discussed.
The Family Genogram Interview (FGI) is the most widely used means of assessing relationship processes in nuclear and extended family systems (McGoldrick & Gerson, 1985; McGoldrick, Gerson, & Petry, 2008). A genogram is a pictorial diagram of a family using data gathered during a semistructured interview to assess for various elements of family functioning, such as conflict, cutoff, enmeshment, and other family relationship patterns. A genogram resembles a “family tree” however the information gathered includes both the simple structure of the family and the emotional patterns and processes present in the nuclear and extended family system. In a typical genogram, circles represent females and squares represent males. Information is often gathered for at least three family generations (McGoldrick et al., 2008). Clinicians typically gather information about the structure of the family (i.e., when individuals were married or how many children each couple had) and family relationships (i.e., Are there any relationships with conflict or cutoff?). While the genogram interview as it is currently used has widespread popularity in clinical and research applications, it also has several drawbacks including limited standardization and a limited theoretical foundation (Coupland, Serovich & Glenn, 1995). The purpose of the current study was to develop a standardized genogram format theoretically grounded in Bowen family systems theory.
Limited Standardization of Genogram Interviewing
The first limitation in genogram interviewing is the unstructured nature of the genogram as a clinical interview tool. Most genogram assessments currently used are based on the writings of Monica McGoldrick and her colleagues (McGoldrick & Gerson, 1985; McGoldrick, Gerson, & Shellenberger, 1999; McGoldrick et al., 2008), and most clinicians tend to rely on their guidelines for conducting a genogram interview (Nichols & Schwartz, 2001). Despite the widespread popularity and acceptance of the family genogram, limited agreement exists about its specific format. While McGoldrick et al. (1999) and McGoldrick et al. (2008) standardized several basic elements of the genogram, they did not provide guidelines for more in-depth, standardized assessment of the emotional and relationship process in families. Specifically, McGoldrick et al.’s (2008) format primarily offers suggestions for how a family’s structural information should be arranged, and which type of symbols to be used, but does not offer a standardized interview protocol to follow. In clinical application, this limits the reliability of genogram as a family assessment tool across clinicians and across clients as often different information is obtained, using different questions in each interview.
The unstructured nature of genogram interviewing also prevents researchers from rigorously testing theoretical and clinical assumptions about family functioning using a genogram assessment (Coupland et al., 1995). Development of a more structured genogram interview—with clear operationalizations of family emotional processes that are replicable across interviewers, within a study, and across studies—would facilitate more empirical research on multigenerational family processes and their implications for individual or family functioning.
Although the research investigating the reliability of genogram interviewing as it is currently conducted in the field is limited, two previous studies confirm the need for greater structure built into genogram interviewing in research and clinical settings. For example, among marriage and family doctoral students, Coupland, Serovich, and Glenn (1995) found the reliability of relationship information obtained by different clinicians was quite low. Further, Rohrbaugh, Rogers, and McGoldrick (1992) investigated whether a mixed sample of expert therapists and family physicians trained in genogram administration could achieve acceptable levels of interinterviewer reliability on several dimensions. While the clinicians were able to come to moderate agreement on the general categories of information that should be included in a genogram (i.e., r = .75 or above), they were unable to consistently agree on the level of importance of different types of information in specific case examples with correlations all below .45 between raters (Rohrbaugh, Rogers, & McGoldrick, 1992). Given the limited reliability in genogram administration, both studies highlight the need for a standardized genogram format for use in clinical and research application.
The second limitation of how genograms are currently used is evidenced by the proliferation of different themed genograms such as a “scripto-trauma” genogram (Jordan, 2006), a career genogram (Chope, 2005), and a brief genogram interview to be used with families experiencing postpartum depression (Holtslander, 2005). While these types of themed genograms may have utility in different types of specific client situations the variety of genograms do not lend themselves to widespread research on family functioning because results across studies are difficult to compare if the genograms used in the studies are significantly different.
Taken together, the lack of agreement about the essential elements needed in a genogram interview and the lack of a consistent theoretical foundation for genogram assessment items hinder the reliability of genogram-based family assessment and research (Coupland et al., 1995). Given the widespread popularity of genogram interviewing, there is a need to improve the standardization and theoretical foundation of how this type of family assessment is conducted both for research and application in couple and family therapy.
To address these standardization and theoretical limitations, the purpose of the study was to develop a standardized genogram interview protocol, the FGI. The goal in creating the FGI was to: (a) improve the standardization of genogram assessment (i.e., reliability and validity) and (b) to improve the theoretical grounding of genogram assessment. To specifically address the need for a theoretically grounded genogram interview, the FGI was developed using the family evaluation diagram described in Bowen family systems theory by Kerr and Bowen (1988) while retaining portions of the traditional genogram format from McGoldrick and Gerson (1985) and McGoldrick et al. (2008) as a foundation for standardizing the FGI format. The FGI was designed to specifically assess and operationalize four emotional processes thought to represent the nuclear family emotional process as described in Bowen theory.
Bowen Family Systems Theory
According to Bowen Family Systems Theory (Bowen, 1978; Kerr & Bowen, 1988), the concept of the nuclear family emotional process describes how basic relationship patterns develop in family systems and often represent where problems or symptoms develop in a family. Within the nuclear family emotional process are four mechanisms by which families respond and adapt to or bind relationship tension, anxiety, or stressors. These four processes consist of: (a) marital conflict, (b) emotional cutoff, (c) symptoms in a spouse, and (d) focus on a child (Kerr & Bowen, 1988).
In the current study, the four family processes outlined in Bowen theory’s definition of the nuclear family emotional process were selected for inclusion in the FGI. Kerr and Bowen (1988) described how these four processes represent the maladaptive ways in which family systems respond to stress, anxiety, and systemic changes. Clinical problems or symptoms usually develop during periods of heightened and prolonged family tension. The level of tension depends on the amount of stressors for a family, how a family adapts to the stress, and the quality of a family’s connection with extended family and social networks. Anxiety increases the activity of one or more of these four relationship patterns. The basic relationship patterns result in family anxiety being focused on or absorbed by certain parts of the family. The more anxiety one person or one relationship absorbs, the less other people in the system must contend with (Bowen, 1978; Kerr & Bowen, 1988).
Kerr described how these four processes are of particular importance to include in a genogram assessment as they are thought to represent important indicators of systemic functioning in families. Specifically, Kerr described how the family evaluation interview and resulting family diagram should represent these emotional processes occurring in the nuclear and extended family (Kerr & Bowen, 1988). However, no prior genogram formats have operationalized these concepts as part of family assessment.
Further, Bowen theory suggests that emotional processes in families are influenced by what is termed differentiation of self (Bowen, 1976, 1978; Kerr & Bowen, 1988). Differentiation of self is thought to influence the extent of marital conflict, emotional cutoff, symptoms in a spouse, and focus on a child in the family emotional system and is critical to mature development and psychological health (Kerr & Bowen, 1988). Differentiation can be thought of on an intrapsychic level as the ability to separate thoughts from feelings, and the ability on an interpersonal level to balance intimacy and autonomy with others.
Research Questions
Given the need for a standardized, theoretically grounded genogram interview two main research questions were investigated using the newly created FGI:
What is the interviewer and subscale item reliability of the FGI? What is the validity of the FGI subscales?
These research questions were examined in two stages. In Stage 1, the FGI was created and developed in an effort to operationalize four Bowen theory family processes (marital conflict, emotional cutoff, symptoms in a spouse, and focus on a child) thought to be indicators of the level of differentiation in family systems. Items were written to create the FGI, and then subscale psychometrics were evaluated to address Research Question 1. The reliability of participant responses on the individual FGI items were assessed based on three criteria: (a) interinterviewer reliability, (b) statistically significant differences in participant responses across different interviews, and (c) subscale reliabilities. The internal consistency reliability of the participant scores on the FGI subscales was also calculated. Based on these analyses, items were deleted and the final version of the FGI was established and used in subsequent analyses in Stage 2.
Stage 2 of the study addressed Research Question 2 using the final version of the FGI. Here, the aims were to report interinterviewer consistency in the FGI administration, and its validity evaluated by examining associations between FGI subscale scores and four existing psychometrically sound measures of similar constructs.
Hypotheses
FGI-marital conflict (FGI-MC) scores, as measured in the nuclear family, were expected to positively correlate with an external measure of distress or discord in marital relationships (Hypothesis 1). FGI-emotional cutoff (FGI-EC) scores, as measured with the family of origin relationships, were expected to positively correlate with a corresponding measure that assesses emotional cutoff in interpersonal relationships (Hypothesis 2). FGI-symptoms in a spouse (FGI-SS) scores, as measured in the nuclear family, were expected to positively correlate with participants rating of self and spouse on a measure of psychological symptoms, social-role functioning, and interpersonal difficulties (Hypothesis 3).
Method
Participants
The sample consisted of 50 heterosexual, married women who had at least one child between 4 and 10 years of age. Participants were on average 37.4 years of age (SD = 4.35, range = 28–47) with an average of 2.62 children (SD = .87, range = 1–5). Participating women were primarily middle to upper-middle class, with a median yearly household income of $70,000 (range = $20,000–100,000). The participants were also relatively well educated, with 74% having at least a bachelor’s level of education. In terms of ethnicity, 98% of the sample was White, and 2% identified as biracial. The women had been married an average of 11.9 years (SD = 3.90, range = 3–24 years).
Mothers selected to participate came from two-parent, heterosexual nuclear families. These criteria were selected in order to maximize variance associated with three of the constructs in the study, marital conflict, symptom formation in a spouse, and focus on a child as Bowen theory most clearly defines these family emotional processes as they are found in heterosexual, married couples (Kerr & Bowen, 1988). The sample consisted of married mothers who were in the life cycle stage of “families with young children” as defined by Carter and McGoldrick (1999) as it is thought that the family emotional processes often become heightened and more pronounced as the anxiety and stress in the relationships increases with the addition of young children (Kerr & Bowen, 1988).
Measures
FGI
The purpose of the study was to develop a standardized genogram interview protocol, the FGI, designed from both the family evaluation diagram described by Kerr and Bowen (1988) and the traditional genogram format (McGoldrick & Gerson, 1985; McGoldrick et al., 1999) to operationalize the four emotional processes in families across both nuclear and multigenerational family systems. Major portions of the FGI are grounded in the traditional genogram format described by McGoldrick and Gerson (1985). In a traditional genogram format, specific symbols are used to denote factual and relationship information across generations of a family system. Like the traditional genogram format, the FGI also includes information about family structure and important nodal events (i.e., births, deaths, marriages, divorces, geographic moves, etc.) across three generations.
Second, the FGI was designed based on the family evaluation process outlined by Kerr and Bowen (1988) and grounded in Bowen theory (Bowen, 1978). Specifically, the FGI was designed to assess nuclear family emotional processes. The complete FGI contains 68 quantitative questions and 84 qualitative questions that assess both the current nuclear family functioning (i.e., spouses and children) and family of origin relationships. The 75-min FGI protocol was designed to assess this relationship information about the four family emotional processes thought to be indicators of the level of differentiation in a family system: (a) FGI-MC, (b) FGI-EC, (c) FGI-SS, and (d) FGI-focus on a child (FGI-FC). These processes are then diagramed in a standard genogram format. The quantitative questions in the FGI provide a measureable means of assessing family functioning. The qualitative questions in the FGI are intended to gather more rich and complex information from participants to give context or elaborate on family processes.
The development of the FGI items was aided by several sources. First, a major influence in question development came from definitions of the family emotional processes as described by Bowen (1976, 1978) and Kerr and Bowen (1988). Questions for each construct in the FGI were created using the theoretical definitions and descriptions of how these processes are thought to occur in family systems. Further, Kerr and Bowen (1988) offered some initial guidance how to assess for these emotional processes in his description of the family diagramming process.
Question development also involved input from experienced family therapists and a preliminary, informal pilot study. First, five experienced family therapists who work from a Bowen theory orientation audiotaped sessions with clients while conducting a genogram interview. Based on the questions asked of clients in the audiotapes, items for the FGI were developed. Input on items was also obtained from a student research team in this stage of development.
Once a set of preliminary FGI items were created a group of clinicians, who were also part of a student research team, informally participated in a pilot study by administering the FGI with their colleagues and acquaintances to provide feedback on the protocol. Based on feedback from this initial pilot study, the FGI items were revised and the original version used in the current study was developed.
While the FGI consists of both quantitative and qualitative items, only quantitative items were used in the current study to aid in consistency of measurement. Also, though the subscales on the FGI are designed to assess both nuclear family (i.e., husband, wife, and children) and family of origin (i.e., parents and siblings), the current analyses targeted either the nuclear family or family of origin, but not both. For example, in the current analyses, nuclear family (i.e., husband, wife, and children) items in the FGI-MC, FGI-SS, and FGI-FC subscales, and not family of origin items (i.e., parents and siblings) were used. For the FGI-EC subscale, only quantitative items assessing cutoff with family of origin were included in the analysis. No qualitative, open-ended items from any of the four subscales were used in the current analyses.
Marital conflict
Each family process in the FGI is operationalized based on Kerr and Bowen’s (1988) description. From a Bowen theory perspective, marital conflict serves to bind the anxiety in the marital dyad. Bowen described how less differentiated couples’ individuality is less well developed and togetherness needs are stronger creating a situation where emotional reactivity is more intense and more easily triggered. The negative or anxious emotional reaction of one partner triggers the same from the other partner, often leading to chronic marital conflict. The subscale marital conflict is operationalized using questions about the duration, severity, and typical patterns of marital conflict in both the nuclear family and the family of origin based on this theoretical description. In this way global marital conflict in couples was assessed. Example quantitative questions used in the current analyses include:
How much or often do you and your partner or spouse have disagreement or conflict? How often do you feel the conflict gets resolved in a way that is mutually satisfying?
The FGI-MC scale consisted of a total of 9 quantitative and 31 qualitative questions assessing both nuclear family and family of origin. However, only the six quantitative, Likert-type scale items assessing the nuclear family were used in the final analyses for the current study. To ensure consistency in measurement, the qualitative items were excluded from subsequent analyses. Additionally, quantitative items assessing family of origin were not included in the final analysis for the current study as Bowen theory suggests that marital conflict in the nuclear family (i.e., current spouses) is often one of the most salient aspects of poor family functioning (Kerr & Bowen, 1988). Subsequent reliability analyses indicated the 6 items retained in the FGI-MC demonstrated good internal reliability (see Tables 2 and 3).
Original FGI Subscale Reliability
Individual Items on the FGI that Did Not Meet the Reliability Criteria 2 and 3
Final FGI Subscales Used in Subsequent Analyses
Emotional cutoff
Second, FGI-EC, is focused on information about the duration, severity, and systemic effects of emotional and physical withdraw between spouses in the nuclear family and among self and members of one’s family of origin. Bowen suggested emotional cutoff may also occur as a means of managing the emotional intensity and anxiety in relationships (Kerr & Bowen, 1988; Titelman, 1998). Emotional cutoff occurs when an individual experiences difficulty managing the emotional intensity in relationships and reduces his or her emotional and/or physical contact with others in response. The emotional cutoff subscale is operationalized using questions about the quantity and quality of contact with both nuclear family and family of origin. For emotional cutoff in relationships within the family of origin, the interviewees were asked specific questions about their relationships with their mothers, fathers, and siblings. This information was used in the current analyses. For example:
On a scale from 1 to 5 how emotionally distant would you say the relationship is? How many people do you have contact with?
The FGI-EC scale consisted of a total of 20 quantitative and 14 qualitative questions assessing both nuclear family and family of origin. However, only the 11 quantitative, (9 Likert-type scale and 2 continuously scored) items assessing the participant’s relationships with family of origin were used in the final analyses for the current study. To ensure consistency in measurement, the qualitative items were excluded from subsequent analyses. Additionally, quantitative items assessing emotional cutoff in the nuclear family were not included in the final analysis for the current study as Bowen theory suggests that emotional cutoff from one’s family of origin is the most clear marker of family dysfunction to assess in a genogram (Kerr & Bowen, 1988). Subsequent reliability analyses indicated the 11 items retained in the FGI-EC demonstrated good internal reliability (see Tables 2 and 3).
Symptoms in a spouse
According to Bowen theory, a poor functioning emotional system can trigger the development of symptoms in one of its members (Bowen, 1978; Kerr & Bowen, 1988). If the system is unbalanced to the point a person feels emotionally isolated or chronically anxious, he or she is vulnerable to dysfunction and the formation of symptoms. Symptoms can take the form of physical, psychological, and/or social symptoms. Items that comprise the symptoms in a spouse subscale focus on the duration and severity of physical, mental, or social symptoms in self and one’s spouse, as well in family of origin. On the FGI-SS the respondent is asked to rate both herself and her spouse on this subscale. Some example questions used in the current analyses include:
How would you rate your physical health? How would you rate your partner or spouse’s physical health?
The FGI-SS scale consisted of a total of 31 quantitative and 23 qualitative questions assessing both nuclear family and family of origin. However, only the 12 quantitative, (9 Likert-type scale and 3 categorical, yes or no) items assessing the nuclear family were used in the final analyses for the current study. To ensure consistency in measurement, the qualitative items were excluded from subsequent analyses. Additionally, quantitative items assessing family of origin were not included in the final analysis for the current study as Bowen theory suggests that symptoms in a spouse in the nuclear family (i.e.. current spouses) is often one of the most common presenting concerns in marital counseling and is a salient aspect of poor family functioning (Kerr & Bowen, 1988). Subsequent reliability analyses indicated the 12 items retained in the FGI-SS demonstrated good internal reliability (see Tables 2 and 3).
Focus on a child
Fourth, the FGI was also designed to operationalize the concept of focusing on a child as a family emotional process. Bowen theory defines focus on a child as a process in which a significant amount of the family anxiety is shifted to one or more children (Kerr & Bowen, 1988). This focus can include overprotective behaviors, emotional dependence on the parent–child relationship, and an inability by the parent to foster the child’s independence. The process of focusing on a child is thought to occur as a means of absorbing or binding the anxiety in the family, which serves to maladaptively stabilize the system (Kerr & Bowen, 1988). Focus on a child was assessed by asking mothers specific questions about the role of the children both in the nuclear family and in the participant’s family of origin. The goal was to identify whether the emotional process of focusing on a child is occurring in the family system and the extent of its presence. Specifically in the FGI, the interviewee was asked questions to identify the child on whom mothers and fathers focus on most, or with whom they are most concerned. Some examples of questions used in the current analyses include:
Which of your children worries you the most? Do you feel your partner or spouse is too focused on your children? When you and your partner or spouse have conversations, how often is the discussion about the children?
The FGI-FC scale consisted of a total of 8 quantitative and 16 qualitative questions assessing both nuclear family and family of origin. However, only six quantitative, Likert-type scale items assessing the nuclear family were used in the preliminary analyses for the current study and the subscale was removed completely in the final analyses due to low internal consistency reliability coefficients of participant responses.
Marital satisfaction inventory—revised
The Marital Satisfaction Inventory–Revised (MSI-R; Snyder, 1997) was used in this study as an external criterion with which to evaluate the construct validity of the FGI-MC subscale. The MSI-R is a 150-item true or false inventory designed to identify both the nature and intensity of relationship distress in distinct areas of couples' interaction. In the current study, four subscales from the MSI-R: Family history of distress, conflict over childrearing, sexual dissatisfaction, and global distress were used. Research has supported the reliability and constructs validity of the MSI-R subscales. Several example items from the MSI-R include: “Two people should get along better than my partner and I do” and “I have considered asking my partner to go with me for relationship counseling.” The internal consistency reliability coefficients for the MSI-R subscales ranged from .75 to .87 in the current analyses. This is consistent with other studies showing an internal consistency coefficient average of .82 and test–retest coefficients averaging .79 (Snyder, 1997).
Differentiation of self inventory—revised
The Differentiation of Self Inventory (DSI-R; Skowron & Friedlander, 1998; Skowron & Schmitt, 2003) was used in this study to assess both emotional cutoff and overall levels of differentiation of self in participants. The emotional cutoff subscale of the DSI-R was used to assess construct validity of the FGI-EC subscale.
The DSI-R is a 46-item self-report instrument grounded in Bowen family systems theory (Bowen, 1978; Kerr & Bowen, 1988) that focuses on individuals, their significant relationships, and their current relations with family of origin. Items are designed to measure one’s level of differentiation of self, such as, “I would never consider turning to any of my family members for emotional support,” or “No matter what happens in my life, I know that I’ll never lose my sense of who I am.” All items are measured on a scale from 1 to 5 with higher scores representing greater differentiation of self. All of the DSI-R subscales have demonstrated good internal consistency reliabilities (ER = .89, IP = .81, EC = .84, FO = .86, and Full scale = .92; Skowron & Schmitt, 2003). In the current study, internal consistency reliability coefficients ranged from .74 to .92.
Outcome questionnaire-45 (OQ-45)
The OQ-45 (Lambert et al., 1996) is a 45-item self-report instrument designed to assess functioning and symptom distress. Sample items from the OQ-45 include: “I feel hopeless about the future,” or “I feel no interest in things.” All items are rated on a scale from 0 to 4 with higher scores representing more symptom distress. Research has indicated that the OQ-45 is a psychometrically sound instrument, with adequate 3-week test–retest reliability (.84) and excellent internal consistency reliability coefficients (.93; Lambert et al., 1996).
The OQ-45 was completed by the participant for both herself and her husband. Using those ratings, a total individual health score was computed for the wife and the husband. A total symptom health score for the couple was also calculated. In addition, a difference score was calculated by taking the absolute difference between the wife and husband’s individual total health score. The internal consistency reliability coefficients for the individual wife OQ-45 self-report were .95 and .89 for responses for her husband. The reliability coefficient for the combined wife and husband OQ-45 scores was .94. These scores were also compared to the FGI-SS subscale.
The Parent Protection scale (PPS)
The PPS (Thomasgard, Metz, Edelbrock, & Shonkoff, 1995) is a 25-item questionnaire assessing parent focus on and protectiveness with their children aged 2–10 years old. The PPS provides an index of the parent’s tendency to be overprotective of his or her child. Sample items include: “I protect my child from criticism,” or “I feed my child even if he or she can do it alone.” This questionnaire asks the parent to rate how often an item applies to them using a 4-point scale (0 = never, 3 = always). Internal consistency of the PPS-Full scale has been shown to be acceptable (.73) and test–retest reliability was .86 over a 3- to 5-week interval (Thomasgard et al., 1995). The internal consistency reliability coefficient of the PPS in the current study was .73.
Demographic questionnaire
The participants were also asked to complete a brief demographic questionnaire. This demographic questionnaire provided information about race or ethnicity, socioeconomic status, level of education, number of years married, and age of children.
Procedures
Participants were recruited from a community database of families who periodically participate in research studies. Initial contact with participants was in the form of a letter. From there, interested mothers contacted the researcher via telephone or e-mail to schedule a time to participate in one, 2-hr session where they completed the FGI and the self-report surveys (MSI-R, DSI-R, OQ-45, PPS, and demographic questionnaire). This research session was conducted in either the mother’s home or in an office on campus depending on the mother’s preference.
The research session began with participants being provided information about the study and written informed consent was obtained. Mothers received a copy of the informed consent form and were then interviewed using the FGI. This portion of the study took approximately 75 min to complete and was audiotaped. Next, participants completed the counterbalanced packet of self-report instruments (i.e., MSI-R, DSI-R, OQ-45, PPS, and demographic questionnaire), in approximately 30–45 min. Participants were compensated $40 for their time.
A small sample of the participants (n = 6) were solicited following their interview and invited to return 1–2 weeks later to complete the FGI with a second interviewer to assess several dimensions of the FGI reliability. All agreed to return and complete the FGI only in a 1-hr second interview. All interviews were audiotaped for subsequent coding to analyze interinterviewer consistency. This subsample received an additional $20 for participating in two interviews, bringing their total compensation to $60.
FGI interviewer training
Four master’s level clinicians who were blind to the research hypotheses and experienced in conducting family genograms were trained to administer the FGI. The interviewers underwent 8 hr of specific training on how to conduct the FGI. As part of the training, a reliability evaluation of the interviewers occurred to assess prestudy interinterviewer reliability, which indicated the interviewers were able to reliably administer the FGI in terms of consistency in asking the FGI questions and yielding similar responses from confederate interviewees.
Results
Stage 1: Development of the FGI
The current study analyses were completed in two stages. Once the FGI was developed, the goal of Stage 1 involved conducting several different analyses on the participant scores at the item level to: (a) assess the internal consistency reliability of the participant scores on the FGI subscales and (b) to assess scores on individual items within the four FGI subscales based on three preset criteria. The purpose of these analyses was to identify and delete any subscales or items with low reliability, to establish the final version of the FGI, which would be used in the subsequent analyses.
The first analysis used to revise the FGI examined the internal consistency of each subscale scores. The overall internal consistency reliability of participant scores within each subscale can be seen in Table 1. The results of Table 1 indicate that participant responses on three of the FGI subscales had good reliability with α coefficients over .80. However, the focus on a child subscale scores did not have adequate reliability with a coefficient of .51. Based on this, the focus on a child subscale was dropped from all subsequent analyses.
Criteria for evaluation of FGI item reliability
The participant scores on individual items within the three remaining FGI subscales were then assessed based on three preset criteria. In order to remain in the FGI, items needed to: (1) demonstrate percent agreement above a .70 level in the interinterviewer reliability analysis, (2) not be significantly different from Time 1 to Time 2 based on a paired t test at the .05 significance level in the interinterviewer reliability analysis, and (3) not reduce the overall subscale internal consistency reliability coefficient. The results of these analyses indicated no items needed to be removed based on criterion 1. A total of 4 items were removed based on Criteria 2 and 3 as seen in Table 2.
Final composition of FGI subscales
Based on these deletions, the final FGI composition used in subsequent analyses consisted of three subscales with a total of 29 questions. Tables 3 and 4 summarize the descriptive statistics, Cronbach’s α coefficients, and the number and type of items in the final FGI composition. Higher scores on the FGI-MC, EC, and SS-Total subscales indicate greater levels of marital conflict, emotional cutoff, and symptoms in a spouse, respectively.
Descriptive Statistics for Final FGI Subscales
Note. FGI-MC = Family Genogram Interview-Marital Conflict; FGI-EC = Family Genogram Interview-Emotional Cutoff; FGI-SS (Total W/H) = Family Genogram Interview-Symptoms in a Spouse Total Score, wife and husband scores combined; FGI-SS (Dif W/H) = Family Genogram Interview-Symptoms in a Spouse Difference Score, the difference between spousal scores. Higher scores indicate greater levels of each emotional process.
Stage 2: Establishing the Psychometric Properties of the FGI
Using the final version of the FGI, Stage 2 of the study involved assessing: (a) interinterviewer reliability, (b) construct validity, and (c) conducting a cluster analysis to examine the how participants’ scores group together based on the FGI subscales.
Interinterviewer reliability
Using the interinterviewer subsample (n = 6), the reliability of each of the pairs of interviewers was again analyzed in terms of percent agreement based on two dimensions, the questions asked and the answers received. Two areas were evaluated: (a) Whether interviewers asked the question (yes or no) and (b) participant responses to the FGI questions were essentially identical across the two interviewers. For the first criterion, interviewers presented all FGI questions with participants, resulting in 100% agreement. For the second criterion, the interinterviewer reliability of the interview pairs was an average of .95. These results indicate the interviewers in the current study had good interinterview reliability for both: (a) asking the same questions across participants and (b) eliciting similar participant responses to the FGI questions.
Intercorrelations between FGI subscales and external construct validity measures
Correlations were examined between the FGI subscale scores and the scores on the external measures (i.e., MSI-R, DSI-R, and OQ-45) to preliminarily examine interrelationships among the scales. Table 5 contains the correlation matrix developed from this preliminary analysis. The results highlight several significant correlations. First, there were a number of significant, positive correlations between the FGI-MC subscale scores and scores on three (i.e., global distress, sexual dissatisfaction, and conflict over childrearing) of the four MSI-R subscales as well as with the FGI-SS subscale scores and the OQ-45 responses. Second, there was a significant positive correlation between the FGI-SS (total and difference) subscale scores and the OQ-45 (total and difference) scale scores. The FGI-EC subscale scores did not significantly correlate with scores on its corresponding scale, the DSI-R-EC. However, the FGI-EC subscale scores were shown to significantly correlate with the DSI-R-emotional reactivity subscale scores as well as with the MSI-R family history of distress and MSI-R global distress participant scores.
Intercorrelations Among the FGI Subscale, MSI-R,DSI-R, and OQ-45 Scores
Note. DSI-R-ER = Differentiation of Self Inventory-Emotional Reactivity; DSI-R-EC = Differentiation of Self Inventory-Emotional Cutoff; DSI-R-IP = Differentiation of Self Inventory-“I” Position; DSI-R-FO = Differentiation of Self Inventory-Fusion With Others; FGI-MC = Family Genogram Interview-Marital Conflict; FGI-EC = Family Genogram Interview-Emotional Cutoff; FGI-SS (Total W/H) = Family Genogram Interview-Symptoms in a Spouse Total Score; FGI-SS-DIF = Family Genogram Interview-Symptoms in a Spouse Difference Score. MSI-R GD = Marital Satisfaction Inventory Revised-Global Distress; MSI-R SD = Marital Satisfaction Inventory Revised-Sexual Dissatisfaction; MSI-R FHD = Marital Satisfaction Inventory Revised-Family History of Distress; MSI-R Chil = Marital Satisfaction Inventory Revised-Disagreement About Child Rearing; OQ-45 Tot. = Outcome Questionnaire-45-Total Score; OQ-45 Dif. = Outcome Questionnaire-45-Difference Score; PPS= Parent Protection scale.
Higher scores on the DSI-R reflect less of each construct corresponding to greater differentiation of self.
*Indicates a significant correlation at the p < .05 level; Bold indicates construct validity correlations.
Construct validity analyses
Construct validity of the FGI subscales was assessed by comparing the scores on the three FGI subscales (MC, EC, and SS) to existing, psychometrically sound measures of similar concepts. It was first hypothesized there would be a significant, positive relationship between the FGI-MC scores and the scores on the four remaining subscales of the MSI-R (Snyder, 1997). An ordinary least squares regression was conducted predicting FGI-MC as a function of the four MSI-R subscales. As Table 6 indicates, this hypothesis was partially supported. This analysis indicated participant scores on the four subscales of the MSI-R were predictive of 54% of the variance in the FGI-MC subscale scores, with global distress and conflict over childrearing scores being the strongest predictors. These results indicate mixed support for the construct validity of the FGI-MC subscale.
MSI-R Subscales as Predictors of the FGI–MC Subscale
Note. R2 = .54; *Indicates significance at the p < .05 level.
Second, it was hypothesized there would be a significant, positive relationship between the FGI-EC scores and the scores on the emotional cutoff subscale of the differentiation of self inventory (Skowron & Friedlander, 1998; Skowron & Schmitt, 2003). Using an ordinary least squares regression was conducted predicting FGI-EC scores as a function of the four DSI-R subscale scores. As Table 7 indicates, this hypothesis was not supported. This result indicates the FGI-EC subscale may not be accurately assessing cutoff as a family emotional process, at least when compared to the operationalization of the construct of emotional cutoff found in the DSI-R. However, given the relationship of the FGI-EC subscale with the DSI-R-ER subscale and the MSI-R-FHD subscale scores, there may be reason to think the FGI-EC is assessing different aspects of emotional cutoff not measured in the DSI-R-EC subscale.
DSI-R Subscales as Predictors of the FGI–EC Subscale
Note. R 2 = .20.
DSI-R-ER = Differentiation of Self Inventory-Emotional Reactivity; DSI-R-EC = Differentiation of Self Inventory-Emotional Cutoff; DSI-R-IP = Differentiation of Self Inventory-“I” Position; DSI-R-FO = Differentiation of Self Inventory-Fusion With Others; FGI-EC = Family Genogram Interview-Emotional Cutoff.
*Indicates significance at the p < .05 level.
Third, it was hypothesized there would be a significant, positive relationship between the scores on the FGI-SS-total score (wife and husband combined score) and difference score (the absolute value of the difference between wife and husband score) when compared with the total and difference scores of participants’ rating of self and spouse on the OQ-45 (Lambert et al., 1996). As noted in Table 5, this hypothesis was supported based on the significant correlation of .59 between the two measures on total score and a correlation of .31 on difference score on the two measures. These results mean the FGI-SS subscale, both total and difference scores, demonstrated construct validity and may be an accurate representation of several components of symptoms in a spouse.
Discussion
The purpose of the study was to develop a standardized genogram interview protocol, the FGI. The results of the current study indicate the relative strengths and limitations of the FGI in its final version. There are implications for the manner in which genograms are used in clinical and research settings, as well as in understanding the role of Bowen theory concepts in family functioning.
Reliability
Interinterviewer reliability
In order to advance family assessment using genogram interviewing, one of the major goals of the current project was to develop a genogram interview that could be administered reliably. The FGI was designed to be a protocol for clinicians to follow step-by-step, with clear instructions and specific questions to ask of the interviewee. In the current study, the results indicate the FGI posted adequate interinterviewer reliability based on data from a reliability subsample in which participants completed the FGI twice with two different interviewers.
Given these results, in clinical application the FGI offers a specific interview protocol for clinicians to follow, having the capability to be administered consistently across interviewers. The FGI provides a common language for assessing clients and therefore has the potential value of reducing variation in how genogram interviews are conducted. For research applications, the standardization and interinterviewer reliability of the FGI would allow for data collected using genogram interviewing to be compared, expanded, and replicated. A structured format would facilitate many new lines of important research since there is currently a paucity of research using genogram interviewing to assess families (Coupland et al., 1995).
Reliability of final FGI subscales
While interinterviewer reliability in the current study was promising, the internal consistency reliability of the FGI subscale scores was mixed. The FGI-MC, FGI-EC, and FGI-SS subscale scores demonstrated good internal consistency reliability, however, the FGI-FC subscale scores did not (Cronbach’s α = .51). There are likely several explanations for this finding.
The concept of focus on a child is likely one of the more difficult Bowen theory constructs to operationally define and the behavioral manifestation of this process may be difficult to measure. This likely meant that the items within the FGI-FC subscale were too diverse from one another in trying to capture the various elements of this construct limiting the internal reliability. Also, in the current study the FGI-FC subscale consisted of six quantitative questions. These items were initially thought to appropriately capture the construct; however, the reliability analysis indicates that this concept was not reliably assessed in a brief manner. Having a longer scale has the potential to increase the internal reliability (Niemi, Carmines, & McIver, 1986). Future revision of FGI-FC subscale will need to involve: (a) increasing the number of items on the subscale and (b) a reassessment of the behavioral manifestations of this emotional process in families to develop items that more fully address the complex nature of this concept.
Construct Validity
Beginning with the construct validity analyses of the FGI-MC subscale, the results are generally consistent with theoretical predictions and have several implications. Specifically, the FGI-MC may be best able to assess global marital conflict in couples. This finding was promising since the FGI-MC was designed to assess marital conflict, broadly defined. Also, the observed relationship between FGI-MC and the two measures of symptoms in a spouse (FGI-SS and OQ-45) indicates support for the proposed interrelationship between symptoms and marital conflict in spouses.
However, the results of the current study indicate the FGI-EC subscale did not significantly correlate with scores on the DSI-R-EC subscale. While this result was unexpected, there may be reason to believe the two scales are computing emotional cutoff differently. FGI-EC scores were developed based on items that specifically assess the participant’s relationships with his or her family of origin, including a focus on amount and quality of contact. In contrast, DSI-R-EC items assess for patterns of emotional cutoff both in significant relationships and current relations with family of origin.
However, despite the nonsignificant correlation to the DSI-R-EC subscale scores, the FGI-EC scores did correlate with the DSI-R-ER, MSI-R family history of distress, and MSI-R global distress subscale scores, indicating some support for the construct validity. Bowen described how, for some, the anxiety created by emotional reactivity in relationships is managed and bound through emotional cutoff (Kerr & Bowen, 1988; Titelman, 1998). This significant finding lends support for the idea that the FGI-EC may be assessing some facet of emotional cutoff, namely that individuals who tend to emotionally cutoff from their family of origin also reported higher levels of emotional reactivity, family history of distress, and distress in one’s marriage (Bowen, 1978).
Third, in terms of assessing symptoms, these results indicate the FGI-SS subscale is likely performing as it was designed. A significant positive correlation with the OQ-45, an existing valid measure of symptom distress and functioning, indicates the FGI-SS is able to assess symptoms present in the spousal relationship. This subscale has the potential to enhance the assessment of symptom development in spouses as it relates to overall family functioning.
Limitations
The results of the study point to important limitations in the psychometric properties of the FGI. While two subscales did have significant results in construct validity tests, but there are still questions as to whether the subscales are accurately representing systemic components of their respective constructs. Specifically, it is likely the subscales still may be limited in this regard, especially when considering what the questions are assessing, who was interviewed, and the number of items in each subscale. Further, in the current study the decision was made to limit the analyses to only quantitative questions in order to more easily facilitate the initial investigation of the psychometric properties of the FGI. The open-ended responses from the qualitative questions within the interview are likely to yield rich detail and an interesting picture of family functioning, as well as provide more information about the psychometric properties of the FGI. Additionally, developing the FGI required making decisions to limit the wide range of material that could possibly included in a genogram interview. It is believed that the potential long-term benefit of a standardized genogram interview will outweigh the relative loss of some of the richness in clinical application.
Another important limitation of the current study is that only one member of the family system was interviewed. The methodology used for the construct validity assessment of the FGI-SS subscale is perhaps the most apparent example of this type of limitation. The FGI-SS and OQ-45 scores were both derived from the data the participant provided rating both herself and her husband on this construct. The results on this subscale therefore reflect only the wife’s perception of her husband’s health and functioning highlighting the need to include more than one family member in future studies.
Further, along with the fairly restrictive preset inclusion criteria, the sample was found to be relatively homogenous on most other demographic characteristics as well. The homogeneity and the nonrandom nature of the sample means the results are best generalized to Caucasian, heterosexual, married women who are upper-middle class and generally well educated, in the lifestyle cycle stage of “families with young children” (Carter & McGoldrick, 1999). In addition, the sample size was relatively small, which further limits the generalizability of the results. It is important to also acknowledge that many family structures do not have a heterosexual, two parents, with children structure. Future research on the wide variety of family types using the FGI would be a valuable addition to the research literature.
Despite the limitations, it is expected that future revision and research that follows up on the current findings will lead to a better understanding of the family emotional processes that contribute to individual and family health and functioning, based on this first step in developing a more comprehensive version of the FGI.
Future Directions
Refinement of the FGI
The overarching need for revision of the FGI involves addressing the construct validity of the subscales, particularly as it relates to the assessment of emotional processes in families from a systemic perspective. This revision should address adding more items and increasing the detail within the items. Most importantly, the current version of the FGI also needs to add more items that are process oriented in nature, with information from various members of the family system to assess the systemic components of each emotional process. The FGI-FC would also likely benefit from rewording of items, a possible reconceptualization of the construct, and a more thorough arrangement of the types of items in the subscale. Further, participant responses to the qualitative portions of the interview may yield important information about family relationships that may be difficult to capture in a quantitative manner.
Once a thorough revision of the FGI is implemented based on the above recommendations, it will be important to first conduct pilot testing of the interview to examine the usefulness of the redesigned subscales and items. From there, a study similar to the current study would be useful in reexamining the psychometric properties of the improved interview. Also, a more complete standardization of the FGI needs to include an investigation of whether or not the psychometric properties of the interview remain consistent with more diverse samples. Overall, with improvements made to the FGI, there are important questions about individual and family systems functioning that could be answered using the FGI that were previously not possible.
Clinical Application
The current study has several implications for clinical application and practice. First, the FGI was designed to be a genogram interviewing protocol that can be used reliably from one interview to the next. In the current study, the FGI was shown to have good interinterviewer reliability, which improves the standardization of how genogram interviews are conducted. For a clinician, the demonstrated interrater reliability of the FGI provides the opportunity to conduct genogram interviews across clients and expect formatting and questions to be the same. This offers the ability to compare clients’ scores as a point of reference, confidence in the consistency of how the genogram interview is conducted, and provides the basis for a common language in genogram interviewing. These possibilities are in contrast to how genogram interviewing is currently conducted in clinical application, namely that there is no common language and no set standardization of questions to be asked, areas to assess, and very limited interinterviewer reliability.
Conclusion
It is hoped the current study serves as an initial step toward developing a standardized genogram interview designed to assess emotional processes in families. The FGI represents a first step toward filling the need for an empirically demonstrated, reliable, and valid means of assessing family functioning, both the nuclear and multigenerational families. Future revision and use of the FGI could be valuable in clinical and research settings toward developing more complex understanding of the role of family in health and functioning.
Footnotes
Acknowledgments
This research was supported by a grant from the American Psychological Foundation, The Randy Gerson Memorial Grant. The authors would also like to acknowledge the Bowen Center for the Study of the Family for their assistance in this research.
Declaration of Conflicting Interests
The author(s) declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.
Funding
The author(s) declared the following financial support for the research, authorship, and/or publication of this article: This research was funded by the American Psychological Foundation; The Randy Gerson Memorial Grant.