Family network typologies of adults with intellectual disability: Associations with psychological outcomes

Abstract Background Based on self‐reported social capital, different typologies of family networks of people with intellectual disabilities were examined. Associations between behavioural and emotional problems or well‐being and typologies were investigated. Method 137 participants with mild intellectual disability were interviewed using the Family Network Method‐Intellectual Disability to assess their emotionally supportive family relationships. Data on participants’ well‐being and behavioural and emotional problems were also gathered. Latent class analysis was used to identify family typologies based on social network measures. Results Four distinguishable typologies were identified, two supportive and two less supportive. A small association was found with behavioural and emotional problems and one of the supportive typologies. Associations with constructs of well‐being were found for both supportive and less supportive typologies. Conclusions A variety of family types were found, with implications for sensitive professional support.

Most research about family networks of people with intellectual disability has only examined the person with an intellectual disability's view of their own relationships with other family members and not the (reciprocal) relations between all network members (Lippold & Burns, 2009;Robertson et al., 2001;Van Asselt-Goverts, Embregts, & Hendriks, 2013). There is a small qualitative literature about people with intellectual disabilities' perspectives on mutual care, between them and their family members. In one study (Walmsley, 1996), 22 adults with an intellectual disability were interviewed about their lives, and giving and receiving care. Results suggested that the participants did not view themselves as being dependent on family care. Instead, they saw themselves as having family roles which allowed for a sense of reciprocity and mutuality. Williams and Robinson (2001) compared the perspectives of 40 parents with those of 45 (young) adults with an intellectual disability.
Based on the interviews, they found that many people with intellectual disabilities and their parents did not feel that there was mutual support. Parents generally defined themselves as carers who took responsibility and exercised control. Both of these studies suggest that a more holistic approach is required, taking into account the complex web of interdependence within families; a model that recognizes mutually supportive relationships and considers the resources needed by the whole family. To capture the multi-dimensional nature of family networks, approaches such as the Family Network Method are required (Widmer, Aeby, & Sapin, 2013). The Family Network Method is an instrument that maps who the participant considers to be their family members. In addition, it assesses how the participant perceives the relationships between these family members (Widmer, Kempf, Sapin, & Galli-Carminati, 2013). This method has been adapted by Giesbers et al. (2019) as the Family Network Method-Intellectual Disability (FNM-ID) to ensure it is accessible and meaningful for use in the population of individuals who have an intellectual disability.
Initial results using the FNM-ID showed that there is likely to be considerable variation in the perceived family networks of people with intellectual disabilities, in terms of size and (reciprocal) emotional support (Giesbers et al., 2020). Examining family typologies can have practical implications, as different types of families might require different professional support. There has been very little research about family network typologies in the field of intellectual disabilities. The few studies published have been focused on parents' perceptions instead of the perception of the person with intellectual disabilities themselves (Mink, Meyers, & Nihira, 1984;Mink, Nihara, & Meyers, 2002). One previous study did take the perspectives of people with intellectual disability into account, to identify typologies of their family networks. Widmer, Kempf, et al. (2013) explored the different family roles instead of the significant emotional support they provide. Widmer, Kempf, et al. (2013) performed a cluster analysis and described four family configurations: professional, kinship, nuclear and friendship family configurations. However, Widmer, Kempf, et al. (2013) did not distinguish who is providing or receiving emotional support to/from the person with intellectual disability, which is the essence of social capital (Bourdieu, 1986).
Given the lack of family network typology studies, based on self-reports of people with intellectual disabilities, the current study examined whether different typologies of perceived family networks can be distinguished in terms of emotional support. Previous research has also shown associations between dimensions of family-based social capital and the behaviour problems (McPherson et al., 2014) and living situation (Widmer, Kempf, et al., 2013) of people with intellectual disabilities. Therefore, a secondary aim of the current study was to examine associations between types of family networks and the personal characteristics of the individuals with intellectual disabilities, their behavioural and emotional problems, and their well-being.

| Participants
The 137 participants had a mean age of 28.2 years (SD = 6.16, range 18-40); 56.2% of the participants were male, 92.0% had a [COUNTRY] cultural background, and 44.9% were officially diagnosed by a certified clinician with a psychiatric or developmental disorder, with autism the most common category (24.1%). Most of the participants lived in a facility in the community (83.9%), and the others lived in a residential facility (see Table 1 When the person with intellectual disability agreed to participate, an appointment was made. Depending on participant preference, face-to-face interviews were carried out at the participants' homes or at the service providers' offices. The researcher carefully explained the purpose of the study, the procedure and the confidentiality of the information. A standard consent procedure (Arscott, Dagnan, & Kroese, 1998) was then followed to assess the capacity of the participant to give consent to take part in the research.
Participants were given a written and verbal overview of the research project, and the researcher asked them the five questions developed by Arscott et al. (1998) After the interview, participants were given a ten euros in recompense for their time. Eleven participants were excluded because their scores on the cognitive assessments were above or below the mild intellectual disability range. Two individuals were excluded because they were not able to answer the questions, leaving a total of 137 participants.
With the participant's consent, their key support worker was interviewed to obtain information about the participants' behavioural and emotional problems and to check additional information about officially diagnosed psychiatric or developmental disorders obtained from personal records.

| Family networks
The Family Network Method-Intellectual Disability (FNM-ID) was used to measure the family networks of people with intellectual disability. The FNM-ID makes it possible to analyse emotionally supportive relationships, by asking participants to estimate relationships among all their family members. Emotional support refers to a belief that love and caring, sympathy and understanding, and/or esteem and value are available from significant others (Thoits, 1995).
The instrument is composed of multiple steps; participants are first asked to provide a list of all individuals whom they consider to be a family member at the time of the interview. The term "family" is deliberately left open to allow the participants to apply their own definitions. Subsequently, participants are asked to list all family members that are significant to them from the listed family members. Finally, they are asked to describe which family members provide emotional support to the participant and to each other from the list of family members, by asking them the question: "If X is feeling 'out of sorts', who is there for X?" (i.e. X represents each individual included the participant's family configuration, considered in turn).
Socio-demographic information on each listed family member is collected, as well as information on the nature of the family tie, the duration of the relationship (if not a relative) and the frequency of contact.
To characterize the family networks of individuals with intellectual disability, seven social network measures are computed from the FNM-ID (Table 2), which are related to a social capital theoretical perspective (Sapin et al., 2016;Widmer, Kempf-Constantin, & Carminati, 2010) and are of interest in terms of the lives of people with intellectual disability.

| Cognitive ability
To check whether a participant met the inclusion criteria of mild intellectual disability, a brief screening IQ-score, based on standard scores and standard errors, was derived because file scores of the participants were often missing, outdated or obtained using unidentified IQ tests. The subtests "Vocabulary" and "Matrix Reasoning" from the [LANGUAGE] version of the WAIS-IV (WAIS-IV-NL) were used (Wechsler, 2012 The raw scores were turned into standardized scores per participant; then, a 95% confidence interval was calculated per subtest with help of the standard measurements of errors. When a participant had a standard score in the intellectual disability range on both subtests (according to the 95% confidence interval), the participant was deemed as having a mild intellectual disability. People with intellectual disabilities often have a varied intelligence profile.
Therefore, participants who score on only one subtest in intellectual disability range were also included in the study. Participants who scored above the intellectual disability range on both subtests were excluded.

| Well-being
Subjective well-being was measured using the Personal Wellbeing

| Behavioural and emotional problems
The Adult Behavior Checklist (ABCL; Achenbach & Rescorla, 2003) was used to measure behavioural and emotional problems. This questionnaire examines a broad range of behavioural and emotional problems: anxious/depressed, attention, withdrawal, aggression, somatic complaints, rule-breaking and intrusive thoughts. Although the ABCL was developed for the general population, a study on the psychometric properties for the use of this instrument with people with intellectual disability has been conducted. Tenneij and Koot (2007) found that the internal consistency coefficient Cronbach's alpha of the ABCL scales, for people with intellectual disabilities, ranged from 0.69 to 0.95 (mean alpha = 0.84); and inter-rater reliability, assessed by the intraclass correlation coefficient, ranged from 0.57 to 0.76 (mean = 0.68). In the current study, the present authors used the scores on the eight subscales. Key support workers were asked to rate whether these items were true of the participants over the past 6 months using a 3-point response scale: "not true" (0), "somewhat or sometimes true" (1) or "very true or often true" (2). Total scores for each scale were converted into T-scores.

| Demographics
Demographic information (participants' sex, age, living situation and cultural background) was obtained during the interview, or afterwards from the participant's file (with consent of the participant).

| Data analysis
The family network data were entered into Excel and then imported and analysed in UCINET (version 6.623), a software package for the analysis of social network data (Borgatti, Everett, & Freeman, 2002).
To identify empirically meaningful family typologies based on the FNM-ID variables, the present authors used latent class analysis (LCA) in Mplus (Muthén & Muthén, 1998-2015. LCA is a probabilistic version of traditional non-hierarchical cluster analysis. The inputs for LCA were the seven social network variables (Table 2) standardized as z-scores.
To identify the ideal number of classes (family typologies), several criteria were used. The first criterion, the measure of parsimony, was the Baysian information criterion (BIC; Kass & Raftery, 1995). After the number of classes was identified, the second step was to test differences between the classes with respect to demographic variables, well-being, and behavioural and emotional problems.
Differences across classes with categorical or binary variables were tested with a chi-square test as described by Lanza, Tan, and Bray (2013) and differences across classes with continuous variables with a chi-square test as described by Asparouhov and Muthén (2014).
Both tests are available in Mplus. An overall chi-square test was carried out for all four classes together. A significant result was followed by chi-square tests comparing each two class combinations.

| Description of the Four Classes
To be able to describe the four different classes and depict them with illustrative graphs (NetDraw, Borgatti, 2002), the raw mean scores for each of the classes on the seven social network variables were calculated (Table 4). To

| Comparisons across classes
To test differences across the four classes, the present authors used a chi-square test for categorical/binary outcomes (Lanza et al., 2013) and a chi-square test for continuous outcomes (Asparouhov & Muthén, 2014), as presented in Table 5. Significant differences across the four classes were found for cultural background, TA B L E 4 Means scores, 95% bootstrap confidence intervals and typology of four family network classes   Note: Only significant overall chi-square tests with p < .05 were followed with chi-square post hoc tests.
TOURNIER ET al. to be close, although the participants in this class do score well on community connectedness.  (Baumeister & Leary, 1995) and have a positive effect on the self-worth of people with intellectual disability (Milner & Kelly, 2009). In turn, this might contribute to the high score for well-being in terms of personal relationships. At the same time, participants in class 3 also scored highest on the subscales rule-breaking and intrusive behaviour of the ABCL.
Typically, people with intellectual disability who have behavioural and emotional problems are among the most disadvantaged and socially excluded in society (Emerson, 2001). The results for class 3 seem at odds with earlier research. There may be at least two possible hypotheses for this result. First, the inflated self-perceptions of people with behavioural problems have been attributed to an illusory positive bias (Barry, Kerig, Stellwagen, & Barry, 2011). Another potential explanation could be that people who are more aggressive are less passive and more demanding, and so are better at maintaining relationships with family members. These hypotheses, and others, could be explored in future in-depth research with people with intellectual disabilities who have families matching the class 3 typology.
Limitations of the current study should be mentioned. First, the present authors focused on emotional support because it has been found to be a stronger predictor for physical and mental health-related outcomes (Berkman, 1995;Thoits, 1995).
However, different results may be obtained if the focus is on instrumental support. Second, no rating of the quality of emotional support was included. Quality of social relations may have an impact on well-being. Positive aspects of supportive relationships appear to provide a sense of security, increasing individuals' positive feelings about themselves and their world (Antonucci, 2001).
Future research could replicate the current methods with a focus on instrumental support and the quality of emotional support-to examine the replicability of the typologies identified. Third, there were some limitations concerning sampling and recruitment. Only 42.4% (n = 150) of the participants who were initially identified agreed to take part in the study meaning that the resulting sample was unlikely to have been representative of the population studied. Unfortunately, no data were available for the non-respondents. Therefore, it was not possible to quantify biases in the sample selection. The present authors only included participants who lived apart from their family in long-term care and who were supported by staff. Their distance from family members and the nature of their support may have had a significant impact on their perceptions of family. In future research, it would be important to examine family typologies in populations living in different contexts, and cultures, and with more representative samples.
The current study was exploratory, and yielded insights about family network typologies of people with mild intellectual disability from their own perspectives. The findings showed that their social capital is low on average but that there is some variability. In terms of practical implications, these findings suggest that people with intellectual disability have different support needs in terms of strengthening or extending their social capital.

ACK N OWLED G M ENTS
The present authors would like to thank the individuals who participated in this study and the staff from the service providers who supported recruitment to the research.