ORIGINAL ARTICLES

Altruistic Social Interest Behaviors Are Associated With Better Mental Health

From the Division of Preventive and Behavioral Medicine, Department of Medicine, University of Massachusetts Medical School, Worcester, Massachusetts (C.S., Y.M., G.R.), and MGH Institute of Health Professions at Massachusetts General Hospital, Charlestown, Massachusetts (J.B.M.).

Address reprint requests to: Carolyn E. Schwartz, ScD, 31 Mitchell Road, Concord, MA 01742. Email: carolyn.schwartz{at}deltaquest.org

Key Words: social interest, • altruism, • social support, • mental health, • physical health, • spirituality, • stress, • response shift.

Although social interest and altruism are considered desirable personality traits and indeed are the cornerstone of mental health in Adlerian psychology (1), their impact on physical and psychosocial well-being has only recently been the focus of empirical investigation. Social interest can be defined as "a cooperative approach toward life and a striving for ideal community" (2) and reflects the "capacity to value the interests and welfare of others even when these have no personal utility" (3). Related to the psychoanalytic concept of generative altruism (4), these constructs reflect a pleasure in fostering the success or welfare of another. Recent work in the emerging discipline of positive psychology has identified altruism as an adaptive mental mechanism that is a mature psychological defense (5). Independent of education and social privilege, altruism can regulate people’s perceptions of those internal and external realities they are powerless to change, and empower them to effect meaningful change (5). The present work investigates the effects on physical health and psychological well-being of altruism as reflected by social interest behaviors.

To date, empirical investigations of the benefit of social interest or altruism suggest that they are associated with better life adjustment (6, 7), perceived meaningfulness of life (8, 9) and marital adjustment (10), and less hopelessness (11) and depression (12). Additionally, social interest has been shown to be a predictor of physical health status (13) and a moderator of life stress (14). In general, helping behaviors that aid survival and material wealth for close relatives are considered more rational, ethical, and obligatory than helping behaviors aimed at nonfamily members (15). However, recent research on preventive behavioral interventions has identified altruism concern for others as a prevalent and salient motivator for engaging in safe sex to prevent human immunodeficiency virus transmission (16), suggesting that focusing on intimacy, affiliation, and altruism has population-level adaptive advantages over focusing on achievement and power. Both orientations have been called "equal and essential partners in human evolution and development" (17), although research comparing psychological well-being and life satisfaction in people with either orientation revealed no between-group differences (18). There is some speculation, however, that the health ramifications of the type A personality (19) are caused by a lack of social interest (1) that may result from not having obtained a sense of belonging, and developing instead an achievement-oriented egocentricity (20).

In religious communities, the social connections may be particularly effective stress buffers and stress deterrents (21). This effect may in part be a result of a tendency to use positive religious coping responses when one receives spiritual support from religious community members (22). The benefit of religion may also be a result of a tendency in religious community members to provide help to others. One large study of Japanese elderly found that greater involvement in religion was associated with helping others more frequently, and that older adults who more often provided assistance to others rated their health more favorably than people who were less involved in helping others (23). Krause et al. (23) suggested that helping others may explain at least part of the relationship between religious involvement and better self-reported health.

In addition to health benefits, helping others may bring a psychosocial benefit to the helper. A 2-year follow-up of patients with multiple sclerosis providing a supportive ear to others was associated with three to seven times the benefit in quality-of-life outcomes (eg, role performance, self-efficacy, coping ability, well-being) among those giving as compared with receiving help (24). The effects were particularly pronounced in aspects of subjective well-being such as purpose in life, self-acceptance, and personal growth. Qualitative interviews exploring these effects revealed that by helping others, the patients were able to reframe their own suffering, derive a stronger sense of meaning in life, and feel a stronger awareness of the existence of a higher power (24). That is, their suffering became a vehicle for helping others (24). Research on the beneficial effects of volunteerism has also suggested that the psychological and physical benefits of volunteer work among the elderly may be a result of the feeling of continuity, involvement, and usefulness (25, 26), and a status-bearing role (27). Thus, in concert with the well documented beneficial effects of receiving social support, the practice of helping others seems to be beneficial. There is a substantial literature that suggests that overdoing a prosocial behavior like helping others can have significant psychological and physical costs (28–31) and that having some balance in the giving and taking of help may be very important.

Thus, there is a growing literature that suggests that altruism or social interest behaviors may have adaptive advantages for the helper. To our knowledge, however, no research has yet addressed whether helping others is more beneficial than receiving help. Further, because social interest behaviors are encouraged in religious communities, it would be worthwhile to investigate whether a marginal benefit of helping others is detectable even in a sample that is more likely than the average population to engage in helping others. The present work sought to evaluate whether helping others was more beneficial than receiving help in a secondary analysis of a large sample of Presbyterian Church members, elders (ie, lay leaders), and pastors. Another aim of the present work was to investigate the additional impact of being overloaded or having a balance of helping others and receiving help. Finally, if helping others was predictive of better quality-of-life outcomes, the present work sought to identify characteristics related to a higher probability of engaging in such altruistic social interest behaviors.

Sample
Members and elders were drawn in a two-stage process in 1996. Elders were lay leaders in the church who invested time and energy into a variety of religious tasks, from assistance in religious services and program planning to membership recruitment and building maintenance (32). They were likely to define themselves by other occupational roles and their positions in the church (32). In the two-stage sampling process, 425 congregations were selected from a population of 11,361 by proportional sampling to give equal representation to congregations of different sizes. For selection of members, the individual congregations were asked to compile a numbered alphabetical list of names of members. The researchers generated seven random numbers for each congregation to correspond to the names of members. The congregations were asked to send the names and addresses of those identified members to the researchers. Seventy-three percent of congregations sent in names, yielding a total of 2163 names, and these people were sent the background survey. Sixty-three percent of members returned the background survey (1360 members, 1314 elders) and were then sent the 1997 survey questionnaire. According to the Presbytery, 1360 members were sent the 1997 questionnaire, and 1025 (75%) returned completed forms; 997 (73%) had complete data for analysis herein. Elders were drawn from the same list of sampled churches, but the names of the elders were obtained from the national offices. Four or five elders from each of the 425 congregations were randomly drawn, resulting in 1759 elders, of whom 1314 (75%) returned the first background survey and were sent the 1997 questionnaire. According to the Presbytery, 1034 (79%) returned the questionnaire; 1019 (78%) had complete data for analysis. This study examines the second questionnaire in the series, administered in 1997. Questionnaires were mailed to the members’ homes and were returned by mail to the Research Services Department of the Presbyterian Church in Louisville, Kentucky. Consent was attained by a completed, returned questionnaire.

Measures
Altruistic social interest was operationalized as giving help to others by summing answers to two Likert-scaled questions that asked how often the person had "made others feel loved and cared for" and "listened to others’ concerns." Receiving help was operationalized by summing answers to two questions that asked how often the "congregation made me feel loved and cared for" and "congregation listened to you talk about private concerns." The responses for each of these questions were 1) never, 2) once a while, 3) fairly often, and 4) very often. Higher summary scores reflecting receiving or giving more help. The balance between giving help and receiving help was evaluated using a ratio of the two summative scores (balance), and feeling overwhelmed by others’ demand (overload) was evaluated by summing answers to two questions that asked how often the congregation "made too many demands on you" and "has been critical of you and the things you have done." Higher overload scores reflected feeling more overwhelmed.

Mental and physical health was measured by the Short Form 36 Health Survey, developed from the Medical Outcomes Study (33). This widely used instrument has extensive evidence for its reliability and criterion validity from various patient and nonpatient populations across cultures (34–39). The physical and mental health scores generated by this measure have possible ranges of 0 to 100.

Positive and negative religious coping was measured by the religion coping measure, the Brief RCOPE (40), which contains positive coping statements (eg, life as part of spiritual force; God as partner; God for strength and guidance; tried to find the lesson; tried to give strength to others) and negative coping statements (eg, God was punishing me; God had abandoned me; questioned God’s existence; expressed anger at God). Positive and negative coping scores were created using principal component analysis. Eleven types of prayer were listed for members to rate according to frequency of use. Stressful events measured included death of a family member, major financial loss, major illness or injury, major disagreement with family members, major disagreement with close friend, and "any other major problem or challenge." A count of these events was an event summary score. Demographic characteristics captured included age, gender, marital status, race, education, employment status, and income.

Statistical Analysis
The present work used bivariate and multivariate linear regression and nested models with likelihood ratio tests (41) to examine how giving vs. receiving help was associated with self-reported physical functioning and mental health and to identify predictors of helping others. The distributions of both physical functioning and mental health scores were highly skewed in the entire sample. For the physical functioning scores, 31% of the respondents had a perfect score (ie, 100), and 64% had physical functioning scores of 90 or more. No transformation would provide symmetric or Gaussian distributions with this type of grouped data. To model physical functioning, we created a dichotomized variable breaking the sample into respondents with scores of less than 90 and greater than or equal to 90. In addition, we performed a "survival-like" analysis as suggested by Dudley et al. (42).

Models investigating mental health adjusted for the significant predictors that were identified in previous work performed by our group using the same data set. This work suggested that better mental health was associated with higher levels of reported positive religious coping, lower levels of reported negative religious coping, and reporting that one had asked God for healing, after adjusting for age, gender, income, general health, and a summative index of stressful events (R² = .26). This summative index included those events that exhibited some variability in the sample, specifically financial loss, disagreement with a family member, and disagreement with a close friend.

A cubic transformation of mental health scores provided a symmetric distribution. This value was then standardized by subtracting the mean and dividing by the standard deviation. This result provided an interpretation of regression coefficients in terms of the standard deviation of the transformed value. The relationship between age and mental health was best fit by a cubic function, such that self-reported mental health was lowest at approximately age 25 years and peaked at approximately age 75 years. Between age 25 and 75 years, mental health seemed to increase. Because of this curvilinear relationship, three age-related parameters were included in the model (ie, age, age squared [age²], and age cubed [age³]).

After examination of unadjusted association of factors with the outcomes, model building for linear and logistic regression was performed with the emphasis on the adjusted association of the primary factors of giving help, receiving help, and overload. We eliminated factors with low variability and provided some variable reduction with the use of coping scores. We first estimated a model without the primary factors. We used backward elimination to create the initial model. We examined factors eliminated —in particular, ones associated in an unadjusted fashion —to determine what adjustment eliminated them and to determine that it made clinical sense. With this best model, we tested the addition of our primary factors using a likelihood ratio test. We realize that this selection process based on p values may result in biased estimates, and strict inferential interpretation of p values may not be appropriate given the multiple premodel testing and selection (43). We considered the full model with all factors. In each case, the primary factor remained significant, and the coefficient estimate was not different by more than 5%. We investigated the primary factors separately and then combined all in the model. When there was moderate correlation between factors, we used conditional regression analysis to investigate the associations. This analysis is suggested by the coplots by Cleveland (44).

Characteristics of Study Subjects
The original sample included 2055 members and elders who completed a baseline questionnaire regarding prayer activities, religious coping, health and well-being, and sociodemographic characteristics. Thirty-nine subjects with missing values of mental health or physical functioning were excluded. This exclusion left 997 members and 1019 elders for this analysis. Participants were predominately white, married, employed, and educated (Table 1). The average age was 56 years. Approximately 85% of participants had annual income greater than $30,000. Subjects missing demographic data did not have a significantly different mental health score from those with demographic information.

Descriptive Statistics of Predictor and Dependent Variables
Descriptive statistics of the variables used to predict the physical functioning and mental health scores are shown in Table 2. Participants in this study tended to have high physical functioning and mental health scores (median = 95 and 84; mean = 85.3 and 78.5, SD = 19.8 and 14.6, respectively). Before participating in the study, 8% of subjects suffered major financial loss, 16.3% had a major disagreement with a family member, and 5.62% had a major disagreement with a friend. Most participants did not encounter any of the stressful events studied, and the average sum of these events was 0.44 (SD = 0.69), with 66% reporting no events. Approximately 70% of the sample reported having asked God for healing for themselves, and the majority of participants believed that the congregation made them feel loved "very often" and listened to them talk about their private concerns "once in a while" or "fairly often" (receiving help mean = 5.8, SD = 1.6). Participants reported giving help slightly less often (mean = 5.7, SD = 1.5), although receiving and giving help were moderately correlated (r = .68, p < .001), and the mean ratio of receiving and getting help (balance) was 1.01 (SD = .27). A small percentage of participants believed that the congregation made too many demands, and was critical "fairly often" or "very often" (12% and 4%, respectively; overload mean = 3.1, SD = 1.2).

Relationship of Giving Help vs. Receiving Help to Reported Physical Functioning
In the bivariate analysis, we found that giving help, receiving help, positive religious coping, negative religious coping, age, and gender were significantly associated with reported physical function. In contrast, balance, race or ethnicity, and level of education were not significantly associated with reported physical function. In the final multivariate model, age and gender remained significantly associated with physical function; all other unadjusted associations were caused by the association of these factors with age or gender. Older people and females reported lower physical functioning in this sample.

Relationship of Giving Help vs. Receiving Help to Reported Mental Health
In the bivariate analysis, we found that receiving help was significantly associated with better mental health (t = 5.3, p < .0001, R² = .014), a relationship that was maintained after adjusting for the sociodemographic, health-related, and religious coping covariates. A likelihood ratio test comparing models with and without receiving help revealed that the two models were significantly different, and receiving help was a significant contributor to predicting mental health (χかい²(1) = 5.48, p < .01). The addition of this variable to the full model did not change the association of the other predictors in the model, with the largest change being a 10% decrease in the positive coping coefficient.

In the bivariate analysis, we found that giving help was significantly associated with better mental health (t = 7.49, p < .0001, R² = .028), and that this relationship was maintained after adjusting for the sociodemographic, health-related, and religious coping covariates (R² = 26%). Adding giving help to the model did not appreciably change the association of the other predictors in the model, although the positive religious coping coefficient decreased in magnitude by 20%. A likelihood ratio test comparing the full model with and without giving help revealed that the two models were different and that giving help contributed to predicting mental health.(χかい²(1) = 12.9, p < .001).

In both the bivariate and multivariate analyses, balance was not significantly associated with mental health. In contrast, overload was significantly associated with worse mental health in both the bivariate (t = -6.21, p < .0001, R² = .019) and multivariate analyses (R² = .26, p < .0001). Because giving help, receiving help, and overload were significant predictors of mental health, we tested a model with all three variables, after adjusting for covariates. Receiving help was no longer significant in this model, reflecting the moderate correlation between giving and receiving help in this sample (r = .68, p < .00001). Adding overload to the full model did not appreciably change the magnitude of the coefficients for the other predictors but did increase the magnitude of the coefficient for overload (from -.09 to -.11), suggesting that the negative influence of feeling overwhelmed by others’ demands was stronger after adjusting for simply giving or receiving help. A likelihood-ratio test revealed that overload was a significant contributor to the model predicting mental health (χかい²(1) = 22.3, p < .001). We thus removed the receiving help score from the model, arriving at a final model (Table 3) that demonstrated that giving help and overload were significantly associated with mental health score, independent of demographics, prayer activities, and prayerful coping. Giving help was associated with better mental health, whereas reported overload was associated with worse mental health. An examination of the model residuals revealed a symmetrical distribution, suggesting that the final model predicting mental health is appropriate.

These analyses suggested that giving help was a more important predictor of reported mental health than receiving help. Given the moderate association between the two variables (r = .68), we investigated whether this selection of variables in the final model was a statistical artifact or was meaningful and could thus be interpreted. Accordingly, conditional regression analysis was performed, looking at the association of one score of mental health within each level of the other score. This analysis revealed that the association of mental health was stronger with the giving help score than the receiving help score, and confirmed that the greater importance of giving help is not a statistical artifact caused by shared variance but rather reflects a substantial difference in the patterns of association.

Predictors of Giving Help
Because giving help was such a salient predictor of mental health, we sought to understand what sociodemographic, prayer activities, and religious coping behaviors were associated with this construct. Table 4 presents unadjusted associations between possible predictors and giving help. All of the prayer activities, satisfaction with prayer life, positive and negative religious coping, age, and gender were associated with giving help. In contrast, frequency of Sunday worship, race or ethnicity, and level of education were not associated with giving help. Because each of the prayer activities was associated in a similar fashion, we created a summary score based on a principal component analysis (45) (eigenvalue = 2.7) for use in the multivariate model. This analytic method for data reduction creates a score based on linear combinations of observed variables to represent the largest amount of variance in the data (45). In the multivariate model, prayer activities, satisfaction with prayer life, positive religious coping, age, female gender, and being a church elder were significant predictors of giving help, whereas frequency of prayer and negative religious coping dropped out of the model (Table 5). The effects of each predictor tended to diminish in the full model, suggesting that there is some association among the different factors. In a similar model looking at predicting balance, only prayer activity and female gender were positively associated (data not shown).

Our findings suggest that helping others is associated with higher levels of mental health, above and beyond the benefits of receiving help and other known psychospiritual, stress, and demographic factors. These findings are particularly notable because the sample studied was quite physically and mentally healthy, and thus may have had less of the requisite variability for demonstrating such relationships. The mental health dimension investigated in this work was composed of the anxiety and depression that plagues most people, rather than more pathological mental illness. Thus, our findings would likely have implications for helping clinically normal people experiencing the vicissitudes of adult life (eg, bereavement, work-related or family-related strains or stressors).

The act of giving to someone else is a way of operationalizing Erikson’s (46) concept of generativity, which refers to behaviors that contribute to family, society, and succeeding generations (47) by being willing to share with another person, guiding others, and establishing a priority of needs that recognizes both the self and others (48). We found that people who were more likely to help others were older and female and tended to be church elders; they also practiced more prayer activities, reported more satisfaction with prayer life, and engaged in positive religious coping. These findings suggest that role expectations and a positive belief system underlie and reinforce altruistic social interest behaviors. For example, elders, females, older people, and more religious people would all have social role expectations of giving to others. These role expectations would likely be reinforced by a positive belief system, as reflected by positive religious coping, engaging in more prayer activities, and being satisfied with one’s prayer life.

The act of giving to someone else may have mental health benefits because the very nature of focusing outside the self counters the self-focused nature of anxiety or depression. Schwartz and Sendor (24) propose that the process resulting from altruistic practice facilitates adaptation via response shift phenomena—that is, changes in internal standards, values, or conceptualization of quality of life (49). They suggest that an outer-directed role allows one to disengage from previous patterns of self-reference, and thereby facilitates an openness to changing internal standards, values, and concept definition (ie, response shift phenomena). These shifts are purported to lead to a renewed perspective on one’s life circumstances, such as one’s illness, stressors, or personal loss (24). Alternative explanations could be that people who are functioning well psychologically are better able and hence more likely to give help, rather than the reverse; or that altruism and psychological health share a common cause, such as being influenced by the same set of genes.

Although our findings suggest that people who help others experience better mental health, our findings also suggest that giving beyond one’s own resources is associated with worse reported mental health. We found that feeling overwhelmed by others’ demands had a stronger negative relationship with mental health than helping others had a positive one. A post hoc analysis revealed that giving help and overload had a small but statistically significant correlation (r = .20, p < .00001), and there was no interaction between the two variables in predicting mental health, suggesting that the effects were additive. This pattern is reminiscent of the dysfunction of cancer cells: reproduction is a necessary and beneficial activity unless it lacks self-regulatory limits, and then it becomes pathological and threatens the viability of the organism. Because giving and receiving help tended to go in tandem in our sample (ie, the mean balance score was 1.01 and the correlation between the scores was .68), it is clear that feeling overwhelmed by others’ demands is not simply a question of giving more than one receives, but rather reflects a qualitative difference in the constructs. Past research on earlier Presbyterian Panel data revealed that elders and clergy were more likely both to benefit from social support and to suffer from feeling overwhelmed by others’ demands (22), suggesting that a leadership role in one’s religious or spiritual community can have significant costs.

The limitations of the present work should be noted so that future research can build on our findings and rectify these limitations. First, our findings are necessarily limited by the cross-sectional design, so we are unable to evaluate whether helping others leads to better mental health, or vice versa. We were unable to test causal hypotheses of this nature in the present data because, although 2-year follow-up data were available on the outcomes of interest, the sample was remarkably stable, and thus, there was little change in physical or mental health (42). This lack of variability made it statistically impossible to predict change and thus to test causal hypotheses about the benefits of helping others. Second, the present work is a secondary analysis of the construct of altruistic social interest, so our operationalization was created post hoc and is not an ideal measure of altruism. Third, our sample consisted of volunteers from the Presbyterian Church, so the external validity of our findings would need to be tested among people not so closely identified with a religion. However, this sample was skewed toward high levels of mental health, so detecting such an effect despite the relatively high mental health of the sample suggests that the effect is not simply a result of being better off psychologically.

Although there is a large body of research on the benefits of receiving social support, the benefits of helping others are relatively unexplored. Future observational research could build on the present work by investigating the predictive significance of helping others and identifying cues for excessive or exhaustive altruism. Intervention research may develop interventions that facilitate a person’s involvement with altruistic social interest behaviors directed at peers who are experiencing a particular stressor (eg, bereavement) and evaluate whether altruistic practice buffers the person from psychological distress or somatic symptoms. It is our hope that this work will stimulate and encourage future research on the benefits of altruistic social interest.

The authors gratefully acknowledge the assistance of Jean Edmunds and Lorna Chiasson in facilitating the literature search. We would also like to thank the Research Services Department of the Presbyterian Church, USA, for providing the Presbyterian Panel data.

Received for publication March 15, 2002 ;

1. Rareshide M, Kern R. Social interest: the haves and the have nots. J Individ Psychol 1991; 47: 464–76.
2. Adler A. Superiority and social interest: a collection of later writings. 3^rd rev. ed. New York: Norton; 1979.
3. Crandall JE. Theory and measurement of social interest: empirical tests of Alfred Adler’s concepts. New York: Columbia University Press; 1981.
4. Seelig BJ, Rosof LS. Normal and pathological altruism. J Am Psychoanal Assoc 2001; 49: 933–59.[Medline]
5. Vaillant GE. Adaptive mental mechanisms: their role in a positive psychology. Am Psychol 2000; 55: 89–98.[CrossRef][Medline]
6. Crandall JE, Lehman RE. Relationship of stressful life events to social interest, locus of control, and psychological adjustment. J Consult Clin Psychol 1977; 45: 1208.[CrossRef][Medline]
7. Zarski JJ, Bubenzer DL, West JD. Tasks of life survey (TOLS). Charleston (WV): Counseling and Human Services Associates; 1982.
8. Crandall JE. Social interest as a moderator of life stress. J Pers Soc Psychol 1984; 47: 164–74.[CrossRef]
9. Mozdzierz GJ, Greenblatt RL, Murphy TJ. Social interest: the validity of two scales. J Individ Psychol 1986; 42: 36–43.
10. Markowski EM, Greenwood PD. Marital adjustments as a correlate of social interest. J Individ Psychol 1984; 40: 300–8.
11. Miller MJ, Denton GO, Tobacyk J. Social interest and feelings of hopelessness among elderly patients. Psychol Rep 1986; 58: 410.[Medline]
12. Crandall JE. A scale for social interest. J Individ Psychol 1975; 31: 187–95.
13. Zarski JJ, Bubenzer DL, West JD. Social interest, stress, and the prediction of health status. J Consult Dev 1986; 64: 386–9.
14. Crandall JE. The effect of stress on social interest and vice-versa. J Individ Psychol 1978; 34: 40–7.[Medline]
15. Kruger DJ. Psychological aspects of adaptations for kin directed altruistic helping behaviors. Soc Behav Pers 2001; 29: 323–30.
16. Nimmons D, Folkman S. Other-sensitive motivation for safer sex among gay men: expanding paradigms for HIV prevention. AIDS Behav 1999; 3: 313–24.[CrossRef]
17. Shapiro Y, Gabbard GO. A reconsideration of altruism from an evolutionary and psychodynamic perspective. Ethics Behav 1994; 4: 23–42.
18. Pöhlmann K. Agency- and communion-orientation in life goals: impacts on goal pursuit strategies and psychological well-being. In: Schmuck P, Sheldon KM, editors. Life goals and well-being: towards a positive psychology of human striving. Kirkland (WA): Hogrefe, Huber Publishers; 2001. p. 68–84.
19. Friedman M, Rosenman RH. Type A behavior and your heart. New York: Knopf; 1974.
20. Scherwitz L, Berton K, Leventhal H. Type A behavior, self-involvement, and cardiovascular response. Psychosom Med 1978; 40: 593–609.[Abstract/Free Full Text]
21. Siegel K, Anderman SJ, Schrimshaw EW. Religion and coping with health-related stress. Psychol Health 2001; 16: 631–53.
22. Krause N, Ellison CG, Wulff KM. Church-based emotional support, negative interaction, and psychological well-being: findings from a national sample of Presbyterians. J Sci Study Religion 1998; 37: 725–41.[CrossRef]
23. Krause N, Ingersoll-Dayton B, Liang J. Religion, social support and health among the Japanese elderly. J Health Soc Behav 1999; 40: 405–21.[CrossRef][Medline]
24. Schwartz CE, Sendor M. Helping others helps oneself: Response shift effects in peer support. Soc Sci Med 1999; 48: 1563–75.
25. Carney JM, Dobson JE, Dobson RL. Using senior volunteers in the schools. J Humanistic Educ Dev 1987; 25: 136–43.
26. Fengler AP. Life satisfaction of subpopulations of elderly: the comparative effects of volunteerism, employment, and meal site participation. Res Aging 1984; 6: 189–212.[Abstract]
27. Hunter KI, Linn MW. Psychosocial differences between elderly volunteers and nonvolunteers. Int J Aging Human Dev 1981; 12: 205–13.
28. Krause N, Ellison CG, Shaw BA, Marcum JP, Boardman JD. Church-based social support and religious coping. J Sci Stud Religion 2001; 40: 657–73.[CrossRef]
29. Rook KS. The negative side of social interaction: impact on psychological well-being. J Pers Soc Psychol 1984; 46: 1097–108.[CrossRef][Medline]
30. Pagel MD, William WE, Becker J. Social networks: we get by with (and in spite of) a little help from our friends. J Pers Soc Psychol 1987; 53: 793–804.[CrossRef][Medline]
31. Finch JF, Okun MA, Barrera M, Zautra AJ, Reich JW. Positive and negative social ties among older adults: measurement models and the prediction of psychological distress and well-being. Am J Commun Psychol 1989; 17: 585–605.[CrossRef][Medline]
32. Pargament KI, Tarakeshwar N, Ellison CG, Wulff KM. The relationships between religious coping and well-being in a national sample of Presbyterian clergy, elders, and members. J Sci Stud Religion 2001; 40: 497–513.[CrossRef]
33. Ware JE Jr, Sherbourne C. The MOS 36-Item Short-Form Health Survey (SF-36), I: conceptual framework and item selection. Med Care 1992; 30: 473–81.[Medline]
34. Anderson EM, Bowley N, Rothenberg BM, Panzer R, Katz P. Test-retest performance of a mailed version of the Medical Outcomes Study 36-Item Short-Form Health Survey among older adults. Med Care 1996; 34: 1165–70.[CrossRef][Medline]
35. Essink-Bot ML, Krabbe PF, Bonsel GJ, Aaronson NK. An empirical comparison of four generic health status measures: the Nottingham Health Profile, the Medical Outcomes Study 36-Item Short-Form Health Survey, the COOP/WONCA charts and the EuroQol instrument. Med Care 1997; 35: 522–37.[CrossRef][Medline]
36. Leidy NK, Palmer C, Murray M, Robb J, Revicki DA. Health-related quality of life assessment in euthymic and depressed patients with bipolar disorder: psychometric performance of four self-report measures. J Affect Disord 1998; 48: 207–14.[CrossRef][Medline]
37. Lin M, Ward JE. Reliability of the MOS SF-36 health status measure in Australian general practice. Aust Fam Phys 1998; 27 (suppl 2): 94–8.[Medline]
38. Revicki DA, Sorensen S, Wu AW. Reliability and validity of physical and mental health summary scores from the Medical Outcomes Study HIV Health Survey. Med Care 1998; 36: 126–37.[CrossRef][Medline]
39. Ware JE Jr, Kosinski M, Keller SD. A 12-item short-form health survey: construction of scales and preliminary tests of reliability and validity. Med Care 1996; 34: 220–33.[CrossRef][Medline]
40. Pargament KI, Smith BW, Koenig HG, Perez L. Patterns of positive and negative religious coping with major life stressors. J Sci Stud Religion 1998; 37: 710–24.[CrossRef]
41. Perez-Hoyos S, Tobias A. Sg111: a modified likelihood ratio test command [reprinted in Stata Tech Bull Reprints 1999;9:171–3]. Stata Tech Bull 1999; 49: 24–5.
42. Dudley R, Harrell FE, Smith L, Mark DB, Califf RM, Pryor DB, Glower D, Lipscomb J, Hlatky M. Comparison of analytic models for estimating the effect of clinical factors on the cost of coronary artery bypass graft surgery. J Clin Epidemiol 1993; 46: 261–71.[CrossRef][Medline]
43. Harrell FE. Regression modeling strategies with application to linear models, logistic regression, and survival analysis. New York: Springer; 2001.
44. Cleveland WS. Visualizing data. Summit (NJ): Hobart Press; 1993.
45. Kim JO, Mueller CW. Factor analysis: statistical methods and practical issues (Series/Number 07–014). Beverly Hills (CA): Sage University Press; 1978.
46. Erikson EH. Childhood and society. New York: WW Norton and Co; 1963.
47. Kozier B, Erb G, Olivieri R. Fundamentals of nursing: concepts, process and practice. 4th ed. Redwood City (CA): Addison-Wesley; 1991. p. 587–9, 593.
48. Fuller J, Schaller-Ayers J. Health assessment: a nursing approach. 2nd ed. Philadephia (PA): JB Lippincott; 1994. p. 550.
49. Sprangers MAG, Schwartz CE. Integrating response shift into health-related quality-of-life research: a theoretical model. Soc Sci Med 1999; 48: 1507–15.