How Can I Connect with Thee: Measuring and Comparing Satisfaction in Multiple Relationship Domains

Procedures

Participants arrived at the laboratory between 8 and 9 A.M., provided informed consent, and then began approximately 8 hours of psychological and physiological testing that included the “Multi-Domain Relationship Satisfaction” (MDRS) questionnaire, demographic questions, standard psychological surveys, interviews, lunch, and a cardiovascular protocol. The procedure was repeated in Year 2, and these data were used to test the temporal stability of scores on the MDRS questionnaire. Participants were paid $126 (U.S.) each year for completing all components of testing.

Measure of demographic variables

Demographic measures of interest were gender, ethnicity, age, education (years of schooling), and annual household income. Annual household income was reported in 12 categories ranging from less than $5,000 to more than $200,000; to achieve a more continuous distribution, the log-transformed median of each category was used in analyses. Missing values for education (2) and annual household income (13) were replaced with means for the corresponding gender by ethnic group combination. The demographic characteristics of the sample are provided in Table 1.

Multi-Domain Relationship Satisfaction (MDRS)

The objective of this scale was to capture, in broad strokes, whether people could rate diverse sources of felt connectedness on identical criteria. The broadly descriptive items in the MDRS followed a format similar to that employed by Hazan and Shaver (1987) in their broadly descriptive items of attachment styles (e.g., variation in degree of secure attachment was based on the item, “1 find it relatively easy to get close to others and am comfortable depending on them and having them depend on me. I don’t worry about being abandoned or about someone getting too close to me”). In our study, instead of varying relationship qualities, we varied the relationship domain and held constant the terms used to describe relationship quality.

The instructions for the MDRS began with the statement, “People vary in how satisfied they are in their relationships with people, pets, and a higher power (e.g., God).” For each relationship domain, participants were then instructed to use a 7-point Likert response scale (1 = not at all, 7 = very much) to answer: (a) to what extent you feel you have a satisfying relationship with the single most important [person in your life/group, association, or institution in your life / pet / God], and (b) the extent to which this relationship effectively serves as a refuge; a source of safety, security, and consolation; and a shelter or protection from danger or trouble. Only those who endorsed a belief in God were asked to complete the items corresponding to the God relationship domain. All but seven of 224 respondents to this item claimed a belief in God. Similarly, only those who owned a pet were asked to complete the items corresponding to the pet relationship domain. Pet owners represented 41 % (N = 89) of 218 respondents to this item. Of the 74 participants who reported the type of pet they owned, 46 individuals (62% of pet owners) owned dogs, 30 owned cats (41% of pet owners), and 8 of these individuals owned both dog(s) and cat(s). Fish and bird owners also were represented (7 individuals).

Validation measures

For each of the validation measures described below, percentages, ranges, means, standard deviations, and Ns are provided in Table 2.

Objective social circumstances

Participants were asked about their marital status (1 = married or living with a partner; 0 = all others), the number of friends and relatives with whom they speak at least every two weeks, whether they have a religious affiliation (1 = yes, 0 = no), how frequently they attend religious services (more than once a week, once a week, 1–2 times a month, several times a year, never), and number of group memberships (e.g., civic group, neighborhood organization, sports club).

Important Life Realms (ILR)

Participants were asked to rank order the importance of each of seven life realms, including (a) a good income, (b) a harmonious family life, (c) good health, (d) meaningful pastimes, (e) a good marital life, (f) a strong faith, and (g) many friends and acquaintances. This method of assessing the relative salience of various aspects of life has been shown in the past to predict depression in later life (Braam, Beekman, Deeg, Smit, & van Tilburg, 1997) and to be related to quality of life (Andrews & Withey, 1976; Flanagan, 1982). For the purposes of our study, we focused on a good marital life and a strong faith and recoded the rank ordering so that higher numbers reflected greater importance of these realms.

Self-esteem (RSE)

We used a reduced 5-item version of Rosenberg’s (1965) 10-item Self-Esteem scale. Participants were asked to use a 4-point Likert scale ranging from 1 (almost never) to 4 (almost always) to indicate how often they felt the way described by each item. Scale scores for each participant were computed by reverse-coding appropriate items and summing the responses to all items (Cronbach’s α = 0.77). The psychometric properties of the scale have been reported by Crandall (1973) and Dobson, Goudy, Keith, and Powers (1979).

R-UCLA Loneliness scale

The R-UCLA has been shown to possess construct validity (Russell et al., 1980) as a measure of general loneliness and degree of satisfaction with one’s social network. Examples of the items are, “I lack companionship” and “I feel in tune with the people around me.” Each of the 20 items is rated on a scale of 1 (never), 2 (rarely), 3 (sometimes), and 4 (often). After reverse scoring appropriate items, loneliness scores were calculated by summing all items (Cronbach’s α = 0.91). Subscales corresponding to the three self-related facets of loneliness/social connectedness (isolation/intimate connectedness, relational connectedness, collective connectedness) were computed as has been described previously (Hawkley et al., 2005).

Religious Weil-Being scale (RWBS)

The Spiritual Weil-Being scale (Paloutzian & Ellison, 1982) consists of two components: (a) religious well-being, which assesses satisfaction with one’s relationship with God, and (b) existential well-being, which assesses purpose in and satisfaction with life (without explicitly referring to a religious component). Participants were asked to rate 20 statements on a scale of 1 (strongly disagree) to 6 (strongly agree). For our purposes, we summed responses to the 10 items constituting the Religious Well-Being scale (Cronbach’s α = 0.94), which includes items such as, “I have a personally meaningful relationship with God” and “I feel most fulfilled when I’m in close communion with God.”

Center for Epidemiologic Studies Depression scale (CES-D)

The CES-D is a 20-item self-report questionnaire that assesses depressive feelings and behaviors experienced during the previous week. Radloff (1977) reported good psychometric properties of this scale. In our sample, Cronbach’s α was 0.89. Responses to each item were recorded using a 4-point Likert scale that ranged from 0 (rarely or none of the time) to 3 (most or all of the time). One item in the CES-D asks whether the respondent felt lonely during the past week, so this item was deleted prior to calculating total score on the CES-D to ensure that any prediction of depressive symptomatology by loneliness was not due to item overlap. We refer to this total score as CESD-ML. A depressive symptoms score was computed by summing the responses to the remaining 19 items, yielding a scale score range of 0 (low depressive symptoms) to 56 (high depressive symptoms). Six cases were missing information on the CESD-ML.

Satisfaction with Life scale (SWLS)

The SWLS consists of five statements (e.g., “In most ways, my life is close to my ideal. “) that the respondent is asked to rate on a 7 point Likert-type scale from 1 (strongly disagree) to 7 (strongly agree; Diener, Emmons, Larsen, & Griffin, 1985). A satisfaction with life score was calculated by summing responses to the five items (Cronbach’s α = 0.86).

Personality

The Big Five (Goldberg, 1992) questionnaire consists of five sets of 20 adjectives (10 positively and 10 negatively worded) that are used to calculate scores on each of five personality dimensions. Using a 9-point Likert scale that ranges from 1 (extremely inaccurate) to 9 (extremely accurate), the instrument asks participants to rate how each of the trait words describes how they see themselves at the present. We administered 60 items to calculate mean scores for three of these dimensions: surgency (Cronbach’s α = .61), agreeableness (α = .67), and emotional stability (α = .63).

Temporal stability of satisfaction in each relationship domain

Sub-scale scores were computed for Year 2 data as they were for Year 1 data. The reliabilities of 2-item subscales were acceptable as indicated by similarly high Cronbach’s αs as were observed for Year 1: Person Satisfaction (0.78), Group Satisfaction (0.84), God Satisfaction (0.87), and Pet Satisfaction (.65). Means for the subscales in Year 1 and Year 2 are displayed in Table 5. Spearman correlations of relationship satisfaction across the 1-year time interval indicated moderate to high temporal stability of satisfaction with the single most important person, God, and pet, whereas group satisfaction exhibited relatively modest temporal stability (see Table 5).

Convergent and discriminant validity

Validation of the MDRS sub-scales entailed first-order and partial correlation analyses with the validation measures described above. Partial correlations between the criterion variables and each of the person, group, God, and pet satisfaction subscales were conducted holding constant the remaining subscales. Results are provided in Table 6. Because pet-owners constituted a small subsample of the total sample, correlations involving satisfaction in the pet domain were sometimes nonsignificant even when the correlation coefficients were as large as or larger than those observed for relationship domains assessed in the total sample. We temper our interpretation of results accordingly.

Person satisfaction

As hypothesized, the largest correlation with person satisfaction was with loneliness, a measure of overall satisfaction with the quality of one’s relationships. We have argued that the Isolation dimension of loneliness reflects feelings of dissatisfaction of the social self at a deeply personal level, whereas the Relational Connectedness dimension reflects dyadic closeness and support, and the Collective Connectedness dimension reflects collective self and group belonging (Hawkley et al., 2005). As evidence of discriminant validity, person satisfaction was somewhat more highly correlated with the Isolation dimension of loneliness than with the Relational and Collective dimensions, and partial correlations enhanced these distinctions.

Additional evidence of validation was provided by significant associations of person satisfaction with marital status (i.e., being married or living with a partner) and the importance of a good marital life, associations that were not evident for group satisfaction. Person satisfaction was also associated with higher self-esteem, a marker of fulfillment of one’s social belonging needs (Leary, Tambor, Terdal, & Downs, 1995). Moreover, partial correlations indicated that self-esteem tended to be associated with person satisfaction but not with group satisfaction, a finding that speaks to the primacy of having at least one satisfactory personal relationship.

Group satisfaction

Group satisfaction, like person satisfaction, was reflected in better overall quality of social relationships as indexed with the UCLA loneliness scale. Group satisfaction was more highly correlated with Collective Connectedness and Relational Connectedness than with Isolation, and this distinction was even more pronounced in the partial correlations.

The group satisfaction subscale was associated with measures of social integration and group participation (i.e., number of close friends and relatives, number of non-religious group memberships, religious attendance), associations that in large part were retained when person, God, and pet satisfaction were held constant. Religious service attendance was also associated with person satisfaction, but this correlation did not withstand statistical adjustment for group, God, and pet satisfaction. In general, partial correlations indicated that higher levels of social integration contributed more robustly to group satisfaction than to person satisfaction.

Group satisfaction retained a significant partial correlation with religious service attendance, but God satisfaction also exhibited a significant partial correlation with religious service attendance. Similarly, group satisfaction and God satisfaction were both associated with the importance of a strong faith. This pattern of correlations suggests that the relevant group for group satisfaction may be a religious or church group. Frequent attenders of religious services may identify a religious group as their most important group, and these same individuals would be expected to be more satisfied with both their group and their God relationship.

God satisfaction

Convergent validation for the God satisfaction sub-scale was provided by associations with religious service attendance and the importance of a strong faith, as well as a high correlation with religious well-being. The partial correlation between God satisfaction and religious well-being remained large and statistically significant when person, group, and pet satisfaction were held constant. Unexpectedly, God satisfaction was also inversely associated with the importance assigned a good marital life.

Pet satisfaction

We had no hypotheses regarding associations with pet satisfaction, and among the validation measures, only one significant association was observed: Pet satisfaction was inversely associated with marital status. This correlation was not diminished when person, group, and God satisfaction were held constant. Pet satisfaction also exhibited a correlation with the religious well-being scale (although nonsignificant in this small sample). This correlation was substantially diminished, however, after adjusting for person, group, and God satisfaction.

Depressive Symptoms (CESD-ML)

Not only high social interest (Bass, Curlette, Kern, & McWilliams Jr., 2002; Ferguson, 2006) but also qualitatively satisfying social relationships are known to protect against depressive symptoms (Antonucci et al., 2002; Cacioppo et al., 2006), and analyses of mean relationship satisfaction across domains was generally consistent with these findings. The four-domain mean (i.e., including pet satisfaction) revealed a nonsignificant inverse correlation with depressive symptoms (r(82) = −.12, p = .3), and the three-domain mean revealed a larger and significant inverse correlation with depressive symptoms (r(206) = −.28, p < .001). Correlations with the individual measures of relationship satisfaction revealed that person and group satisfaction were each significantly associated with lower levels of depressive symptomatology. Moreover, partial correlations augmented the inverse association between depressive symptomatology and person satisfaction, and partial correlations abolished associations with satisfaction in other relationship domains. Neither pet nor God satisfaction was correlated with depressive symptoms.

Satisfaction with Life scale

Happiness is associated with good social relations; indeed, Diener and Seligman (2002) reported that among the predictors they tested, “No variable was sufficient for happiness, but good social relations were necessary” (p. 81). In our study, life satisfaction was correlated with overall relationship satisfaction, whether measured across all four or only three domains (r = .26, N = 81; r = .36, N = 199, respectively). When examined separately by relationship domain, however, person and group satisfaction but not God or pet satisfaction, were significantly correlated with life satisfaction. As was true for depressive symptoms, partial correlations showed that only person satisfaction was uniquely associated with life satisfaction.

Individual differences in satisfaction within, across, and among relationship domains

An initial repeated measures analysis of variance (within: person, group, God, and pet relationship domains) among pet owners (N = 84) revealed a significant difference in satisfaction among relationship domains (F(3, 249) − 8.536, p < .001). This was attributable to greater person and God satisfaction than group satisfaction (p < .001) and greater God satisfaction than pet satisfaction. Pet satisfaction did not differ from person or group satisfaction, nor did person satisfaction differ from God satisfaction (p > .05). This analysis was repeated comparing only the person, group, and God relationship domains, and results using this larger sample (N = 210) mirrored those seen using the subsample of pet owners (see Table 5 for the means of satisfaction in each domain). Subsequent multivariate analyses focused on the larger sample available for the three-domain models. Models that employed all four relationship domains in the subsample of pet owners produced a similar pattern of results; these results are available from the first author upon request.

Personality differences

In the multivariate model examining personality variables, agreeableness was associated with significantly higher satisfaction ratings across relationship domains (F(1, 189) = 19.012, p < .001). The between-subject effects of surgency and emotional stability were not significant The pattern of domain differences in relationship satisfaction was attenuated and nonsignificant when personality variables were held constant (F(2, 378) = 1.865, p > . 1). This attenuation could not be attributed to domain-specific effects of the personality variables on relationship satisfaction. Neither agreeableness, nor surgency, nor emotional stability interacted with relationship domain to predict differential satisfaction ratings (p > .2 for each).

Demographic differences

In the multivariate model examining demographic variables, between-subject effects were significant for gender (F(1, 203) = 15.270, p < .05) and household income (F(1, 203) = 5.166, p < .05). Women and those with a higher household income reported significantly higher satisfaction ratings across relationship domains. Holding demographic variables constant, satisfaction ratings differed among domains (F(2, 406) = 4.208, p < .05), and the pattern of differences mirrored that of the unadjusted means. In addition, age, household income, and Black race/ethnicity exhibited significant interactions with relationship domain holding constant all demographic variables (p < .05). Consistent with socioemotional selectivity theory (Carstensen, 1992), age was associated with significantly higher ratings of person satisfaction, β (unstandardized) = 0.06 (SE = 0.02, p = .025). Age was not associated with ratings of group satisfaction, and contrary to some previous findings (e.g., Wink & Dillon, 2002), age was not associated with God satisfaction (p > .9) in this sample of older adults. Income was associated with higher ratings of person satisfaction, β = 0.41 (SE = 0.12, p= .001) but was not associated with ratings of group and God satisfaction. Consistent with prior findings (e.g., Ortega, Crutchfield, & Rushing, 1983), God satisfaction was greater in Black participants than White participants, β = 0.70 (SE = 0.22, p = .001), in a model that adjusted for differences in age, gender, education, and household income. Person and group satisfaction did not differ among racial/ethnic groups.

This research was supported by an award from theTempleton Foundation, and by grants from the National Institute of Aging, P01-AG18911 and R01-AG034O52.