The Beach Center Family Quality of Life Scale is an internationally validated instrument for measuring family outcomes. To revise the scale for better alignment with the Family Quality of Life theory, the authors excluded non-outcome items in this revision. In this study, we examined reliability and validity of the revised scale (i.e., the FQoL Scale-21) and its scores for Taiwanese families of children and youth with intellectual disability and developmental delay (age 0–18). Results from 400 Taiwanese respondents suggested that the FQoL Scale-21 has the potential to be used as an indicator of positive outcomes in intervention evaluation, policy making, and service delivery.
Historically, the majority of disability research in Taiwan has centered on outcomes for people with disabilities and generally overlooked the well-being of families of people with disabilities (Chou, Lin, Chang, & Schalock, 2007; Hsu, 2007). In the past decade, there has been an increased recognition of the significance of supporting families as a way to improve outcomes for people with disabilities (Chou, Lee, Lin, Kröger, & Chang, 2009; Turnbull, Turnbull, Erwin, Soodak, & Shogren, 2011).
Family studies are especially important in the context of the traditional values in Taiwan, and this importance is reinforced by law. Similar to other Asian countries, family members in Taiwan consider themselves responsible to take care of each other (Chan & Lee, 2004). While 92.8% of people with disabilities live at home, family members assume primary caregiver roles for more than half of them (Department of Statistics at Ministry of Interior, 2012). The People with Disabilities Rights Protection Act (2015) holds local government agencies accountable for delivering services to family caregivers (i.e., caregivers who co-reside with and are related either by reason of biology or marriage to people with disabilities) and further requires local government agencies to evaluate and provide technical assistance periodically to organizations responsible for service delivery.
Presumably, the long-term impact of individualized and appropriate family services in Taiwan is enhanced family outcomes (Brown, R. I., Hong, Shearer, Wang, & Wang, 2010). However, research on family outcomes remains scant in Taiwan. Therefore, this research aims to examine the reliability and validity of a family quality of life (FQoL) Scale for Taiwanese families of children and youth with intellectual disability or developmental delay (aged 0–18). In the following sections, we describe family outcome studies and the FQoL conceptual framework. Then, we summarize relevant research on measuring FQoL.
Family Outcome Studies and FQoL Conceptual Framework
The construct of FQoL provides an important ecological framework for family outcome evaluation. FQoL is defined as “a dynamic sense of well-being of the family, collectively and subjectively defined and informed by its members, in which individual and family-level needs interact” (Zuna, Summers, Turnbull, Hu, & Xu, 2010, p. 262). Zuna et al. (2010) conducted a comprehensive literature synthesis and proposed an overarching conceptual theory as a foundation to build an FQoL theory. Figure 1 depicts an updated conceptual framework derived from a more recent literature review (Chiu et al., 2013). Family strengths, needs, and priorities provide input for systemic factors, family-unit and individual member factors, and support factors. The multiple factors interact with each other to produce the FQoL outcome. Finally, the FQoL outcome contributes to new family strengths, needs, and priorities.
In the past decades, researchers have developed measures with reliability and validity to collect data on FQoL. For example, one study (Hu, X., Summers, Turnbull, & Zuna, 2011) reviewed 16 existing measures published between 1980 and 2009. Two of the FQoL measures identified by the authors were validated and used in disability studies on more than one ethnic group. The two tools, the FQoL Survey-2006 (Brown, I., et al., 2006) and the Beach Center FQoL Scale (Hoffman, Marquis, Poston, Summers, & Turnbull, 2006), both cover domains related to Family Interaction, Emotional Well-Being, Physical/Material Well-Being, and Disability-Related Services. In addition to the overlapping domains, the Beach Center FQoL Scale (Hoffman et al., 2006) has one domain labeled Parenting. In contrast, the FQoL Survey-2006 (Brown, I., et al., 2006) contains additional domains labeled Support from Other People, Influence of Values, Careers and Planning for Careers, Leisure and Recreation, and Community Interaction.
Both of these measures are intended to be used as outcome measures in studies focusing on the impact of factors on FQoL. The use of the FQoL construct as an outcome measure is increasingly considered an outcome of choice because it is a more neutral construct, allowing for the possibility of both positive and negative experiences (Chiu et al., 2013). Previously, the majority of research on families has been focused on negative outcomes such as depression or stress (Kyzar, Turnbull, & Summers, 2012). Further, the majority of research to date has been descriptive research investigating the influence of child and/or family characteristics on family outcomes (Turnbull, Summers, Lee, & Kyzar, 2007). These authors called for more family research investigating the influence of services and supports on family outcomes, and in particular on FQoL as an outcome.
The challenge is that both of the existing FQoL measures include subscales and items, which measure the availability of services and supports as part of the FQoL construct. For example, the Beach Center FQoL Scale includes a Supports subscale with items that assess family satisfaction with these supports. This subscale would be tautological in a study investigating whether services are effective in improving FQoL, in that the study results would in effect be saying that effective services are related to effective services. However, there is very sparse research to determine whether the Beach Center FQoL Scale is a psychometrically strong measure after removal of these service-related items, with the exception of one study validating the scale without the Supports subscale on families with typically developing, young children (Zuna, Selig, Summers, & Turnbull, 2009). Therefore, it is an important feature in the continuing line of research related to FQoL, to determine the psychometric validity of a measure in which support-related items are removed.
Instrument Used to Measure FQoL in Taiwan
As FQoL research evolved, there has been increased attention on FQoL studies in non-Western countries (Wang, 2010). Unfortunately, the majority of family research in Taiwan has reported family caregivers' individual quality of life, where the caregivers primarily are mothers; only a few studies in Taiwan examined FQoL at a family-unit level (Hsu, 2007; Tang et al., 2005). Consequently, information on instruments is limited. We found two studies published in Taiwan using the term “FQoL” that attempted to investigate quality of life at a family-unit level with items related to all family members. The first study, conducted by Tang et al. (2005), centered on a 35-item FQoL measure derived from two Western-based studies (Brown, I., Anand, Fung, Isaacs, & Baum, 2003; Poston et al., 2003). Using data collected from 152 Taiwanese families, the research team performed initial analyses based on the data-driven approach and found a 6-factor solution for the tool. The second study also centered on scale validation, in which Hsu (2007) collected responses from 397 families of young children with disabilities (aged 0–7) and proposed a 3-factor 17-item scale. Nevertheless, results from both studies need to be interpreted with caution, given that both contained methodological concerns (e.g., inadequate sample size, inadequate number of items for one factor, and insufficient data on psychometrics). Finally, neither measure has yet to be employed in further studies.
In addition to the two self-developed measures reported in the aforementioned studies, the Family Quality of Life Survey-2006 (Brown, I., et al., 2006) was also used to investigate FQoL of Taiwanese families with children with disabilities. In a cross-cultural comparative study, R. I. Brown et al. (2010) examined responses on the FQoL Survey- 2006 (Brown, I., et al., 2006) from 83 Taiwanese families of children with autism (aged 1-14) in one urban area. The results of this study, based upon a small and specific sample (urban families of children with autism), might not reflect FQoL and potential group differences among Taiwanese families of children with other disabilities. The study solely focused on reporting outcomes and did not report the translation and validation processes in the report.
To sum up, there is a lack of psychometrically sound instruments to understand FQoL of Taiwanese families. Among the existing scales, we selected the Beach Center FQoL Scale in this study because of its cross-cultural validity and short length. In addition to numerous items that confound service satisfaction with family outcomes, the FQoL Survey- 2006 requires respondents to rate items across six underlying concepts of importance, opportunities, initiative, attainment, stability, and satisfaction. Thus the measure presents complexities in scoring and interpretation, as well as increased respondent burden. The shorter Beach Center FQoL Scale, without the non-outcome items and using satisfaction only, has the potential to be a useful tool in combination with other measures in research and large-scale evaluation of services and policies. In order to examine the construct validation of 21 items of the Beach Center FQoL Scale (hereafter referred as the FQOL Scale-21) for Taiwanese families, we addressed the following research questions:
Does the FQoL Scale-21 produce reliable scores to measure four subdomains (i.e., family interaction, parenting, emotional well-being, and physical/material well-being) and the overall construct of quality of life of Taiwanese families of children and youth with disabilities?
Does a four-factor model of the FQoL Scale-21 fit the data well? If yes, does this model further demonstrate the convergent and discriminant evidence?
Does a higher-order factor model of the FQoL Scale-21 provide good model fit?
Which model provides a better fit to the data: a four-factor lower-order model or a single higher-order factor model?
Due to restrictions from the Personal Information Protection Act (2015) in Taiwan, we employed an indirect survey distribution strategy. Upon approval from the University Human Subjects Committee, we first contacted agencies responsible for service delivery and parent support in Taiwan with geographic quota sampling strategies to ensure sample representativeness. Next, we mailed 500 survey packets through Chunghwa Postal Service (equivalent to U. S. Postal Service) to 26 service providers (e.g., social workers, teachers) selected by their agency administrators. These service providers from eight local early intervention centers, five parent support groups, and 11 schools then followed researcher-developed instructions to distribute survey packets to family respondents. Within four weeks of distribution, we received 409 completed surveys (81.8% return rate). After data screening, the sample was reduced by nine respondents to 400. The nine respondents who were eliminated did not respond to 15% or more items of the FQoL section (missed more than three questions). The majority of respondents were mothers (n = 290, 72.50%), non-aboriginal Taiwanese (n = 366, 91.50%), and married/ living with a partner (n = 321, 80.30%). Over half of them had a son with disabilities (n = 256, 64.00%) and school-age children with disabilities (n = 225, 56.25%). Table 1 provides detailed information on demographics of the respondents.
Each participant received a stamped self-addressed envelope to use in returning the survey packet and a pre-incentive of a gift card for 100 New Taiwan dollars (approximately three U.S. dollars). The survey packet contained the FQoL Scale-21 and demographic questions in Traditional Chinese.
The Beach Center Scale (Hoffman et al., 2006) was designed to measure families' perceived satisfaction in five domains of life: Family Interaction, Parenting, Emotional Well-Being, Physical/Material Well-Being, and Disability-Related Services. Scores obtained from the Beach Center Scale were reliable (α =.88 for the whole scale, .80–.92 for subscales) and valid as evidenced by moderate correlations between the Family Interaction domain with the Family APGAR (r = .68), as well as the Physical/Material Well-Being domain with the Family Resource Scale (r = .60; Hoffman et al., 2006; Summers et al., 2005). The item-level overall FQoL higher-order model had acceptable fit: χ2 (270) = 439.24, p < .001, CFI = .92, and RMSEA = .05. Furthermore, in another study, the satisfaction rating from 442 Chinese families had an acceptable internal consistency (α = .88), test-retest reliability (r = .60–.77); fit for the subscale level model was acceptable: χ2 (265) = 748.15, p < .001, CFI = .97; and RMSEA = .07 (Hu, X., Wang, & Fei, 2012). To ensure cultural appropriateness of the Chinese Beach Center FQoL Scale for Taiwanese participants, Tseng and Liu (2013) recruited social workers and parents of children with disabilities to refine the items. The revisions included minimal changes in written text (i.e., from Simplified Chinese to Traditional Chinese) and wordings (e.g., change from “handicap” to “disability”) deemed appropriate by Taiwanese participants. Because this study aimed to refine the Beach Center Scale by retaining only outcome-related items, we removed the four items related to Disability-Related Services and retained 21 relevant items across four domains in this study. Relying on the same rating scale (5-point Likert) and anchors (1 - very dissatisfied to 5 - very satisfied), Taiwanese families responded to each item to indicate their level of satisfaction. (Refer to Table 3 for a complete list of all 21 items.)
The survey included multiple questions about the respondents (i.e., gender, relationship to the child, nationality, date of birth, marital status, employment status, educational level, geographical location, household income, additional support at home) and their children with disabilities (i.e., gender, date of birth, severity of disability).
Given the strong theoretical structure of the FQoL Scale-21 (Chiu et al., 2013) and previous confirmatory factor analysis on families of children without disabilities (Zuna et al., 2009), we chose a confirmatory factor analyses (CFA) procedure to establish construct validation of the FQoL Scale-21 using the sample collected in Taiwan. On average, there was 0.7% rate of missing data on 21 items (range: .03%–2.3%); full information maximum likelihood (FIML) estimation was used to recover these missing data (Schlomer, Bauman, & Card, 2010). We specified the models using a fixed-factor method of identification and used the maximum likelihood estimation in Mplus version 7.0 (Muthén & Muthén, 1998–2012).
Research question 1—Reliability indices
We evaluated the internal consistency of the FQoL Scale-21 scores to examine “the extent to which individual differences in measured scores are consistent and producible” (Widaman, Little, Preacher, & Sawalani, 2011, p. 41). Specifically, coefficient alpha (Cronbach, 1951) and coefficient omega (McDonald, 2013) were calculated for four subscales and the overall FQoL construct. Coefficient alpha is the most frequently used internal consistency index. However, it assumes that all items of a given construct are equally good indicators of the latent construct being evaluated, which does not occur in the FQoL Scale-21. Thus, we also reported coefficient omega that assumes all items on a scale differentially explain the variances of latent constructs.
Research questions 2 to 4—Model fit/Model comparisons
The confirmatory factor analysis was performed to test the four-factor structure of the FQoL scale-21 (i.e., family interaction [6 items], parenting [6 items], emotional well-being [4 items], and physical/material well-being [5 items]). To evaluate model fit, we used standardized root mean square residual (SRMR), root mean square error of approximation (RMSEA), comparative fit index (CFI), and Tucker-Lewis index (TLI; Brown, T. A., 2015). SRMR is the average discrepancy between the observed and model-predicted correlation. With zero representing a perfect fit, a smaller SRMR value indicates the better model fit. RMSEA estimates the lack of fit in the model as compared to a saturated model, with values between .08 and .10 considered mediocre, values between .05 and .08 considered acceptable, and values below .05 considered excellent fit (Browne, Cudeck, Bollen, & Long, 1993; MacCallum, Browne, & Sugawara, 1996). CFI assesses the misfit between our model and the worst-fitting model (i.e., independent model) as a proportion, whereas TLI evaluates fit between our model and the worst model in terms of degrees of freedom. We considered that CFI and TLI values greater than .90 represented acceptable fit as values close to 1.00 indicate better fit. (Hu, L., & Bentler, 1999).
Once a four-factor structure was confirmed, we reported factor loadings, residual variances, and R2 (i.e., total variance – residual variance). When factor loadings are significantly loaded on their distinct latent constructs, convergent validity is achieved (Anderson, 1987). To empirically examine discriminant validity, we followed the procedure introduced by Bagozzi, Yi, and Phillips (2001). Specifically, we performed chi-square difference tests between each set of nested models (a hypothesized model with correlations being freely estimated vs. an alternative model with a correlation between a pair of latent constructs being fixed to one). When each set of nested chi-square evaluation indicated that latent correlations were significantly different from one (χ2 diff (1) > 6.64, α = .01 [we used an alpha level of .01 throughout this paper due to chi-square test's sensitivity to sample size]), we regarded discriminant validity as attained.
After examining (a) the model fit of the four-factor structure and (b) the magnitude and pattern of correlations among factors, we conducted a higher-order confirmatory factor analysis by reproducing the correlations among four factors of an initial CFA model with a more parsimonious higher-order FQoL factor structure. We then evaluated the model fit of a higher-order solution based on the same model fit indices and criteria mentioned previously. When the higher-order solution provided an acceptable fit to the data, we performed a chi-square difference test between nested models (i.e., an initial four-factor model vs. a higher-order model) to examine which model provides a better fit to the data.
Research Question 1
Table 2 provides two types of internal consistency indices for four subdomains and an overall FQoL construct. Cronbach alphas exceeded .8 for all subscales and the composite construct, demonstrating satisfactory internal consistency (Ponterotto & Ruckdeschel, 2007). Coefficient omegas, which were more appropriate reliability indices for the FQoL Scale-21, were slightly larger than corresponding Cronbach alphas (range: .82–.96) and represented good internal consistency.
Research Question 2
Results from the confirmatory factor analysis on this Taiwanese sample indicated adequate fit of the sample data to the four-factor structure: χ2 (183) = 763.555, p < .001, SRMR = .053, RMSEA = .089 (90% CI: .083–.096), CFI = .907, and TLI = .894. Parameter estimates from this solution, including factor loadings, residual variances, and R2 are presented in Table 3. All standardized factor loadings (i.e., standardized regression coefficients that explain the degree of standard deviation change in items when one standard deviation changes in the respective latent factor) exceeded the predetermined .4 (ranged .62 to .91) and were significant at .01 alpha level, providing convergent evidence of the FQoL Scale-21.
Next, we examined latent correlations among the four subdomains of the FQoL Scale-21. As provided in Table 4, latent correlations ranged from .80 to .94 and were significant at the alpha level of .01. To examine discriminant validity, we then performed six sets of chi-square difference tests by comparing a model without a constraint and a model with a constraint on a given correlation (Table 5). Because every χ2 diff exceeded 6.64 (the critical value at the alpha level of .01 with df = 1), six correlations among four latent constructs were all significantly different than one, empirically supporting the discriminant evidence of the FQoL Scale-21.
p < .01.
Note. CFA = confirmatory factor analyses; FI = Family Interaction; PA = Parenting; EW = Emotional Well-Being; PW = Physical/Material Well-Being.
Model serves as a baseline model (correlations were freely estimated).
Research Question 3
Because all four subdomains were highly correlated in Research Question 2, we hypothesized a second-order construct would provide a more parsimonious representation of factor structure for the FQoL scale (the second model in Figure 2). Results from the higher-order CFA suggested equally acceptable model fit to data: χ2 (185) = 771.845, p < .001, SRMR = .053, RMSEA = .089 (90% CI: .083–.096), CFI = .906, and TLI= .894. The unstandardized factor loadings, standardized factor loadings, residual variances, and R2s are presented in Table 6. Each of the first-order factors (i.e., four domains) significantly loaded on the higher-order factor, ranging from .82 to .98. The residual variances represent the proportions of variance in the first-order factors that were not explained by the higher-order factor. For example, the physical/material well-being construct had 34% of variance that was not predicted by the overall FQoL construct. In turn, R2s (i.e., total variances—residual variances) in Table 6 indicated that higher-order factor accounted for about 67%–95% of the variance in the first-order factors.
Research Question 4
To identify which model (first-order CFA vs. higher-order CFA) provided a better fit, we performed a follow-up nested chi-square comparison: Δχ2 (2) = 8.29, p = .016. Because the higher-order model did not produce a significant decrease in model fit using the alpha level of .01 for decision making (i.e., the higher-order model provided comparable fit when compared with a lower-order model, but was more parsimonious), the higher-order solution was preferred to conceptualize FQoL.
We proposed research questions to evaluate the psychometric properties of the FQoL Scale-21 for Taiwanese families. Using the confirmatory factor analysis that adjusted for measurement error, the underlying structure of the FQoL Scale-21 was supported using our Taiwanese sample. In this section, we address the limitations of the study; summarize the findings; and propose future directions for research, policy, and practice.
Limitations of Study
Several limitations should be considered in deriving conclusions from the results. First of all, the majority of the respondents were families of children and youth with developmental delay or intellectual disability. The sample limited generalization of the findings to families of people with other disabilities. Second, similar to the majority of family studies in the disability-related field, respondents in the study were primarily mothers (Turnbull et al., 2007). To build on the research findings from this study, future researchers should obtain responses from multiple members to the maximum extent culturally appropriate, and collect more detailed demographic information on respondents and non-respondents to more fully account for extraneous variables that could potentially influence findings. Third, western researchers may question the high response rate in survey returns for this study. This is not out of line with other survey studies in Asian cultures (Lin, Lo, & Chiu, 2008; Wang, Huang, & Chiu, 2009) given the strong emphasis in this Confucian society on respecting the authority of teachers. However, it cannot be discounted without further research, that the responses could have been biased given the fact that the surveys were distributed to the families by their children's teachers and service providers.
Summary of the Findings
This study provided evidence of strong psychometric properties of the FQoL Scale-21. First, the reliability indices for the four subdomains and the overall FQoL construct were satisfactory, indicating that the subscale scores and the total scores of the FQoL Scale-21 demonstrated consistency across replications of an assessment procedure (American Educational Research Association [AERA], American Psychological Association [APA], & National Council on Measurement in Education [NCME], 2014). Second, we found favorable results regarding construct validation of the FQoL Scale-21. Each set of items that measured a given construct was theoretically similar and the four factors were significantly intercorrelated (i.e., convergent validity). At the same time, a set of items on a certain construct was less correlated with other sets of items on other constructs (i.e., discriminant validity). Establishing convergent and discriminant validity was a critical step to evaluate levels of quality of life of families with an individual with disability and to provide necessary supports for these families. Future research on criterion validity is needed to further conceptualize the latent constructs of the FQOL Scale-21 by identifying their relations with other criterion measures. In addition, future research should examine what characteristics of children or family influence or moderate the level of FQoL. Third, multiple goodness-of-fit indices supported a good fit to the data for both lower-order factor and higher-order factor models. When a statistical comparison between two nested models was performed, we found that a higher-order model showed a better fit of the model than a lower-order model, providing support for stakeholders to use a composite score of the overall FQoL and/or four subdomain scores (i.e., family interaction, parenting, emotional well-being, and physical/material well-being) depending on their purposes of assessment. When interpreting this finding, however, we should acknowledge that the p-value of the nested model was at the borderline of the decision criterion (p = .016; we used the significance level .01 with sample size of 400). This finding suggests a need for future replication research to further elaborate the meaning and interpretation of scores of the FQoL Scale-21. The validation process for assessments such as this scale requires “accumulating relevant evidence to provide a sound scientific basis for the proposed score interpretations” (AERA et al., 2014, p. 11). Thus, the findings of the current study provided a good starting point to gather pertinent information on validity of the FQoL Scale-21 scores in Taiwanese context.
This study provides empirical evidence to document and understand the overall FQoL of Taiwanese families. The FQoL conceptual framework provides an ecological perspective in interpreting the results. Nevertheless, there are further research, policy, and practice questions related to FQoL for Taiwanese families.
Future directions for research
There are additional steps to be taken related to continued validation of the measure. First of all, it is necessary to conduct a study that examines how well the measurement model of the FQoL Scale-21 is generalizable across subgroups of families of students with disabilities. Since the FQoL Scale-21 was designed to be administered in heterogeneous family groups, the equivalent measurement properties should be established across subgroups of the population (e.g., early childhood vs. youth and young adults with disabilities, different disability subgroups, etc.; Brown, T. A., 2015). In addition, it would be interesting to compare measurement properties of the FQoL Scale-21 internationally to examine if the underlying constructs are measured in the same operational manner across different cultural contexts.
Second, it is necessary to conduct studies that compare latent mean levels of satisfaction/ quality of life on each domain of the FQoL scale (e.g., family interaction, parenting, emotional well-being, and physical/material well-being) or latent correlational relations among the four domains across subgroups of the population. This is particularly important because the findings can be used to improve services for families of students with disabilities, inform system decisions, enhance policy, and ultimately lead to positive outcomes. To build a system of family support, effort should be made to understand how input (i.e., family needs, strengths, and priorities) and systemic factors (i.e., societal values, policy, systems, and programs) interact with the influential factors and how input and systemic factors simultaneously impact the FQoL outcomes.
Third, the results of this study suggest that the FQoL Scale-21, i.e., a measure of family quality of life stripped of items related to family perceptions about the adequacy of services and supports, has sufficient validity for use in outcome studies to evaluate the influence of services and supports on families. Therefore, this measure holds promise for use in evaluations of supports and service delivery on FQoL as an outcome. Such research, on services that are amenable to change rather than stable demographic characteristics of the child and family, may provide information leading more directly toward identifying components of service and support models that are effective in improving FQoL (Kyzar et al., 2012). Researchers should continue to explore types (i.e., emotional, informational, instrumental, and physical) and sources (i.e., formal and informal) of support in terms of their effects on FQoL. By doing so, researchers will be able to understand what affects FQoL and how to identify strategies and resources to improve FQoL accordingly.
Fourth, although there are studies verifying that family characteristics and parent participation do influence child outcomes in terms of academic achievement, emotional health, and behaviors (Davis-Kean, 2005; Epley, Summers, & Turnbull, 2011), more research is needed to confirm the relation between family and child outcomes.
Last, an FQoL tool with strong psychometric properties has scientific value in terms of comparing and contrasting systemic factors (e.g., social values, policies, systems, and programs) affecting FQoL. Chiu and colleagues (2013) have summarized research on systemic factors related to FQoL to date which is quite limited nationally and especially internationally. Given the socialecological model of disability, an FQoL tool with sound cross-cultural psychometrics enables researchers to respond to U.S. leaders' challenges to address context, broadly, as both an intervening variable and integrative concept (Shogren, 2013), as well as to address public policy (Shogren et al., 2009), specifically in this case related to family outcomes. For example, the effect of public policy on FQoL can be analyzed by documenting FQoL outcomes in countries that are similar and different in terms of policy incentives and disincentives for family support.
Future directions for policy and practice
Having a conceptual framework and an instrument with strong psychometric properties not only aids in research but also helps identify areas for improvement that derive from Regulations of Services for Family Caregivers of Individuals with Disabilities (2012). Policy leaders may choose to use FQoL research data as one factor in determining who the beneficiaries of family support policy should be and in determining whether family support service-delivery systems are effective in enhancing FQoL in the judgment of families.
The FQoL conceptual framework has implications for researchers, practitioners, and policy makers in understanding family outcomes within context of systemic, family-unit, individual-member, and support factors (Chiu et al., 2013; Zuna et al., 2010). The findings of this study suggest that the FQoL-21 has the potential to be used as an indicator of positive outcomes in intervention evaluation, policy making, and service delivery. Furthermore, the current study is essential in establishing validity and reliability of the FQoL Scale-21 in Taiwan.