Abstract
This article reports on a confirmatory factor analytic study of an adapted version of an instrument designed to assess family functioning of Chinese families. The Chinese Family Assessment Instrument, originally designed for completion by adolescents, was adapted for completion by parents. A sample of 700 parent dyads of elementary school children (382 girls and 318 boys) completed the adapted questionnaire. Initial factor analyses showed that the existing five-factor structure used for adolescents’ responses was not a good fit for these data. Instead, a four-factor solution emerged where the factors were positive family functioning, negative family functioning, tolerance for family members, and parental understanding. This structure was the same for both mothers and fathers. Further studies of the Chinese Family Assessment Instrument parent adaptation are required to test the factor structure that emerged. Following such studies, validation studies will be required.
The rapid economic growth and increasing Westernization experienced by China since the adoption of the “Open Door Policy” in the 1980s has resulted in a shifting cultural climate characterized by conflict between modern and traditional values (Lee & Lee, 1996; Ma, 2007). These differences have direct implications for family life as parents bring up children in an environment that is radically different from that in which they grew up (Ma, 2007, 2008). This is exemplified by the fact that while many young people in China today are embracing autonomy, independence, and personal success that are the hallmarks of individualistic Western cultures, their parents were raised to value collectivist ideals such as cooperation, obedience, and group cohesion. These intergenerational differences are likely to impede effective communication between parents and children (Jung & Forbes, 2007) and therefore have an impact on family functioning and family life.
Another factor that may have an impact on family functioning in China is the “one child only”’ policy, introduced in 1979 to curb the growth of China’s rapidly growing population. As a result of this policy, many urban-dwelling Chinese children are likely to be only children. The lack of siblings, as well as the weakening of familial bonds, may threaten traditional understandings of self that are embedded in kinship relationships (Falbo & Poston, 1993). Indeed, as well as having no siblings, many only children are likely to have no aunts and uncles and, therefore, no cousins. On the other hand, they are likely to be the focus of attention of their grandparents as well as their parents, since they are likely to be the only grandchild as well as the only child.
Given these changes, it seems important to understand the functioning of modern-day Chinese families. However, Shek (1998, 2001a, 2001b) argued that Chinese people tend to see their families in an undifferentiated manner because they lack the language to describe the related experiences, raising the possibility that it may be difficult to assess family functioning. In an attempt to provide a means of doing so, he created the adolescent-report Chinese Family Assessment Instrument (C-FAI; Shek, 2002), the first indigenous instrument to examine family functioning in Chinese populations. In developing the C-FAI, Shek questioned whether dimensions of healthy families derived from Western contexts were applicable to Chinese culture. Three possible reasons as to why the Western dimensions of happy family life may not be applicable were proposed: (a) in Chinese families there is a stronger emphasis on harmony in family and social relationships, (b) Chinese families are also more hierarchical in structure than families in Western cultures, and (c) face-saving is an important concern in family or social problem solving (Allison, 1997; Ho, 1996; Shek & Lai, 2000; Ting-Toomey, 1988). In support of this argument, Shek (2001a, 2002) reported that the factor structures of translated Western instruments used in China usually do not hold up.
Shek (2002) based the C-FAI on the themes derived from semistructured interviews with parents and adolescents from 416 families in Hong Kong about the characteristics of a happy family (see Shek, 2001c; Shek & Chan, 1998). Absence of conflict, interpersonal harmony, mutuality, connectedness, and positive parent–child relationships emerged as the major themes. Shek and Chan (1998) noted that emotional expressiveness and communication, assumed characteristics of happy Western families, played a lesser role.
The C-FAI comprises 33 items that make up five stable and reliable dimensions: (a) mutuality (e.g., mutual support among family members), (b) communication and cohesiveness (e.g., family members talk to each other, few barriers among family members), (c) conflict and harmony (e.g., frequent fighting among family members), (d) parental concern (e.g., parents love their children), and (e) parental control (e.g., parental control is too harsh). Siu and Shek (2005) argued that these dimensions reflect the major features that Chinese parents and children consider to be important, adding that the parental control factor could be seen as a subset of the flexibility dimension in Western cultures, and the conflict and harmony dimension reflects the importance attributed to the “absence of conflict.” Furthermore, they suggested that parental concern and parental control are related to functioning of parents, which demonstrates the role of parents in family functioning in Chinese families.
Shek (2002) and Siu and Shek (2005) have demonstrated that the C-FAI possesses excellent psychometric properties. It is internally consistent (Cronbach’s α = >.90), stable across time (test–retest reliability = .84), and has acceptable convergent and construct validity in both clinical and nonclinical samples. In a further recent study with a sample of 3,649 Hong Kong students, Shek and Ma (2010) found support for the factor structure of the C-FAI, concluding that the instrument is an objective measure of Chinese family functioning, with high factorial validity that can be used in family practice contexts with Chinese families.
Although the C-FAI appears to be a useful advance in terms of the assessment of functioning within Chinese families, it is limited to the extent that it only assesses this functioning from the perspective of the adolescent members. Research suggests that it may also be important to consider the perceptive of parents. Discrepancies between parents’ and their adolescent offspring’s perceptions of family functioning have been consistently reported in Western studies, with adolescents tending to have a more negative view of the family than their parents (Carlson, Cooper, & Spradling, 1991; Noller & Callan, 1986; Ohannessian, Lerner, Lerner, & von Eye, 2000). In Carlson et al.’s (1991) study, adolescents rated their families as significantly less cohesive, less expressive, and less independent than did their parents. However, there were no significant differences in perceptions of family organization and control. Similarly, Ohannessian et al. (2000) assessed perceptions of family cohesion, adaptability, and adjustment. They found that adolescent girls reported significantly lower levels of family cohesion and adjustment than did both their mothers and fathers. Boys’ perceived level of family cohesion was also significantly lower than that of their mothers and fathers, and while they reported significantly lower levels of family adjustment than their mothers, they did not differ from their fathers on perceptions of this dimension. There were no significant differences in adolescent and parent ratings of adaptability. In another study of perceptions of family functioning, Safyer, Thompson, Maccio, Zittel-Palamara, and Forehand (2004) found that parents reported significantly higher levels of family functioning than did their adolescent children. Specifically, discrepancies were found in ratings of communication, rituals, worry, and positive affect. However, there were no differences between parents and adolescents in their perceived level of family conflict.
The importance of being able to assess concordance between parent and adolescent ratings of family functioning is demonstrated by research showing that discordance is associated with increased levels of stress, conflict, and problems within the family (Olson et al., 1983; Reiss, 1981). Extreme disagreement leads to limited interaction between family members and has an adverse effect on the adjustment of individual family members (Shek, 1999). For example, disagreement between parents and adolescents on family functioning has been related to lower levels of adolescent self-esteem and self-competence (Carlson et al., 1991) and higher levels of deviance (Davey, 1994), anxiety, and depression (Ohannessian, Lerner, Lerner, & von Eye, 1995). In contrast, agreement between family members can create positive affect and lead to improved interaction and adjustment (Carlson et al., 1991).
In light of the above, a useful advance in relation to the C-FAI would be to adapt it for parent report so that the perspectives of various members of Chinese families can be compared. Indeed, Siu and Shek (2005) make the point in their article that it would be interesting to use the C-FAI to determine how family functioning may differ based on different perspectives. They note that this may be particularly important in China, where the use of family therapy is developing rapidly. Thus, the development of the C-FAI to include a parent version would broaden its use as both clinical and research tools. Given that the dimensions reflected in the C-FAI reflect the major features that both Chinese parents and children consider to be important to a happy family (Siu & Shek, 2005), it can be expected that an adaptation for parent report would not be inappropriate. The purpose of the current study was to adapt the C-FAI for parent report and assess whether or not the factor structure outlined by Shek (2002) was maintained. This would open the door for validation studies of the adapted instrument.
Method
Participants
Participants in the study were both parents of 729 primary school children in Chengdu, the capital of Sichuan province, located in southwest China. When outliers were excluded, data from 382 parent dyads for girls and 318 parent dyads for boys remained for analysis. The average age of the children was 8.7 years (SD = 1.69), there were no significant differences between the females and males with regard to age distribution.
Materials
As stated above, the C-FAI includes 33 items that make up five stable and reliable dimensions: mutuality (12 items), communication and cohesiveness (9 items), conflict and harmony (6 items), parental concern (3 items), and parental control (3 items). Respondents use a 5-point response format to rate their agreement with the items, with 1 = strongly agree and 5 = strongly disagree. Items include statements such as “Parents often talk to their children,” “Family members care about each other,” and “There is a lack of harmony among family members.” Responses to each item on the subscale are added to obtain the subscale score, and a total score can be obtained by adding the subscale scores. Higher scores reflect greater dysfunction within the family.
The C-FAI was designed for completion by adolescents. In the present study, the wording was modified slightly to make it amenable to parent response. This modification was made by a bilingual academic (XX) and double checked by a postgraduate Chinese student studying in Australia. For example, questions from the adolescent version such as [In my family] “In general, parents’ discipline toward their children is over vigorous/too strict” and “Parents generally understand their children’s thinking (mentality)” were changed to “In general, I am very strict toward my child” and “In general, I am able to understand my child’s thinking.”
Procedure
The project was conducted in accordance with approval gained from the University Human Research Ethics Committee and within the rules and guidelines for research conducted through Sichuan Normal University. Data were collected in Chengdu, China. Prospective participants were given a pack that included an invitation to participate in the study, the plain language statement, the questionnaire set, and a return envelope. One pack for each parent was sent home with each child in the selected classes. If parents were willing to participate in the study, they completed the questionnaire separately and returned it in a sealed envelope to the school. The completed questionnaires were then collected from the schools. Eight hundred surveys were distributed; the return rate was 94%.
Analytic Procedure
Using Mplus 6, a multiple group confirmatory factor analysis (CFA) was conducted to test whether the established factor structure of the scale found in children was also appropriate for the mothers’ and fathers’ data. As recommended by Hu and Bentler (1999), model fit was evaluated using four fit indices: comparative fit index (CFI), Tucker–Lewis index (TLI), root mean square error of approximation (RMSEA), and standardized root mean square residual (SRMR). Consistent with the procedures outlined by Ho (1996), the model was fitted simultaneously for men and women. An unconstrained model (where loadings and intercepts are free to vary across mothers and fathers) was compared with a constrained “measurement invariance” model (where loadings and intercepts are constrained to equality across mothers and fathers). Measurement invariance is established if the fit of the constrained model does not fit the data significantly worse than the unconstrained model. Nested models are traditionally compared using chi-square statistics. However, with large sample sizes such as in this study, chi-square becomes too sensitive and can show significant differences between models even when measurement noninvariance is present (Meade, Johnson, & Braddy, 2008). Given this, nested models were compared using difference in CFI as recommended by Meade et al. (2008).
Results
Results from the multiple group CFA showed that although the five-factor structure was invariant across mothers and fathers (ΔCFI = .01), it was not a good fit for the data, χ2 (1, 026) = 8,012, p < .05, CFI = .66, TLI = .65, RMSEA = .10, SRMR = .10. Given this, an exploratory factor analysis (EFA) was conducted, and the resulting factor structure was then validated using multiple-group CFA.
Results from the EFAs were evaluated using fit indices, scree plots, and the extent to which the results could be interpreted substantively. Inspection of fit indices and the scree plots for both mothers and fathers suggested a four- or five-factor solution was most appropriate for both samples (see Table 1 for fit indices). However, the five-factor solutions for mothers and fathers contained one factor that was defined almost completely in terms of cross-loading items, making interpretation of the factor structure difficult. In the four-factor solution, a number of items had high cross-loadings or showed poor loadings (<.30) across all factors and were removed from the analysis. Subsequent EFAs on the remaining items supported the same four-factor solution, which was completely consistent across mothers and fathers.
Fit Indices for Exploratory Factor Models of the Adapted C-FAI.
Note. C-FAI = Chinese Family Assessment Instrument; CII = comparative fit index; TLI = Tucker–Lewis index; RMSEA = root mean square error of approximation; SRMR = standardized root mean square residual.
A second multiple-group CFA was conducted to validate the four-factor solution revealed by the EFA. Results showed that the model was a good fit for the data, χ2 (320) = 1,730, CFI = .89, TLI = .88, RMSEA = .08, SRMR = .08. Furthermore, the model was invariant across mothers and fathers (ΔCFI = .01). The four factors were interpreted as positive family functioning (8 items), negative family functioning (conflict, 3 items), tolerance of family members (3 items), and a parent’s understanding of their child (4 items). See Table 2 for factor loadings.
Factor Loadings for the Adapted C-FAI.
Note. C-FAI = Chinese Family Assessment Instrument.
Reliability of the Adapted C-FAI and Its Subscales
Based on the subscale composition, the overall internal consistency of the 18-item C-FAI was very good (Cronbach’s α = .83 for mother and father report). Cronbach’s alpha for the four subscales were also generally strong. Cronbach’s alphas for father report and mother report for positive family functioning were both .82. For negative family functioning they were α = .60 and .59, respectively; for tolerance of family members they were .75 and .80, respectively; and for parents’ understanding of their child they were .77 and .81, respectively.
Table 3 summarizes the scores of mothers and fathers of boys and mothers and fathers of girls on each subscale of the C-FAI, as well as the total scores. Since there are no normative data for this questionnaire when it is completed in mainland China, and completed by adults, the scores cannot be interrupted in terms of level of family functioning.
Means, Standard Deviations, and Ranges of Mother and Father Reports of Family Functioning According to Gender of Their Child.
Note. C-FAI = Chinese Family Assessment Instrument.
Discussion
The C-FAI is a culturally specific scale designed to assess adolescents’ perceptions of functioning in Chinese families. In this study, we adapted the scale for completion by parents because it is often appropriate to assess both parent and adolescent perspectives of family functioning, since research shows that they do not necessarily concur. Given that the dimensions of family functioning assessed by the scale were derived from interviews with parents and adolescents, we argued that they are relevant for assessment of parent perceptions and those of adolescents. Our purpose was to establish whether or not the factor structure described by Shek (2002) and confirmed with his collaborators (Shek & Ma, 2010; Siu & Shek, 2005) would be maintained in the adapted parent-report C-FAI.
Our analyses indicted that the adapted parent-report C-FAI produced a different structure to that established by Shek and colleagues for the adolescent version. There was no difference between mother and father reports in this emergent structure. This may suggest that mothers and fathers experience family functioning in similar ways. The dimensions were labeled positive family functioning, negative family functioning (conflict), tolerance of family members, and a parent’s understanding of their child. The reliabilities of the subscales and total scale were strong, apart from the negative family functioning subscale (Cronbach’s α coefficients were .60 and .59 for mothers’ and fathers’ reports, respectively). These lower coefficients may be due to the small number of items on this subscale or the negative nature of the items being answered inconsistently by the participants. It is also possible that there may be different types of negative functioning in families, so future directions for researchers could be to include more items on this subscale and assess their fit. It is also notable that coefficient alpha is just one of the criteria that can be used to determine the reliability of a measure. It should be noted that the factor loadings were all very good.
The different factor structure found for parent respondents compared with adolescent respondents may be due to a number of factors. One is that parents may see family functioning through a different lens than adolescents. This may be especially the case in China were the demographic, economic, and cultural transitions discussed earlier in this article may have an impact on generations in different ways. Another is that our data were collected in Chengdu, in South Western China, whereas the original scale was developed in Hong Kong. Family functioning, and life in general, in Hong Kong may have been affected by the long history of contact with Europeans via colonization and the exclusion from the constraints of the communist era that dominated before liberalization on the mainland. A third possibility is that in the current study, the participants were parents of primary school–aged children, whereas the C-FAI was designed for adolescents. Because the parents in this study had younger children, and families were likely to be at a different stage of the family life cycle, the nature of family functioning and family dynamics may be different. As noted earlier, Shek (1998, 2001a, 2001b) suggested that Chinese people tend to see their families in an undifferentiated manner because they lack the language to describe the related experiences. This could explain why the “positive” and “negative” dimensions of family functioning emerged from the findings.
It is notable that on both this parent version and the adolescent version of the C-FAI the number of items on some subscales is small. On the original C-FAI, two subscales, concern and control, each have only three items. In the adapted version, negative family functioning (conflict) and tolerance of family members each have only three and parent’s understanding of their child has four. Given this, it may be appropriate to consider the introduction of some additional items to strengthen the reliability of those subscales.
Although this study did not replicate the factor structure of the C-FAI, it provides a basis for further exploration of measurement of family functioning in China. There are very few Chinese scales assessing family functioning, perhaps none based on parent report, so this study makes an addition to the literature. However, there are several limitations to the study. First, although the sample was relatively large, it was not random, as it was recruited through a small number of schools in one city in China. Therefore, the findings may not be generalizable. Second, although the CFA identified a different set of four factors in the parent version of the C-FAI to those in the adolescent version, further studies are needed to assess the psychometric properties of these four factors. For example, it would be important to assess the validity of the factors against other measures of family functioning. It is notable though that at this stage there are no other indigenous Chinese measures of family functioning that could be used as a benchmark. It may be possible to use the measure in a clinical context to ascertain whether clinically referred families, whose family functioning is rated by clinicians as being disrupted or impaired (e.g., marked by parent–adolescent conflict or poor problem solving), score differently to our community sample. It would also be useful to collect test–retest data on the parent version of the C-FAI.
Conclusion
This study investigated the factor structure of an adapted version the C-FAI. Originally designed to be a culturally sensitive measure of family functioning from the perspective of adolescents, our adapted version was for completion by parents. A four-factor structure, supported by both mother and father data, that varied from that of the original was obtained. The factors were positive family functioning, negative family functioning, tolerance for family members, and parental understanding. Further investigations of the version of the instrument are required.
Footnotes
Declaration of Conflicting Interests
The authors declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.
Funding
The authors received no financial support for the research and/or authorship of this article.