A meta-analytic investigation of the relationships between different dimensions of family strengths and personal and family well-being

Objective: Evaluate the relationships between five family strengths dimensions (commitment, communication, cohesion, coping & competence) and personal and family well-being. Background: Noted family strengths experts contend that families who describe themselves as strong share a number of broad qualities or traits. Family strengths are viewed as internal resources that are hypothesized to be related to enhanced personal and family well-being. Method: Meta-analysis was used to determine the effect sizes between different family strengths dimensions and both parent and family well-being. The meta-analysis included 14 studies (N = 3,491 participants) conducted in 10 countries. The focus of analysis was the sizes of effects between each family strengths dimension and both parent and family well-being and whether the strengths of relations between family strengths and well-being differed for each family strength dimension. Results: Publication bias analysis indicated no differences in the sizes of effects for peer-reviewed and non-peer-reviewed research reports. Each of the family strengths dimensions was significantly related to both personal and family well-being. The size of effect for the relationship between family strengths and family well-being was larger than that for personal well-being. The between type of family strengths dimension analyses showed that the size of effect between family competence and personal and family well-being was smaller than those for each of the other family strengths dimensions. Conclusion: The findings support the hypothesis that different dimensions of family strengths are important predictors of personal and family well-being. Results also indicated the strength of the relationship between family strengths and well-being differed between the five family strengths dimensions.


Introduction
Noted positive psychology theorists emphasize the role human strengths play in engaging people in positive life experiences and how those experience contribute to healthy wellbeing and functioning (e.g., Aspinwall & Staudinger, 2003;Linley, Joseph, Harrington, & Wood, 2006;Lopez, Pedrotti, & Snyder, 2018). Lopez (2009) and Sheridan, Warnes, Cowan, Schemm, and Clarke (2004) both argued that family strengths are important for healthy family and family member functioning but noted a lack of attention to these relationships in the positive psychology literature. Lopez (2009), for example, noted that "Despite the good being done on positive psychology and family functioning…we know little about how family strengths and how [strengths] come together to define a healthy family" (p. 692).

Family strengths
"Family strengths are those relationship qualities that contribute to the emotional health and well-being of the family" (DeFrain & Stinnett, 2002b, p. 637). This succinct and cogent statement includes two interrelated constructs: The qualities that define a strong or healthy family and how those qualities are related to or influence family well-being. Numerous researchers and practitioners have focused their work on identifying the qualities of strong families (e.g., Beavers & Hampson, 1990;Curran, 1983;Otto, 1962;Stinnett & DeFrain, 1985). There is, surprisingly, little empirical evidence demonstrating a relationship between the presence of different qualities of strong families and different dimensions of well-being.
Efforts to identify the qualities of strong families has resulted in lists of the beliefs, values, behavior, and interactional patterns that are commonly present in successful and healthy families. For example, Stinnett and DeFrain (1985) identified six qualities of strong families, Curran (1983) identified 14 traits of healthy families, and Otto (1962) identified 11 components of strong families. Table 1 includes lists of the qualities of strong and healthy families identified by these family strengths experts. DeFrain and Stinnett (2002b) noted that despite differences in the conceptual frameworks that led to the identification of different qualities of strong families, "researchers around the world have found remarkable similarities in families in different cultures. Families that describe themselves as strong commonly share a number of broad qualities or traits" (p. 639). This can easily be seen in Table 1 where many of the qualities or traits of strong families in the three lists are much the same. For example, all three sets of characteristics include the ability to communicate positively and openly as one quality of strong and healthy families. A content analysis of 10 compilations of the qualities of strong families resulted in a list of 12 common traits and characteristics .  (Otto, 1962) Strong Families (Stinnett & DeFrain, 1985) Healthy Families (Curran, 1983)  scale to measure family strengths where strengths were empirically related to personal, family, and child well-being found, with only one exception (Arshat & Baharudin, 2014), that the majority of studies of the relationships between family strengths and well-being used the Family Functioning Style Scale (FFSS) to measure family strengths. The literature search identified 33 FFSS studies that related family strengths to personal, family, and child well-being. The results were meta-analyzed to identify the sizes of effect between the total family strength scores and the different types of well-being (Dunst, Serrano, Mas, & Espe-Sherwindt, in press).

Qualities of strong families
The 33 studies were conducted in 12 countries and included more than 7000 participants. The participants included biological parents, foster parents, adoptive parents, graduate students, and adolescents. The independent variable was the total FFSS scores. The dependent variables included five dimensions of well-being: Personal well-being (e.g., Abidin, 1997;Bradburn, 1969), personal belief appraisals (e.g., Gibaud-Wallston & Wandersman, 2001;Rosenberg, 1965), positive parenting practices (e.g., Buri, 1991;Groza, Ryan, & Cash, 2003), family well-being (e.g., H. I. McCubbin, Comeau, & Harkins, 1981;Summers et al., 2005), and child well-being (e.g., Achenbach & Ruffle, 2000;Epstein & Sharma, 1998). The weighted average correlations between the total FFSS scores and each dimension of well-being were used as the size of effects between family strengths and well-being. Family strengths were related to each dimension of well-being and ranged between r = .26 (95% CI = .18, .33, p = .0000) for child well-being and r = .54 (95% CI = .43, .63, p = .0000) for family well-being. The results indicated that family strengths were related to well-being as hypothesized DeFrain and Stinnett (2002b) and others (e.g., Greeff & le Roux, 1999;Lingren et al., 1987) and were similar to findings in other meta-analyses of strengths-related family functioning measures (e.g., Leeman et al., 2016;Van Schoors et al., 2017). Moore, Chalk, Scarpa, and Vandivere (2002) noted, however, the need for additional research to determine which kinds of family strengths are related to which types of child, parent, and family well-being.

Purpose of the study
This paper includes additional analyses of a subset of studies in the Dunst et al. (in press) meta-analysis. Fourteen of the 33 studies in the meta-analysis included the correlations between different family strengths dimensions and either or both personal and family well-being. Most investigators used results reported in Trivette, Dunst, Deal, Hamer, and Propst (1990) and Trivette, Dunst, Deal, Hamby, and Sexton (1994) for assessing the relationships between five FFSS subscale dimensions (scores) and the well-being measures. Table 2 includes the subscales, brief descriptions of each of the family strengths dimensions, and examples of scale items. The subscales were identified by factor analyses of the 26 FFSS scale items and were found to be inter-related in an expectant manner. This was consistent with observations by Otto (1962) who noted that "family strengths are not isolated variables, but form clusters and constellations which are dynamic, interrelated, and interacting" (p. 80).
Findings from the meta-analysis of the relationships between each of the family strengths dimensions and personal and family well-being are reported in this paper. The findings were expected to (1) shed light on how different dimensions of family strengths are related to personal and family well-being and (2) ascertain if family strengths are related to healthy well-being and functioning in the same was as individual human strengths are related to healthy functioning (e.g., Baumann & Eiroa-Orosa, 2017;Macaskill & Denovan, 2014).

Method
Detailed descriptions of the search terms, search sources, inclusion criteria, independent and dependent measures, and methods of analysis are included in Dunst et al. (in press). The search methods, participant characteristics, and study measures are briefly described next to place the analyses and results in methodological and procedural context. The American Psychological Association Reporting Standards for Quantitative Meta-Analyses were used to structure the conduct of the research synthesis and to report the results from the meta-analysis (Appelbaum et al., 2018).

FFSS Constructs and Examples of the Subscale Items Family Commitment
Ability of each family member to behave in ways that contribute to the well-being of the family unit (e.g., Family members can depend on one another) Family Communication Ability to communicate and interact with one another in positive and constructive ways (e.g., We pitch in and help one another) Family Cohesion Ability to "stick together" in both good and bad times (e.g., We make personal sacrifices to benefit the family) Family Coping Ability to deal effectively with adverse life events (e.g., We believe it is good in even the worst situations) Family Competence Ability to obtain informal and formal support and resources to meet family needs or achieve desired goals (e.g., Our family is able to ask for outside help when needed)

Search terms
The primary search terms were "family functioning style scale" and different variations (e.g., "family functioning style" AND "scale OR instrument OR survey"). The secondary search terms were different variations of the scale names used by investigators in their primary studies (e.g., "family functioning style questionnaire" and "family functioning style survey"). The non-English names of the FFSS were also searched to locate studies (e.g., Escala de Estilos de Funcionamiento Familiar).

Search sources
Seven different electronic databases were searched for FFSS studies (e.g., PsycNET, ProQuest Central, ProQuest Dissertations and Theses, Google Scholar). These were supplemented by hand searches of the reference sections of all retrieved papers and searches for papers citing the primary sources of the FFSS Deal, Trivette, & Dunst, 2009;Trivette et al., 1994;Trivette et al., 1990).

Inclusion criteria
Studies were included in the meta-analysis if the correlations between any of the FFSS subscale scores and either or both personal and family well-being were assessed in the research reports. (Too few studies in the larger meta-analysis included the correlations between the FFSS subscale scores and personal beliefs, positive parenting practices, or child well-being to be meta-analyzed.) No limitations were placed on the type of research report or the language in which the research reports were published.

Search results
The search procedures identified 425 non-duplicated papers that referenced the FFSS. The majority (N = 392, 92%) were excluded because they did not include the correlations between the FFSS and one or more types of well-being measures. Among the studies reporting the correlations between FFSS scores and well-being measures (N = 33), 14 included the correlations between three or more FFSS subscales and well-being. Table 3 shows selected characteristics of the 14 research reports that met the inclusion criteria. The studies included 3,491 participants. Three studies were conducted in the USA (Algood, 2013;Trivette et al., 1994;Trivette et al., 1990), two in Germany (Sarimski, 1997a(Sarimski, , 1997b, two in Portugal (Magina, 2011;Santo, 2017), and one each in Canada (Franks, 2007), Chile (Soto, 2013), India (Ara & Shah, 2015), Panama (Zelenka, 1994), Slovakia (Banovcinova & Gal, 2019), Spain (Guijarro, 2010), and South Africa (Koen, van Eeden, & Rothmann, 2013). Three of the studies were published in peer-reviewed journals, seven were dissertations or theses, one was a book chapter, and one was a conference proceeding.
The participants were the biological mothers or both the biological mothers and fathers in 11 studies, adolescents in two studies, and graduate students in one study. The average age of the adult participants ranged between 23 and 42 years. These participants completed, on average, 11 to 17 years of formal education. The average age of the adolescent study participants was 16 years. They completed an average of 11 to 13 years of formal education.

Measures
The primary study investigators used different numbers of FFSS subscales in their studies. Three studies included three subscales, five studies included four subscales, and six studies included all five subscales (Table 2). Four studies included personal well-being measures, five studies included family well-being measures, and five studies included both personal and family well-being measures. The two types of well-being measures differed in terms of the attributional targets of the scale items (Bugental, Johnston, New, & Silvester, 1998). The targets of appraisals of the personal were the participants' judgments of his or her personal psychological health. The targets of appraisals of the family well-being measures were the participants' judgments of family psychological health and quality of life.
Seven different scales were used to measure personal well-being and five different scales were used to measure family well-being. All of the scales were self-report measures where the scale items were rated on either a 4-point or 5-point Likert scale. Most scales have been widely used to assess different dimensions of personal and family well-being (see e.g., Cooke, Melchert, & Connor, 2016;Olson et al., 2000;Swain, Harrigan, & Woog, 1995;Zhang & Chen, 2019) and have well established psychometric properties.
The instruments used to measure personal well-being were the Parenting Stress Index (Abidin, 1997), Psychological Well-Being Index (Bradburn & Caplovitz, 1965), State-Trait Anxiety Inventory (Spielberger, Gorusch, & Lushene, 1970), KIDSCREEN Psychological Well-Being Subscale (Ravens-Sieber et al., 2005), World Health Organization Quality of Life Psychological Health Subscale (World Health Organization, 1996), Perceived Stress Scale (Cohen, Kamarck, & Mermelstein, 1983), and the Rosenberg Self-Esteem Scale (Rosenberg, 1965). The instruments used to measure family well-being were the Family Inventory of Resources and Management Mastery and Health Subscale (H. I. McCubbin et al., 1981), Family Satisfaction Scale (Olson & Willson, 1982), Family Assessment Measure-III (Skinner, Steinhauer, & Santa-Barbara, 1995), Beach Center Family Quality of Life Scale (Summers et al., 2005), and the Family Quality of Life Scale (Troster, 2004). Note: Participant mean age and mean years of school completed (education) were estimated in a number of studies based on available information in the research reports. Percent married includes participants living with a partner. NR (not reported) indicates that that the information was not reported or insufficient information was provided to estimate the participant characteristics. NA indicates that the characteristic was not applicable due to the sample that was the focus of investigation (e.g., marital status of adolescents).
The dependent measures differed in terms of whether a higher score indexed either better well-being or attenuated well-being. In the latter case, the signs of the correlation coefficients were reversed so that higher scale scores reflected better well-being (e.g., the correlations between family strengths and parenting stress were expected to be associated with less stress resulting in a negative correlation between measures).

Methods of synthesis
The meta-analysis was performed using Meta-Essentials (Suurmond, van Rhee, & Hak, 2017;Van Rhee, Suurmond, & Hak, 2015). Fisher's z transformation of the zero-order correlations between the FFSS subscale scores and well-being measures were used to compute the weighted average sizes of effect for the independent-dependent scale relationships. These z-scores were converted back to correlation coefficients for reporting purposes. Random effects models were used to ascertain the average sizes of effects given the heterogeneity of the study and sample characteristics ( Table 3) and because of the differences in the scales used to measure personal and family well-being.
Publication bias was examined to determine if the type of research study was related to effect size differences. Publication bias, or the file drawer effect, "occurs when results of published studies are systematically different from results of unpublished studies" (Song, Hooper, & Loke, 2013, p. 71). Because the studies in the meta-analysis included a mix of published and unpublished research reports (Table 3), we compared the sizes of effects for peer-reviewed journal articles with non-peer-reviewed research reports to determine if there was any publication bias.
The analyses for the family strengths--well-being relationships included the number of effect sizes for each FFSS subscale (k), number of study participants (N), the average weighted effect size (r), the 95% confidence interval (CI) for the average effect size, the Ztest for determining how much the average correlation differs from zero, and the p-value associated with the Z-statistic. The heterogeneity in the sizes of effects between studies was evaluated using the I2 where values close to zero indicate similar study results and values close to 100 indicate inconsistency in the study results (Higgins, Thompson, Deeks, & Altman, 2003).
Between FFSS subscale comparisons were assessed using the QBetween (QB) procedure. This test is analogous to a one-way ANOVA for within-study comparisons (Hedges, 1994). QB was also used to determine if the personal and family well-being results could be combined or needed to be assessed separately for the two types of well-being.

Types of well-being comparison
The average size of effect for the relationship between family strengths and personal wellbeing is r = .28, 95% CI = .25, .31, Z = 18.00, p = .0000, and the average size of effect for the relationship between family strengths and family well-being is r = .44, 95% CI = .38, .48, Z = 16.19, p = .0000. The between type of well-being comparison was significant, QB = 24.25, df = 1,80, p = .0000, indicating that the relationship between family strengths and the two well-being constructs are not the same.
Although both average correlations differed significantly from zero as evidenced by significant Z-test results, the size of effect for the relationship between family strengths and family well-being was larger compared to the size of effect between family strengths and personal well-being. All other analyses are therefore reported separately for the two types of well-being. Table 4 shows the results for the relationships between the five family strengths dimensions and both personal and family well-being. Each of the family strengths dimensions was significantly related to both types of well-being as evidenced by significant Z-test results.

Family strengths dimensions and well-being
The heterogeneity tests indicated that there were similar results in the different research reports for personal well-being but considerable inconsistency in the results for 4 of the 5 family strengths dimensions for family well-being. Examination of the forest plot for the family strengths--personal well-being relationships showed that the confidence intervals for all of the sizes of effects overlapped. The forest plot for the family strengths-family well-being relationships showed that the confidence intervals for two studies (Banovcinova & Gal, 2019;Koen et al., 2013) did not overlap with many of the other studies. These two studies had large sample sizes, and as noted by Rucker, Schwarzer, Carpenter, and Schumacher (2008), inconsistency tends to increase as sample size increases. Note: k = Number of effect sizes, N = Number of study participants, r = Average weighted effect size, and I 2 = Inconsistency (homogeneity) in the effect sizes in the primary research reports.

Between type of family strengths comparisons
The between type of family strengths subscale comparisons differed significantly for both personal well-being, QB = 10.49, df = 4,38, p = .0330, and family well-being, QB = 11.68, df = 4,34, p = .0200. In both analyses, these differences appear to be the result of the smaller sizes of effects for the relationship between family competence and personal and family well-being compared to the sizes of effect for the relationships between each of the other family strengths dimensions and the two well-being measures. This was confirmed by analyses without family competence as a within-family strengths dimension measure.
The relationships between types of family strengths and personal well-being was not significant, QB = 2.21, df = 3,29, p = .5300, and the relationships between family strengths and family well-being was also not significant, QB = 1.37, df = 3,34, p = .7120.

Positive psychology and family strengths
Findings from studies of the relationships between individual strengths and different dimensions of well-being show that the presence of more strengths is associated with enhanced positive and attenuated negative well-being (Linley, 2013;Macaskill & Denovan, 2014;Park, Peterson, & Seligman, 2004). Results from the present study also showed that family strengths were related to both personal and family well-being. The findings provide support for contentions by Lopez (2009) and Sheridan et al. (2004) that the intersection between positive psychology and healthy family functioning is family strengths.

Differential relationships between family strengths and well-being
Results from the meta-analysis add to our understanding of the relationships between family strengths and healthy functioning by showing how strengths are differentially related to different dimensions of well-being. Results showed that the size of effects for family strengths was larger for family well-being compared to personal well-being. This finding is consistent with the hypotheses that the presence of family strengths would be related to better family well-being (e.g., DeFrain & Stinnett, 2002b;Greeff & le Roux, 1999). The results add to the knowledge base in terms of how family strengths are related to personal and family well-being (Moore et al., 2002). The between type of family strengths comparisons indicated that sizes of effect for four of the strengths dimensions (commitment, communication, coping, & cohesion) were larger than the sizes of effects between family competence and both personal and family well-being. This is most likely due to differences in the attributional targets (Bugental et al., 1998) of the FFSS scale items. The targets of appraisals of family commitment, communication, coping, and cohesion all involve the family or individual family member beliefs, behavior, and practices involving one another (e.g., "We make personal sacrifices if it helps our family", "we are always willing to pitch in and help one another"). In contrast, the targets of appraisal of the family competence items are the family's relationship to persons or organizations outside the family (e.g., "Friends and relatives are always willing to help with our problems"). Family competence may, therefore, be a help-seeking or social support construct rather than a family strengths construct.
As described in the Results section, there were larger degrees of heterogeneity in the sizes of effects for the family well-being results compared to the personal well-being results. It was noted that this may have been due to a few studies with larger sample sizes. There is likely one other unmeasured factor that could account for the differences in the homogeneity of the sizes of effects for personal well-being and the heterogeneity of the sizes of effects for family well-being. The personal well-being measures each assessed some dimension of subjective well-being or psychological health (e.g., stress, anxiety, depression, well-being). In contrast, the family well-being measures assessed more varied constructs (e.g., satisfaction, quality of life, well-being). As part of coding the outcome measures in the larger meta-analysis (Dunst et al., in press), it was noted that many of the measures of the different types of well-being seemed, in some cases, more different than alike. This was the case for family well-being. It was also noted that too few studies in the meta-analysis used the same scales measures of well-being to do comparisons between the type of well-being measures.

Implications for research and practice
The fact that family commitment, family communication, family cohesion, and family coping each had similar sizes of effect with personal and family well-being is not an indication that any particular family strength dimension might be the focus of research or practice at the exclusion of other strength dimensions. The qualities of strong families are best understood as unique combinations of strengths. As noted by Otto (1962), "family strengths are not isolated variables, but form clusters and constellations which are dynamic, fluid, and interacting" (p. 80). Lewis, Beavers, Gossett, and Phillips (1976) as well noted that "optimally functioning or competent families appears to be due to the presence and interrelationship of a number of factors" (p. 205). Dunst et al. (1988), as part of a content analysis of different compilations of the qualities of strong families, concluded that "A combination of qualities appear to define strong families, with certain combinations defining unique family functioning styles" (p 25). Therefore, different families are likely to be characterized by particular family strengths dimensions where different life circumstances are likely to activate different strength qualities or dimensions. It would therefore be of interest to know if the presence of different combinations of family strengths are not only differentially related to different types of well-being but if and how different dimensions of family strengths are related to areas of personal and family functioning (e.g., parent-child interactions, belief appraisals).
The meta-analysis of the relationships between different dimensions of family strengths and personal and family well-being is, to the best of the author's knowledge, the first research synthesis of studies using a family strength scale to assess the qualities of strong families (Deal et al., , 2009 and to ascertain the sizes of effects with two types of well-being. As described earlier in the paper, only one study was located (Arshat & Baharudin, 2014) that used another family strengths scale where the items were based on the qualities of strong families (Stinnett & DeFrain, 1985) where the total scale score was only related to child well-being.
Our understanding of how the qualities of strong families are related to family wellbeing and individual family member well-being (DeFrain & Stinnett, 2002b) could be advanced by studies using family strengths scales other than the FFSS to determine the relationship between family strengths and well-being. This includes meta-analyses of studies using family strengths scales development from different conceptual and theoretical perspectives (Olson, 1989). Two such scales are the Family Strengths Scale (FSS; Olson, Larsen, & McCubbin, 1983) and the Family Hardiness Index (FHI; M. A. McCubbin, McCubbin, & Thompson, 1986). The FSS measures two types of family strengths (pride and accord) and the FHI measures three types of family strengths (commitment, challenge, and control). Meta-analyses of studies using these scales would permit assessment of how other family strengths dimensions are related to well-being. Comparisons of the results from these meta-analyses with the results reported in this paper would provide a basis for knowing whether family strengths operationalized from different perspectives yield similar or dissimilar findings.