Three models of hierarchical multiple regression analyses with subscales of physical health (PHS) as the dependent variables (N = 282).
Bowen family system theory describes family interactional processes that are carried across the generations and determine an individual’s level of autonomy and emotional reactivity as well as the global functioning of the family. According to the theory, any personal, health-related, or relational issues can be explained as a result of diffused anxiety produced by destructive interactional patterns among family members. Although many studies are revealing the relationship between early family life experiences and functioning in adulthood, there is still a lack of studies exploring the complex mediational models based on Bowen theory that would reveal associations between different family-of-origin variables and adults’ health as well as psychological well-being. The chapter defines the main assumptions of Bowen theory as well as summarizes the main results of three studies demonstrating how family and personal factors defined by Bowen theory, such as family emotional system, triangulation, differentiation of self, relate to adults’ health and psychological well-being.
- Bowen family system theory
- family emotional system
- differentiation of self
- interparental relationship
The quality of life of an adult is determined by the interaction of many factors, yet psychological well-being and physical health can be considered as the essential variables ensuring the quality of life. Therefore, researchers seek to understand what factors and how they contribute to a person’s psychological well-being and physical health. Numerous studies reveal the importance of individual psychological and physical factors for these variables, but the significance of family interactional dynamics has become increasingly recognized over the last few decades. There is strong evidence that more favorable interactions with a spouse/partner usually predict better health as well as psychological well-being outcomes of an adult [1, 2, 3]. Additionally, the newest data suggest that broader family interactional context accounts for the significant changes in a person’s physical health and morbidity as well as in psychosocial functioning (e.g., [4, 5, 6, 7, 8, 9]). Thus, the body of research provides clear evidence that supports the necessity to understand physical health and psychological well-being from a systemic perspective. However, understanding the associations between variables without a clear theory is difficult to achieve. Bowen family systems theory (BFST) is a systemic theory binding together family processes and individual factors. It has been vastly studied and has gained recognition for explaining different intrapsychic as well as interpersonal phenomena in the interactional context of the nuclear and genetic family [7, 10, 11, 12, 13, 14, 15, 16]. Bowen  claimed that an individual’s physical and emotional issues could be explained by the interactional processes in the family system. The chapter defines the main assumptions of the BFST as well as summarizes the main results of three studies demonstrating how family and personal factors defined by the BFST, such as family emotional system, triangulation, differentiation of self, relate to adults’ health and psychological well-being.
2. The main ideas of the Bowen family systems theory
The BFST describes different interactional processes that are carried across the generations and determine an individual’s level of autonomy and emotional reactivity as well as the global functioning of the family [17, 18]. According to the theory, any personal or relational issue/symptom can be explained as a result of diffused anxiety produced by destructive interactional patterns among family members [15, 17]. Bowen described six interrelated concepts that address family processes, with differentiation of self (DoS), regarded as the central one as it appears in the manifestation of every other phenomenon described by Bowen . It explains two fundamental aspects of psychosocial functioning: the ability to separate own feelings from thinking and to remain capable of making decisions under stress as well as the ability to develop close intimate relationships while remaining autonomous [12, 14, 16, 17, 20]. The DoS develops within the significant relationships with parents whose DoS shapes their mutual relationship as well as the relationship with their children and is transferred to the next generation through the nuclear family emotional system (NFES), triangulation, and family projection .
NFES can be referred to as a cumulative phenomenon since it binds the rest of the five concepts together (DoS, triangulation, multigenerational transmission process, family projection process, emotional cutoff) and explains the pathways of symptom development. The evolution of NFES begins even before an adult decides to have an intimate relationship. NFES develops gradually and depends on the quality of relationships with families-of-origin, the adjustment of partners to each other before having children as a two-person system, and their adjustment as a three-person system when a child is born. The partners’ level of DoS plays a critical role in the development of NFES since the lower DoS predicts more fusion and anxiety between the partners. Bowen  postulated that a person usually subconsciously chooses a partner with a similar level of DoS. The lower DoS predicts more fusion and anxiety between the partners. Partners with low differentiation are emotionally reactive, have little personal autonomy, seek emotional support, appraisal, and needs’ gratification. Under stress, they cannot think clearly and make effective decisions, tend to become overreactive and overwhelmed by the emotions. Marital discord usually develops ‘when neither spouse will “give in” to the other in the fusion, or when the one who has been giving in or adapting refuses to continue’ [17, p. 115]. The conflict can manifest with a wide range of interactional patterns from simple quarrels to overt violent acts as well as alienating from each other. Despite the tendency to establish a partnership with a person who has a similar level of differentiation, one partner usually has a slightly higher level of differentiation, and, in turn, he or she absorbs less undifferentiation and anxiety by communicating over-protectiveness, worry, or criticism to the less functional partner. The latter soaks up more of undifferentiation and anxiety, accepting her helplessness, powerlessness. Such interactions allow avoiding a conflict; however, they strengthen the unequal positions of the partners by allowing the more functional partner to gain more strength while the less functional partner develops emotional, social, or physical symptoms. Such interactional dynamics result in high levels of anxiety in the couple that has to be channeled somewhere else. Within the triadic systems, parents with low DoS tend to transfer their diffused anxiety to children through triangulation processes. Kerr and Bowen  noted that the two-person system is unstable because of the constant fluctuation of anxiety levels between partners and as it increases, the two-person system forms a triangle by involving a third person in their relationship. After pulling a child into their conflictual relationship, parents usually seek the child’s compassion, emotional support, or openly set a child against a partner. In this way, a child is forced to support the side of one of the parents, mediate in their conflict, comfort, etc., which may lead to a loyalty conflict in a child. Therefore, continual triangulation may result in children’s mental or physical health problems and has detrimental effects on a child’s development as well as functioning in adulthood [21, 22]. Involvement in triangulation does not enable children to become more differentiated than parents, restricts their autonomy, and strengthens emotional reactivity and anxiety [15, 18, 23, 24, 25]. Summarizing, NFES reflects the levels of differentiation and anxiety in the family system through three presumable patterns of the symptoms: chronic marital problems, physical or psychological illness of a partner, impairment of a child .
This chapter aims to present and discuss the results of three studies based on the Bowen family system theory. Thus, the next sections will present the summary of the studies aimed at examining the role of various family-of-origin variables on adults’ physical health and psychological well-being. Since some study measures and data analysis methods are shared among all three studies, the sections Measures and Data analysis, as well as General discussion, are integrated, while the sections Participants and Results are separate for each study.
Interparental relationship quality was measured by Retrospective Measure on Interparental Relationship Quality  in Study 2 and 3. This is a 21-item self-report questionnaire that retrospectively measures interparental relationship quality of adult respondents who grew up with both of their parents. The IPRQ has a structure of three factors: (1) destructive relationship (conflicts, violence, expression of anger, unhappiness and fear, non-constructive ways of conflict resolution, the involvement of children into interparental conflicts); (2) harmonious relationship (constructive ways of conflict resolution, expression of positive emotions, happiness, mutual respect and regard, accord over parenting and domestic duties); (3) avoidant relationship (avoidant and indifferent interactional patterns). The items are evaluated on a 5-point scale from 1 (totally disagree) to 5 (totally agree). The items defining destructive interactional patterns have reversed scoring, and higher scale scores reflect fewer signs of destructive communication. For the present sample, Cronbach’s α were: (1) .91, (2) .93, (3) .53, and for total score – .93.
Differentiation of Self was assessed by Differentiation of Self Inventory (DSI)  in Study 2 and 3. The DSI consists of 43 items and contains four subscales: (1) Emotional Reactivity (ER) (the extent to which one responds to anxiety-provoking situations with intensified emotions); (2) I-Position (IP) (the ability to adhere to one’s beliefs despite an external pressure); (3) Emotional Cutoff (EC) (feelings of excessive vulnerability in relations with others); (4) Fusion with Others (FO) (emotional over-involvement as well as over-identification with parents). All items are scored on a 6-point scale from 1 (not at all true of me) to 6 (very true of me). The higher scores of the subscales reflect a lower level of DoS. Cronbach’s α for DSI (total score), ER, IP, EC, and FO in the present study were .87, .82, .75, .80 and .61, respectively.
Satisfaction with couple relationship was evaluated using Couple Relationship Satisfaction Scale (CRSS)  in Study 2 and 3. The scale can be used with couples irrespective of their developmental stage (dating, cohabitation, marriage). The CRSS has 13 items and contains two subscales: (1) Satisfaction with an emotion-focused relationship; (2) Satisfaction with a behavior-focused relationship. All items are scored on a 10-point scale from 1 (totally dissatisfied) to 10 (totally satisfied) with an additional answer point of 0 (not applicable) for those specific situations when couples live separately or do not have children or do not have sexual intimacy and the like. The higher scores of the subscales reflect higher satisfaction with the relationship. For the present sample, Cronbach’s α were: (1) .95, (2) .93, and for a total score – .96.
Psychological well-being was measured by Ryff’s Psychological Well-Being Scales (PWBS)  in Study 2. The 54-item version was used. The PWBS consists of 6 scales (each having 9 items): (1) Autonomy (A); (2) Environmental mastery (EM); (3) Personal growth (PG); (4) Positive relations (PR); (5) Purpose in life (PL); (6) Self-acceptance (SA). All items are scored on a 6-point scale from 1 (strongly disagree) to 6 (strongly agree). Negatively phrased items are recoded and the higher scores of the subscales reflect a higher level of well-being. Cronbach’s α for PWBS (total score), A, EM, PG, PR, PL, and SA in the present study were .94, .73, .83, .72, .81, .77 and .84, respectively.
Nuclear family emotional system (NFES) in Study 1 was measured with Nuclear Family Functioning Scale (NFFS) , adjusted to internet survey . The 25-item scale consists of four subscales: (1) Personal distress (Cronbach α = .83). (2) Destructive relationship with a partner (Cronbach α = .88). (3) Constructive relationship with a partner (Cronbach α = .89). The subscale has a reverse scoring, thus before statistical analysis, the items’ scoring must be recoded, and a higher subscale score represents a less constructive relationship with a partner. (4) Child’s problems (Cronbach α = .81). The total score of the NFES is calculated by adding the response values of all 25 items (Cronbach α = .92). Higher subscales’ scores represent more severe dysfunction in the evaluated domain of the nuclear family.
Family-of-origin emotional system (FOES) in Study 3 was measured by applying the short version of the Family-of-Origin Scale (FOS)  which retrospectively measures manifestation of autonomy and intimacy within adults’ family-of-origin. The scale consists of 22 items (e.g., “I remember my family as being warm and supportive”) and has a one-factor structure. The items are evaluated on a 5-point scale from 1 (strongly agree) to 5 (strongly disagree). Ten items are recoded before adding the total score and the higher scores of the scale reflect lower levels of perceived intimacy and autonomy within the family-of-origin. Cronbach’s α for the present sample was .96.
Adults’ physical health in Study 1 was assessed with Physical Health Scale (PHS) . The 9-item scale had two subscales: (1) subjective health indicators (Cronbach α = .91); the scale covers the subjective evaluation of one’s health (e.g., ‘I feel perfectly healthy’). The items were scored on a 5-point scale from 1 (‘totally agree’) to 5 (‘totally disagree’). (2) objective health indicators (Cronbach α = .72), such as frequency of visiting physicians during the last year (never/rarely/often/regularly); using medications due to health problems (no/yes); reduction of working capacity due to an illness or other health problems (no/yes). Higher scores of subscales represent a worse evaluation of physical health.
Triangulation. In Study 3 the triangular relationship inventory (TRI)  was used to measure triangulation in the family of origin. The inventory is a self-report 24-item instrument having four subscales: (1) Balanced (e.g., “My parents handle the tension between one another without including me”); (2) Mediator (e.g., “Both of my parents use me to communicate with the other”); (3) Cross-Generational Coalition (e.g., “I have to take sides when my parents disagree”); (4) Scapegoat (e.g., “My parents seem to work together only when they are dealing with my behavior”). Although the TRI is intended for late adolescents’/early adults’ in this study it was used for adults asking to evaluate their relationships with parental figures retrospectively. For this purpose, the tense of the items was changed from present to past. The items are evaluated on a 5-point scale from 1 (strongly disagree) to 5 (strongly agree). The items of the Balanced subscale are recoded and the higher scores of all subscales correspond to more triangulation. For the present sample, Cronbach’s α were: (1) .84, (2) .83, (3) .86, (4) .87 and for the total scale score – .90.
Physical and emotional health. In Study 3 the RAND 36-Item Health Survey (Version 1.0)  covers different physical and emotional health aspects. Scoring of the instrument is a two-step process: (1) precoded numeric values are recoded per the scoring key provided by the authors. Each item is scored on a 0 to 100 range so that the lowest and highest possible scores are set at 0 and 100; (2) items on the same scale are averaged together. All items are scored so that a high score defines a more favorable health state. The physical health scale was compounded of the 21 items reflecting physical health aspects (physical functioning, bodily pain, role limitations due to physical health problems, general health perceptions). The emotional health scale was made up of 14 items corresponding to emotional health aspects (role limitations due to personal or emotional problems, emotional well-being, social functioning, energy/fatigue). Cronbach’s α were .88 and .92, respectively.
Sociodemographic information. In all three studies, respondents were asked to indicate their gender, age, couple status, duration of the relationship, and if they have children. Additionally, in Study 2 and 3 participants were asked to refer their education level and if they grew up with both biological parents, while in Study 1 – the age of every child and health issues in the presence.
3.2 Data analysis
In Study 1 and 2 hierarchical series of multiple regression analyses were conducted to explore the contribution of study variables to the prediction of psychological well-being (six subscales of PWBS) as the dependent variable. To estimate the effect size for a hierarchical multiple regression in each step Cohen’s f2 was calculated. According to Cohen’s  guidelines, f2 ≥ 0.02, f2 ≥ 0.15, and f2 ≥ 0.35 represent small, medium, and large effect sizes, respectively.
In Study 3 a mediation analysis was performed to test the impact of interparental relationship quality, triangulation, family emotional system, the DoS, and satisfaction with a couple relationship on adult’s physical and emotional health. A path analysis was carried out using the Maximum Likelihood method and the following goodness-of-fit indices : model Chi-Square, the Root Mean Square Error of Approximation (RMSEA), the Standardized Root-Mean-Square Residual (SRMR), Tucker-Lewis Index (TLI), Goodness of Fit Index (GFI) and Comparative Fit Index (CFI). TLI, GFI, and CFI values greater than 0.90 are considered acceptable, whereas the ones higher than 0.95 are considered excellent. RMSEA and SRMR values lower than 0.08 are considered acceptable, whereas values close to 0.05 are considered as good . Bootstrap re-sampling was applied to test the significance of the mediation paths, using 5000 bootstrap samples and 95% confidence intervals.
The data were analyzed using SPSS Version 23 and AMOS Version 23.
3.3 Study 1
The objective of the study was to test the prognostic value of adults’ current family dynamics reflected by NFES on adults’ physical health.
3.3.1 Participants and procedure
The study was conducted online in 2017 in Lithuania. Respondents were recruited from various social networking websites intended for family issues. Before starting their participation, respondents were presented with an informed consent covering the primary goal of the study, voluntary participation, confidentiality, data protection, and contacts of the researcher.
Inclusion criteria were the following: intimate relationships at the moment of participation in the study and having a child from 3 to 18 years. The final sample consisted of 282 participants of whom 95% (n = 267) were women. The average age of the participants was 36.27 (SD = 5.78) years, ranging from 21 to 53 years. Most participants (n = 247, 88%) were married, the rest 35 (12%) cohabitated with one’s intimate partner. The average duration of the relationship was 13.08 years (SD = 6.03). Twelve percent (n = 34) of the participants indicated within 5 years being diagnosed with physical illnesses, such as arthritis, diabetes, epilepsy, hypertension, migraine, cancer, etc.
Two three-step hierarchical multiple regressions were performed for both subscales of PHS (Objective health indicators, Subjective health indicators) as the dependent variables. Each hierarchical multiple regression analysis consisted of three blocks of independent variables which were subscales of NFES (Personal distress, Constructive relationship with a partner, Destructive relationship with a partner, Child’s problems), presence of illness at the moment of the survey, and control variables (gender and age). The variance inflation factors (VIF) and tolerance factors for each of the single predictor variable were no larger than 4 (ranging between 1.1 and 2.6) and no smaller than 0.25 (ranging between 0.39 and 0.95), respectively, suggesting no collinearity between independent variables. The results of the final regression models are presented in Table 1.
|Blocs of predictors||Objective health indicators||Subjective health indicators|
|Model 1 (β)||Model 2 (β)||Model 3 (β)||Model 1 (β)||Model 2 (β)||Model 3 (β)|
|1st block: NFES|
|2nd block: HI|
|Presence of illness||.57***||.57***||.32***||.32***|
|3rd block: CV|
|ES (Cohen’s f2)||0.06||0.46||—||0.21||0.11||—|
Analysis unraveled quite apparent differences between the results of regression analysis when predicting objective and subjective health indicators in terms of significant independent variables and effect size. Although in the first model personal distress, the only scale of NFES, significantly predicted objective health, its effect size was small and after adding the presence of illness into the regression the personal distress lost its predictive value. The final model explained 36% of the variance of adults’ objective health and the only significant predictor in the context of other independent variables was the presence of illness: those adults who have diagnosed illness reports more objective indicators of health’s problems. Differently, in the regression analysis of subjective health three scales of NFES sustained their significant predictive values after adding the presence of illness and the effect size of the latter was small when the effect size of NFES was medium. The final model explained 25% of the variance of the dependent variable and revealed that higher personal distress, more signs of a destructive relationship with a partner, more severe child’s problems, and presence of illness predict more indicators of subjectively evaluated physical health’s problems. The similarity between both models was associated to control variables (gender and age) – neither of them had predictive power and the effect size was equal to zero in both cases.
The hierarchical multiple regression analysis was repeated with the total score of NFFS instead of its scales aiming to examine the predictive value of the NFES as a whole (see Table 2).
|Blocs of predictors||Objective health indicators||Subjective health indicators|
|Model 1 (β)||Model 2 (β)||Model 3 (β)||Model 1 (β)||Model 2 (β)||Model 3 (β)|
|1st block: NFES|
|2nd block: HI|
|Presence of illness||.59***||.59***||.34***||.35***|
|3rd block: CV|
|ES (Cohen’s f2)||0.02||0.52||—||0.14||0.13||—|
The results presented in Table 2 are very similar to those in Table 1, nevertheless, one difference was found. When predicting objective health, the NFES (total score) sustained significant predictive value after adding the presence of illness into the equation. Thus, the worse objective health of adults is predicted by the worse NFES and presence of illness, although the effect size of the first was small and of the second one was large.
3.4 Study 2
The objective of the study was to examine the prognostic values of interparental relationship quality observed during childhood, DoS, and satisfaction with a current couple relationship on adults’ psychological well-being.
3.4.1 Participants and procedure
Lithuanian online survey software program (
There were three main inclusion criteria for participation in the study: being 18–55 years old, growing with both parents during childhood, and being in a romantic relationship at the moment of participation in the study for at least one year. The final sample consisted of 905 respondents who satisfied all inclusion criteria and who consented to their data being used in the study. Ninety two percent (n = 833) of respondents were women, the mean age was 38.49 (SD = 9.46) years. Most of the study participants (n = 653, 72%) were married, 136 (15%) cohabitated, and 116 (13%) were dating with a romantic partner. The average duration of the relationship was 13.87 years (SD = 8.61). Education of the respondents was as follows: 587 (65%) respondents completed university studies; 134 (15%) studied in the college; 58 (6%) completed professional training; 115 (13%) finished secondary education, and education of 11 (1%) respondents was lower than secondary education.
The research objective was implemented by running six four-step hierarchical multiple regressions for each subscale of PWBS (Autonomy, Environmental mastery, Personal growth, Positive relations, Purpose in life, Self-acceptance) and PWBS total score as the dependent variable. Each hierarchical multiple regression analysis consisted of four blocks of independent variables which were subscales of interparental relationship quality (Destructive relationship, Harmonious relationship, Avoidant relationship), DoS (Emotional reactivity, I-position, Emotional cutoff, Fusion with others), satisfaction with couple relationship quality (Satisfaction with an emotion-focused relationship, Satisfaction with a behavior-focused relationship) as well as control variables (gender, age, and education). The variance inflation factors (VIF) and tolerance factors for each of the single predictor variable were no larger than 4 (ranging between 1.0 and 3.5) and no smaller than 0.25 (ranging between 0.29 and 0.96), respectively suggesting no collinearity between independent variables. The results of the final regression models are presented in Table 3.
|Blocs of predictors||Autonomy||Environ-mental mastery||Personal growth||Positive relations||Purpose in life||Self-acceptance||Well-being (total score)|
|1st block: IPRQ|
|2nd block: DoS|
|3rd block: CRS|
|4th block: CV|
The results revealed that 44.8% of the variance of the participants’ autonomy can be attributed to the main three blocks of predictors tested in the analysis. Only 0.4% of the variance can be explained by control variables and the change in R2 adding the control variables was nonsignificant. Although all four models were statistically significant (FModel1 = 7.58, p < .001; FModel2 = 106.43, p < .001; FModel3 = 82.63, p < .001), the greatest change in R2 was produced by the block of the DoS – this block accounted for 42.9% of the variance of the participants’ autonomy with a large effect size (f2 = .75), while effect sizes of the rest of the independent variables’ blocks were small. The final model (see Table 3) explained 45% of the variance and showed that a better autonomy of adults is predicted by their stronger I-position, lower emotional cutoff, lower fusion with others, and younger age with I-position as the strongest predictor in the final model.
The results of hierarchical multiple regression analysis predicting environmental mastery revealed that all four models were statistically significant (FModel1 = 21.27, p < .001; FModel2 = 99.81, p < .001; FModel3 = 86.79, p < .001), though the variance explained varied depending on the predictors’ blocks. The main three blocks of predictors explained 46.1% of the variance of the environmental mastery variable, whereas control variables explained only 0.7% of the variance. The most significant change in R2 was provided by the block of the DoS explaining 37% of the variance of the dependent variable (a large effect size; f2 = 0.59). The final model (see Table 3) accounted for 46.8% of the variance and showed that a more harmonious interparental relationship perceived in childhood, lower emotional reactivity, stronger I-position, lower emotional cutoff, higher satisfaction with a behavior-based couple relationship, and older age predict higher environmental mastery of adults. The strongest predictor in the final model was I-position.
All four models predicting personal growth were statistically significant (FModel1 = 10.11, p < .001; FModel2 = 41.51, p < .001; FModel3 = 32.34, p < .001) with the main three blocks of predictors accounting for 23.8% of the variance of dependent variable, whereas control variables explained only 2.4% of the variance. Similarly, the most considerable change in R2 was made by the block of the DoS. The block explained 21% of the variance of the functioning variable and demonstrated a medium effect size (f2 = 0.27). The final model (see Table 3) accounted for 26.2% of the variance and showed that stronger I-position, lower emotional cutoff, younger age, and a higher level of education predict a higher level of an adult’s personal growth.
The results of the prediction of positive relations were similar to those predicting previous dependent variables as all four models were significant (FModel1 = 34.26, p < .001; FModel2 = 95.82, p < .001; FModel3 = 76.73, p < .001) and the DoS block had the largest predicting power (explained 32.4% of the variance and demonstrated a large effect size, f2 = 0.49). The final model (see Table 3) explained 43.5% of the variance and indicated that more positive relations of an adult are predicted by fewer signs of interparental avoidances perceived in childhood, lower emotional reactivity, stronger I-position, lower emotional cutoff, higher fusion with others, higher satisfaction with an emotion-based couple relationship, and female gender. The strongest predictor of positive relations in the final model was cutoff.
The four models predicting purpose in life were statistically significant (FModel1 = 19.91, p < .001; FModel2 = 52.66, p < .001; FModel3 = 41.69, p < .001) and the largest parts of the variance of dependent variable was accounted by DoS block (explained 22.7% of the variance and demonstrated a medium effect size, f2 = 0.30). The final model (see Table 1) explained 30% of the variance and revealed that purpose in life is predicted by more harmonious interparental interactions with fewer signs of avoidances perceived in childhood, lower emotional reactivity, stronger I-position, lower emotional cutoff, younger age, and a higher level of education with the cutoff as the strongest predictor.
The results predicting self-acceptance did not distinguish from previous ones. The four models predicting the dependent variable were statistically significant (FModel1 = 2254, p < .001; FModel2 = 93.03, p < .001; FModel3 = 86.96, p < .001) and the largest parts of the variance of dependent variable was accounted by DoS block (explained 34.9% of the variance and presented a large effect size, f2 = 0.54). The final model (see Table 3) explained 46.4% of the self-acceptance’s variance. It revealed that the strongest predictor is I-position, which together with more harmonious interparental relationship perceived in childhood, lower emotional reactivity, lower emotional cutoff, higher satisfaction with an emotion-based couple relationship, and a higher level of education predict higher self-acceptance of an adult.
Finally, the results of regression analysis with PWBS total score as the dependent variable were in line with those predicting separate scales of PWBS. The block of DoS scales had the strongest effect (explained 31.5% of the variance and presented a large effect size, f2 = 0.88) and the final model explained the substantial part (57.4%) of the dependent variable’s variance. The results revealed that higher psychological well-being can be expected among those who in the family-of-origin observed more signs of harmonious and less of avoidant relationships between parent figures, have stronger I-position and stronger emotional ties with significant persons, are less emotionally reactive, are more satisfied with emotional interactions with a partner and have a higher level of education.
3.5 Study 3
The objective of the study was to examine a complex mediation model that includes the triangulation and family-of-origin emotional system (FOES) perceived during childhood, DoS, and satisfaction of current couple relationship as the mediators between retrospectively assessed interparental communication and adult children’s physical and emotional health.
3.5.1 Participants and procedure
The online questionnaire was generated using SoSci Survey  and was made available to users via
There were three main inclusion conditions for participation in the study: being 18 years or older and being in a romantic relationship lasting for at least one year at the moment of participation in the study. The sample consisted of 257 respondents most of whom were women (n = 225; 88%), the mean age was 38.82 (SD = 9.29) years. Most of the study participants (n = 192, 75%) were married, 42 (16%) cohabitated, and 23 (9%) were dating a romantic partner. The average duration of the relationship was 13.52 years (SD = 9.07), 69% (n = 1716) of the respondents had children.
Table 4 presents basic descriptive statistics and correlations among the study variables (total scores) that were included in the mediation path model analysis.
|1. Interparental relationship quality||—||72.86||17.91|
|3. Family-of-origin emotional system||−.66***||.44***||—||57.86||18.73|
|4. Differentiation of self||−.24**||.27***||.26***||—||139.76||27.51|
|5. Couple relationship satisfaction||.31***||−.15*||−.33***||−.31***||—||95.58||26.63|
|6. Physical health||.15*||−.26***||−.11||−.30***||.08||—||81.84||13.91|
|7. Emotional health||.32***||−.35***||−.30***||−.58***||.37***||.46***||67.05||20.46|
Correlation analysis demonstrated statistically significant correlations (p < .001) among all variables except correlations between FOES as well as couple relationship satisfaction and physical health. The strong negative correlations were found between interparental relationship quality and triangulation as well as FOES, which implies that better interparental relationship quality relates to lower triangulation and better nuclear family emotional system. Adults’ better emotional health was moderately correlated with better interparental relationship quality, lower triangulation, better FOES, better couple relationship satisfaction as well as better physical health, and strongly correlated with higher DoS.
Next, the multiple mediation model was established to test the direct and indirect effects of independent variables on adults’ physical and emotional health. Figure 1 demonstrates the theoretical path mediation model based on the BFST. According to the theory it was expected that family-of-origin dynamics (interparental relationship quality, triangulation, and FOES) will have an indirect effect on adult children’s health through DoS and satisfaction of couple relationship. On the other hand, it was anticipated that FOES and triangulation will play as the mediators in the association between interparental relationship quality and adult children’s DoS.
The statistical analysis consisted of two steps: testing for model fit and for mediation with bootstrapping. According to the theory, the initial model was set to be recursive and to include paths from every independent variable to the supposed mediator and the dependent variables. However, after the correlation analysis revealed no significant correlations between FOES as well as couple relationship satisfaction and physical health, the paths were removed from couple relationship satisfaction and FOES to adults’ physical health. Figure 2 presents the results of the model for adults’ health demonstrating significant direct effects.
The mediation model provided an excellent fit to the data: χ2(5) = 4.23, p = .52, CFI = 1.00, TLI = 1.01, GFI = 1.00, SRMR = .017, RMSEA = .00. The model explained 12% of the variance of the variable adults’ physical health and 41% of the variance of adults’ emotional health. The model confirmed all predicted direct effects and additionally demonstrated the direct effect of retrospectively evaluated triangulation on adults’ physical and emotional health as well as the direct effect of retrospectively evaluated family emotional system on current satisfaction with couple relationship.
To explore multiple mediations in detail, the estimates of specific indirect effects and their confidence intervals were calculated separately (see Table 5). Based on the bootstrapping confidence intervals, significant indirect effects (full mediation) were shown from interparental relationship quality through triangulation and family emotional system on the DoS as well as through the FOES on couple relationship satisfaction. Although interparental relationship quality significantly predicted triangulation and FOES and explained a considerable part of their variance (46% and 37%, respectively), the independent variable as well as both mediators explained only 9% of the DoS variance.
|Mediation hypothesis||Direct effect||Indirect effect||95% CI||Results|
|IRPQ→T → DoS||.03 (ns)||−.20*||[−.36; −.05]||Full mediation|
|IRPQ→ FOES→DoS||.03 (ns)||−.22**||[−.37; −.08]||Full mediation|
|IRPQ→ FOES→CRS||.10 (ns)||.21*||[.07; .36]||Full mediation|
|FOES→DoS → CRS||−.21*||−.07*||[−.13; −.02]||Partial mediation|
|IRPQ→DoS → EH||.02 (ns)||−.02 (ns)||[−.13; .08]||No mediation|
|IRPQ→CRS → EH||.02 (ns)||.02 (ns)||[−.01; .08]||No mediation|
|T → DoS → EH||−.16*||−.12*||[−.21; −.03]||Partial mediation|
|DoS → CRS → EH||−.48***||−.03**||[−.06; −.01]||Partial mediation|
|IRPQ→T → DoS → EH||.02 (ns)||.07*||[.02; .13]||Full mediation|
|IRPQ→ FOES→DoS → EH||.02 (ns)||.08**||[.03; .14]||Full mediation|
|IRPQ→ FOES→DoS → CRS → EH||.02 (ns)||.01*||[.00; .02]||Full mediation|
|T → DoS → PH||−.22*||−.04*||[−.08; −.01]||Partial mediation|
|IRPQ→DoS → PH||−.07 (ns)||−.01 (ns)||[−.05; .03]||No mediation|
|IRPQ→T → DoS → PH||−.07 (ns)||.03*||[.01; .05]||Full mediation|
|IRPQ→ FOES→DoS → PH||−.07 (ns)||.03**||[.01; .05]||Full mediation|
Full mediation was found assessing multiple mediation models demonstrating that interparental communication impacts adult children’s physical and emotional health indirectly through the mediators: triangulation, FOES, DoS, and satisfaction with current couple relationships in the case of emotional health. Interestingly, retrospectively evaluated triangulation in childhood had both significant indirect (through the mediator DoS) and direct effect on adults’ physical and emotional health.
3.6 General discussion
The primary aim of this chapter was to present and discuss the results of three studies based on systemic ideas proposed by the BFST [17, 18]. According to the theory , interparental as well as parent–child interactional dynamics profoundly affect family members’ psychological, social, and even physical functioning. DoS is a core phenomenon linking intrapsychic and interpersonal relationships as well as between-generational transmissions of some interactional patterns that are repeated in different generations again and again. Partners alongside some level of DoS unconsciously bring into the couple’s life interactional patterns perceived and experienced in a family-of-origin thus shaping specific nuclear family emotional system. Lower levels of partners’ DoS accounts for the development of destructive interactions and consequently adverse family emotional system. Children in such a family are at risk to be involved in interparental conflicts through the triangulation process and to develop accordingly lower level of DoS. Whereas Bowen suggested that any relational, mental, and physical health problem can be explained by interactions of the mentioned phenomena, three studies presented in this chapter were to explore the significance of retrospectively evaluated FOES as well as triangulation, DoS, NFES and satisfaction with couple relationship for adults’ health and psychological well-being.
The associations between family variables and adults’ physical health. Despite the sound theory, the pathways of how interparental relationship quality, triangulation, and family emotional system affect an adults’ physical health are unclear. One possible path could be through the DoS: a high level of anxiety and problems in the family system decrease DoS and produce higher emotional reactivity, which in turn weakens the immune system of family members leading to their health worsening. However, there is no available research based on the BFST that would test either the role of DoS as the mediator or the family variables themselves, for the adults’ physical health status.
The physical health in Study 1 was evaluated through subjective indicators such as satisfaction with own physical health as well as objective indicators such as using medications due to health problems. The results revealed a quite substantial difference in the role of the nuclear family emotional system on objective and subjective indicators of adults’ physical health. NFES explained only 10 percent of the variance of objective indicators and almost the third (32%) of subjective indicators. The more severe personal distress, the more significant problems of a child, and the more destructive relationship with a partner predicted worse subjective health indicators. In general, the high number of personal and interactional problems in the family had a negative effect on the subjective evaluation of personal health. The results confirm the assumption of BFST  that the family emotional system reflecting the generated level of anxiety and problems in the family system influences the physical functioning of adult family members. However, as mentioned before, although the higher personal distress and the higher total score of NFES were significantly associated with worse objective health indicators, the overall score of NFES explained only 10% of the dependent variable’s variance. The results are ambiguous for two reasons. First, the state of physical health did not relate to couple discord and less constructive couple communication, although many studies demonstrated that dyadic interaction has an impact on personal health outcomes [1, 2, 3, 39, 40]. One possible explanation for such results could be related to the health measures used in the studies. The respondents were asked to report their objective physical health indicators such as physician visit frequency, the use of medications due to health problems, and reduction of working capacity due to a disease or other health problems. The cited research used subjective evaluations of physical health. Our study demonstrated that despite a significant moderate correlation (r = .57, p < .001), objective and subjective physical health indicators do not necessarily constitute a single phenomenon. Besides, the results also demonstrated that a couple relationship has an association with subjective health indicators. Thus, it could be that a family member who is experiencing high anxiety and other negative emotions due to a discord between him/her and a partner and emotional and behavior problems of a child, subjectively rates his/her physical health as deteriorated due to overall ill-feeling. However, it might be that the subjective evaluation of health conditions does not always mean actual physical health problems, and dyadic issues influence the objective state of health indirectly through other psychological variables. Second, the results, after all, correspond to BFST , proposing that the NFES relates to family members’ physical functioning. However, NFES explained only 10 percent of objective health indicators’ variance. According to the theory, one would expect a much higher percentage, because, according to the theory, family members’ state of physical fitness is an outcome of NFES. Such results suggest that physical health, even though it is related to NFES, is possibly not its direct outcome, as the theory implies. Moreover, it might be that an association between NFES and adult’s objective health is far more complicated than it was covered in the present study. Thus, more variables mediate or moderate the association.
Study 3 helped to look more in-depth on the associations between different family interactional variables and adults’ physical as well as emotional health. The main goal was to examine a complex mediation model based on the BFST showing that family-of-origin dynamics (interparental relationship quality, triangulation, and FOES) in the past will have an indirect effect on adult children’s health through DoS and satisfaction of present couple relationship. The results confirmed the Bowen’s proposition, that family members’ health (physical and emotional) in the present can be explained by the significant interactions in family-of-origin in the past as well as in the present. Specifically, interparental communication through the mediators – triangulation, family emotional system, and DoS – had an indirect effect on adults’ physical health. In the case of emotional health, the mediation model was very similar except for additional mediator – satisfaction with couple relationship. More interestingly, the FOES did not have a significant correlation with adults’ physical health and the result was very similar to that obtained in Study 1, demonstrating a very low correlation between NFES and adults’ physical health. On the other hand, triangulation experienced in the family-of-origin not only had an indirect effect through the mediator DoS but also had a direct effect on adults’ health (both physical and emotional). Lastly, the mediation model demonstrated the significance of DoS predicting adults’ health as well as mediating between other family-of-origin variables and adults’ health. Such results corroborate postulates of The BFST that different interactional processes experienced in the family-of-origin have a critical role in adult children’s emotional well-being and physical functioning.
The scientific data on associations between different family-of-origin variables and adults’ psychological and physical functioning are rather extensive, however somewhat inconsistent. For example, Stuart-Parrigon and Kerns  found that more severe partners’ conflict does not predict later child anxiety that can be considered as a component of DoS and an essential criterion of psychological functioning. However, Cowan and Cowan  noted that the intervention aimed at the marital relationship’s improvement has the potential to shape parent–child communication as well as a child’s functioning. The similar results were found by other authors [43, 44, 45] showing that better parent–child relationships are found among the adults whose relationship with a partner is better. Those studies confirm the spill-over hypothesis proposing a positive association between interparental relationship quality and parent–child relationship quality [43, 46] that in part confirms Bowen’s statement regarding the interparental communication as a channel to children’s DoS.
Additionally, research data show that parents’ ability to discuss family and partners’ relationship issues openly and constructively reduce the feelings of triangulation in children that eventually impair their well-being [10, 47, 48] and is positively associated with young adult children’s higher self-esteem and less perceived stress [49, 50]. Several longitudinal studies [51, 52] confirm the data produced by cross-sectional studies providing empirical evidence about the significant correlations between family relationships during adolescence and functioning in adulthood. More specifically to the BFST, Peleg  demonstrated that individuals who have experienced more stressful life events in childhood and adolescence are of lower DoS and more prone to be involved in intergenerational triangulation, which, in turn, leads to inadequate coping with future stressful events. Such results alongside the results obtained in Study 3 tend to the conclusion that good interparental relationship quality creates a positive emotional environment for children and serves as a protective factor for children’s well-being in adulthood. On contrary, poor interparental relationship quality is a detrimental factor contributing to the development of the negative family emotional system as well as to the involvement of children in the interparental discord through the triangulation process and acting as a risk factor for children’s physical and emotional functioning in adulthood. However, as was mentioned before, both studies (Study 1 and 3) demonstrated quite a weak effect of family variables on adults’ physical health, as they explained only 10 to 12 percent of the dependent variable’s variance. This means that even a higher number of independent variables in Study 3 produced very similar results as in Study 1. Such results might question the posture of the BFST regarding the effect of family on its members’ physical health. More studies are needed to re-examine the associations and look for other potential mediators linking family interactions to its members’ physical health status.
The associations between family variables and adults’ psychological well-being. Study 2 aimed to examine the predictive value of family variables such as interparental relationship quality, DoS, and satisfaction with a couple relationship on adults’ psychological well-being. The results revealed that adults’ DoS had the highest significant impact on their psychological well-being in comparison to other study variables such as the interparental relationship observed during childhood and satisfaction with the current couple relationship. These results are in line with other body of research [4, 5, 6, 16, 53], demonstrating that higher DoS predicts fewer psychological symptoms, higher expression of happiness, well-being, and satisfaction with life. The findings corroborate the statements of Bowen [17, 18] that the DoS is a core characteristic that determines a person’s relationship with self and others, his/her ability to function under stress, and develop meaningful and close relationships which shape the experience of general well-being and functioning daily. Additionally, the results confirm the weightiness of the DoS predicting psychological well-being in adulthood, which shows that it is the most critical phenomenon in comparison to other systemic factors explored in this study.
Interestingly, separate elements of DoS had different prediction power for the particular components of well-being. However, two almost equally strongest predictors of every component of adults’ well-being were I-position and emotional cutoff, suggesting that a higher level of well-being can be found among those who have a clearer self-view, are somewhat autonomous, and maintain warm relationships with significant others . In some other studies, the emotional cutoff is found to predict marital problems and dyadic adjustment [20, 53] leading to lower satisfaction with life which reflects worse psychological well-being . The emotional cutoff is a process of emotional alienation from parents, siblings, and other family members with whom a person has unresolved emotional and attachment issues. The decision to reduce or cut off emotional and/or physical contact with significant others is a way to cope with the anxiety that is generated by these relationships. However, emotional cutoff does not lead to problem resolution. Rather, they become dormant and prompt greater emotional reactivity as well as fusion with current partner/spouse. Thus, adults who are capable of maintaining close and satisfactory relationships with the members of the family-of-origin have the potential to expand their emotional support system and to cope better in their current personal, social, and professional life.
Although DoS was the strongest predictor of adults’ well-being, higher interparental relationship quality (specifically harmonious relationship and less avoidant relationship), as well as satisfaction with an emotion-focused couple relationship played a significant role in predicting some of the components and the total score of the psychosocial well-being as well. This suggests that adults who growing up observed supportive, affectionate, committed, and constructive interactions between parent figures and are more satisfied with emotional communication with their partner in adulthood are more prone to a higher sense of well-being. Study 3 demonstrated that interparental relationship quality perceived in childhood indirectly through the family emotional system and DoS relates to satisfaction with couple relationships in adulthood. Thus, it can be assumed, as Bowen suggested , that patterns of interaction between partners through triangulation, family emotional system, and DoS are transferred through generations. In adulthood, an individual with low DoS has difficulties in creating and sustaining satisfactory relationships [16, 20, 53] because one could not develop appropriate skills in the family-of-origin. Thus, adults’ relationship problems, alongside with high vulnerability to stress, eventually can lead to broader psychosocial functioning difficulties and a low sense of well-being .
The findings propose that adults’ DoS has a significant effect on their well-being both directly and indirectly – higher DoS leads to a more satisfying relationship with a partner which, in turn, has a positive impact on adults’ well-being. On the other hand, higher DoS creates a positive context for a person’s functioning in daily life. Results conform to other research body, providing evidence for the role of DoS as well as couple relationship in adults’ psychosocial functioning [4, 5, 6, 16, 54]. Adults, who have reasonable control over their emotions in stressful situations as well as in relationships with others, and who can develop and maintain healthy and harmonious relationships while sustaining autonomous and clearly defined self, may tend to make better personal and relational decisions leading to a better adjustment during life shifts as well as a more satisfactory life in general. In line with this reasoning, the findings support the postulates of the BFST  regarding DoS as a central mechanism by which personal, as well as relational well-being, is generated. The DoS is a product of interactional processes in the family-of-origin and is transmitted over generations.
Limitations. Some limitations of the studies should be considered when interpreting the results. Although the statistical analysis and theory assume some causal relationships among studied variables, a cross-sectional design that as used in all three studies does not establish the actual causality. A longitudinal research approach, along with a more significant number of different variables pertinent to adults’ physical health and well-being, could provide insights on the pathways that connect family interactions with the health of family members as well as its changes in the systemic context. Another limitation of all 3 studies is related to the lack of gender balance as the samples primarily consisted of women; thus, the generalizability of the results in the men population is considerably restricted. Future research should attempt to involve more male participants to find out if established associations are applicable in both gender groups.
Summarizing the results of the three studies, some general conclusions can be drawn:
As Bowen postulated, the family emotional system that develops gradually in the family-of-origin depending on the interactional patterns between partners as well as between parents and children significantly relates to the psychological well-being of adult children. Adults who were growing up in the family system shaped by parents whose communication was marked by intense discord, conflicts, lack of mutual respect, and who were involving children into their conflicts, tend to develop lower DoS and have poorer emotional health as well as feebler psychological well-being.
Although there were found some associations between adults’ physical health and the communication patterns in the family-of-origin and nuclear family, the data do not allow to judge regarding the family’s role unambiguously, as the phenomena under investigation explained a very small part of the physical health’s variance. Nevertheless, the results revealed that experience of triangulation in the family-of-origin might have a long-lasting detrimental impact on personal health.
DoS is a phenomenon that significantly relates to adults’ satisfaction with couple relationships, emotional health, and psychological well-being. However, the notion of the BFST, that DoS is a product of the communication processes progressing in the family-of-origin might be reconsidered, as researched family variables, although significantly correlated with DoS, together explained a very small portion of the variance of adults’ DoS. Such results prompt considering and looking for other developmental variables that together with phenomena described in the BFST shape the DoS.
Acknowledgments to Simona Celiauskaite, who participated in the development of Physical Health Scale and recruitment of the participants in Study 1.
Conflict of interest
The author declares no conflict of interest.