Proposal research about "the protective effects of religiosity on depression and anxiety"

Sot Scr Med Vol 3.2, No 11, pp 1257-1262, 1991 Pnnted m Great Bntam All nghts reserved

0277-9536/91 $3 00 + 0 00 Copyright 0 1991 Pergamon Press plc

RELIGION AND PSYCHOLOGICAL DISTRESS IN A COMMUNITY SAMPLE

DAVID R WILLIAMS,’ DAVID B LARSON,* ROBERT E BUCKLER,.’ RICHARD C HECKMANN~ and CAROLINE M PYLE’

I Departments of Sociology and Epldemlology and Pubhc Health, Yale Umverslty, P 0 Box 1965, Yale Statlon, New Haven, CTO6520, US A, *NatIonal Institute of Mental Health, 5600 Fishers Lane, Rockvdle, MD 20857, U S A, ‘Department of Psychology, Western Seminary, 5511 E Hawthorne, Portland, OR 97215, U S A 4Department of Psycluatry, Umverslty of Colorado School of MedIcme, 4200 E 9th Ave. Denver, CO 80262, U S A and SDepartment of Epldemlology and Pubhc Health, Yale School

of Medlcme, 60 College Street, New Haven, CT 06520, U S A

Abstract-This paper exammes the effect of rell@ous attendance and affibatlon on psychologlcal distress m a lonptudmal commumty study of 720 adults Rebglous atlibatlon IS unrelated to mental health status In contrast, although rehgous attendance does not directly reduce psychologlcal distress, It buffers the deletenous effects of stress on mental health That IS, m the face of stressful events and physlcal health problems, rebgous attendance reduces the adverse consequences of these stressors on psychologlcal well-bemg

Key words-rehglon, psychologlcal &stress, stress

The relationshIp between rehglon and health status has been recelvmg mcreasmg sclentlfic attention in recent years One mdlcator of this interest 1s the growing number of reviews focused on rehglous vanables that have appeared m the medical and social science hterature [l-7] In terms of mental health outcomes, the literature indicates that more often than not, religion measures are inversely associated with indicators of psychological distress Bergm [S] revlewed 26 studies that assessed the assoclatlon be- tween rehglon and mental health status He reported that almost half of the studies found an inverse assoclatlon between rehglon and psychological symp- toms with the remainder about equally divided be- tween those that found a positive relationship and those that reported no assoclatlon However, given that 80% of the studres reviewed by Bergm [S] utlhzed student samples, it IS difficult to draw conclustons about the generahzablhty of these findings

Studies employmg more representative samples present a slmllar mixed pattern of findings Two studies based on national probability samples have reported an inverse assoclatlon between rehglous attendance and psychological distress [8,9] Slml- larly, several community studies have reported m- verse assoclatlons between measures of religion and scores on screening scales of global distress [IO-131 At the same time, other community studies report no assoclatlon between rehglon and mental health status [14-161

The hterature assessing the mental health conse- quences of rehglon IS plagued with conceptual and methodological hmltatlons which require that great caution should be exercised m mterpretmg the find- mgs For example, with few exceptions [16, 171 most of the existing studies have used cross-sectional designs m which rehglous mvolvement and mental health status are measured simultaneously A given level of psychological functlonmg can be either

a cause or a consequence of rehglous behefs and behavior In cross-sectional analyses it 1s lmposslble to detect causal dlrectlonahty m the relatlonshlps observed Researchers have also given inadequate attention to the measurement of the rehglous vanabie and to the underlying processes by which rehglon may affect health status [2, IO, 17-201

One way m which rehglous mvolvement may affect health status 1s by modlfymg the relatlonshlp between stress and illness Stress has been shown to have pervasive negative effects on physical and mental health [21], but psychosocial resources can compen- sate for or moderate the impact of stress on health [22] Recently, Krause and Van Tran [23] docu- mented that rehglous mvolvement 1s a cntlcal psycho- social factor that counteracts the adverse effects of stress on feelings of self-esteem and mastery The literature on stress recognizes that a given psychoso- cial resource, such as rehglon, may affect psychologl- cal distress by directly enhancing mental health status, lrrespectlve of the level of stress, and/or by buffenng the effects of stress on health [22] The buffenng hypothesis postulates that m the face of stress, religion can protect the mdlvldual from the potentially negative consequences of stress To our knowledge, there have been no attempts to empln- tally assess the dynamics of the assoclatlon between religion, stress and psychological distress

This paper seeks to enhance our understandmg of the relatlonshlp between rehglous behavior and mental health by exammmg how two measures of rehglous mvolvement, rehglous attendance and reh- glous affiliation, combme with stress to affect psycho- logical distress In 1967, a random sample of rest- dents of metropolitan New Haven were mtervlewed Lmdenthal et al [13] have reported on the cross- sectional assoclatlons between rehglon and mental health status They found that both rehglous affiha- tlon and rehglous attendance were inversely assocl-

1257

1258 DAVID R WILLIAMS et ~1

ated with psychological distress Two years later, a second wave of data was collected from these New Haven residents To date, no analyses have related the 1967 religion measures to distress m 1969 In addition, although controls were utilized for SOCIO- demographic vanables m the ongmal study, no attempts were made to assess the extent to which the assoclatlon between religion and psychological distress vanes for structural charactenstlcs such as race or socloeconomlc status A growing body of evidence indicates that stress, the resources to cope with stress, and the efficacy of these resources vary for groups occupying different structural positions m society [22]

This paper focuses on the ongmal respondents who were reinterviewed m 1969 We assess the extent to which the pattern of findings m the cross-sectional analyses remam robust m the more ngorous prospec- tive analyses Specifically, we address the followmg research questions

1 How do religious attendance and affiliation relate to psychological distress?

2 Do the consequences of religious mvolvement vary by major soclodemographlc charactenstlcs such as age, race, education and gender7

3 To what extent can measures of rehglous mvolvement buffer or moderate the effects of stress on health?

METHODS

The analyses reported here use data from the Myers et al [24,25] lonBtudma1 study of mental health m New Haven, Connecticut The sample con- sists of 720 adults who were reinterviewed m 1969 from an ongmal random sample of 938 respondents who were first interviewed m 1967 Table 1 lists the means, standard deviations and mtercorrelatlons among the vanables utilized Our sample IS 44% male, 11% black, 26% unmarned, and has a median education level of 12 years and a mean age of 44 8 years

Psychological distress 1s measured by the Gurm ef al [8] symptom checklist scale This scale consists of 20 statements of psychophysiological symptoms that indicate the presence of moods of depression and anxiety. The symptoms of the Gurm scale were selected from among those most frequently mentioned by patients m treatment and they allow

for respondents to be ordered on a contmuum of reported distress Respondents reported the fre- quency with which each symptom was expenenced Scores on the Gunn scale thus range from 20 (all symptoms expenenced ‘often’) to 80 (all symp- toms occurrmg ‘never’) In contrast to our use of the Gunn scale as a contmuous measure, the scale 1s sometimes used qualitatively to dlstmgmsh between the mentally impaired (score = 66 or lower), and the non-impaired We believe that our contmuous measure of psycholoDca1 distress 1s more theoreti- cally appropnate for the study of the assoclatlon between religion and mental health than a more qualitative dlstmctlon between psychlatnc cases and normals If rehglon has positive effects on mental health, they are hkely to be evident throughout the continuum of mental health status and not only at the extreme of the dlstnbutlon

Two measures of rehglous commitment at wave one (1967) are utilized Religious attendance measures the usual frequency of attending rehglous services (values range from 1 = never to 6 = more than once a week) To facilitate interpretation of product terms m the regression analyses, the religious attendance measure was converted to a standard score based on the mean and standard devlatlon of the total sample, and a constant was added to this standardized vanable so that the lowest actual value 1s zero The religious affiliation measure IS based on the response to the questlon “Are you affiliated with any church or religious group?” (1 = yes, 0 otherwise)

Two summary measures of stressful life expen- ences, occurnng dunng the two years between the mtervlews, are utlhzed Both measures of stress are listed m the Appendix The first 1s an index of undesirable life events The second stress measure 1s a sum of the number of physical health problems experienced To avoid confounding between the measure of psychological distress and the health problems index, followmg Kessler and Cleary [26], we excluded those health complamts that mtultlvely appeared to have a strong psychosomatic compo- nent From a list of 44 symptoms, we selected those 16 health complamts for which a psychosomatic component would be muumal

Ordinary least squares (OLS) regression analyses utlhzmg the regression program m SAS [27] are used for estimating the magmtude and statistIca slgmficance of the relatlonshlps among religious

Table I Means, standard dewatmn and mtercorrelatmns (dwmals onutted) among vdrldbles

I 2 3 4 5 6 7 8 9 IO II I2 Standard

Mean dewauon

I Age - 2 Education’ -39 - 3 Marital staus (married) -08 06 - 4 Gender (male) IS 05 I4 5 Race (black) -15 -15 -20 6 Attendance 1961 -01 09 03 7 Attendance 1969 03 02 02 8 Health problems 20 -15 -07 9 Life events -II -04 -01

10 Gunn 1967 06 I5 I2 II Gunn 1969 00 I5 06 I2 Affihauon 1967 02 I2 08

- -09 -10 -13

02 -01

I4 IO

-04

44 81 I6 61 3 37 I 50 074 044 044 0 50

- 0 II 0 32 -01 - I68 100 -01 54 - 219 100

08 -06 -08 - 0 73 0 97 07 -04 -06 14 - 0 82 I I4

-12 17 I4 -27 -15 - 72 05 8 I9 -03 IO I2 -37 -32 53 - 72 85 7 74

00 41 27 -04 -01 09 02 - 0 75 049

‘The education vanable IS coded as follows 1 = less than 7 years, 2 = 7-9 years, 3 = 10-11 years, 4 = 12 years, 5 = 13-15 years, 6 = college graduate and 7 = graduate on professional tranung

Rehgon and psychologrcal dtstress m a commumty sample 1259

mvolvement, stress and psychological distress OLS regression is fully appropnate for our continuous dependent vanable The correlation matnx from whtch the regression models were estimated is pre- sented m Table 1 Pauwtse present correlations were used m all regression analyses The analyses pro- ceeded m a senes of steps m which we estimated the effects of rehgrous mvolvement on psychological distress This relationship was then adlusted for potentially confounding sociodemograpmc factors The soctodemographrc vanables utrhzed are age (m years), education (nommally scaled vanable coded from 1 = less than 7 years of education to 7 = graduate or professtonal training), gender (1 = male), manta1 status (1 = marned, 0 otherwtse), and race (1 = black, 0 otherwtse) Subsequent regression models assessed the association between stress and psychological distress and the extent to which rehgtous mvolvement may buffer the effects of stress on health

A final step m all analyses mvolved entenng the Time 1 Gunn score as a predictor of Time 2 Gunn The use of the Ttme 1 distress measure effectively converts the Time 2 outcome mto change scores, Thts 1s appropnate m these analyses because tt allows us to determine the extent to whtch any improved mental health functiontng found among those high on rehgtous mvolvement ts stgmficantly greater than any improvement found among those having lower scores on the rehgion measures

RESULTS

Relrglon and psychologrcaI dzstress

Table 2 presents the results of three regression analyses that assess the assoctation between psycho- logical distress and religion In the first model, Ttme 2 (1969) Gunn scores are regressed on the Time 1 (1967) religious attendance and affihatton In the second regression model, controls are introduced for soctodemographtc factors (age, education, manta1 status, gender and race) that were measured at Time 1 The final model adds the Time 1 (1967) Gunn score as a predictor of the Time 2 Gunn score

The first model m Table 2 indicates that although rehgtous affiliation IS unrelated to psychological drs- tress, rehgrous attendance IS positively associated with the Time 2 (1969) Gunn score Persons who attend rehgtous services regularly report lower levels of psychologtcal distress than infrequent attenders and non-attenders This relattonshtp remains robust when adlusted for the soctodemographtc variables but tt IS reduced to non-stgmficance when controlled for Time 1 (1967) psychologtcal distress Religious attendance at Time 1 is not associated with increases m psychologtcal well-being, as measured by the Gurm scales. Thus, m the face of ngorous statistical con- trols for the possible confounding of public rehgtous partictpatton with scores on the Gurm scale, we find that attendance IS unrelated to psychological d:stress Our prospecttve analyses have failed to replicate the inverse assoctattons between religious commitment and psychologtcal distress that were reported for the cross-secttonal analyses at Time 1 [ 131

We tested for nonhneanty m the assoctatton be- tween rehgtous attendance and mental health status

Table 2 Analyses of the assoclatlon between Tune 2 (1969) Gunn scores and the rehgton measures at

Tlmc 1 (1967)

i II III

Independent vanables (SE) (A) (&

Attendance 0 83’. 0 84.’ 0 16 (0 32) (0 32) (0 28)

Affihatlon -046 -0 88 -0 88 (0 75) (0 74) (0 64)

Age 003 (0 02) (%5

Education 0 94.1 0 48’. (0 22) (0 19)

Mamed 0 87 0 II (0 69) (0 60)

Sex (men= 1) I 60** 0 54 (0 60) (0 52)

Race (black = 1) 0 63 I 30 (0 97) (0 84)

Gunn 1967 0 49” (0 03)

Constant 71 80 66 14 35 81 R2 0010 0 050 0290

** = P < 0 01, 2-taded tests b = unstandardized regrewon coefficients

Shaver et al [28] reported a curvihnear relatton- ship between religion measures and psychologtcal symptoms The very rehgtous and the non-rehgtous enloyed the best reported health Accordingly, to a regression equation that included the demographic vanables and Ttme 1 (1967) rehgious attendance, we added the squared coefficient for religious attend- ance (quadratic term) A srgmficant quadratic term would indicate that the associatton between rehgtous attendance and distress ts curvtltnear The quadrattc term was not stgmficant (analysts not shown), mdicat- mg the absence of curvihneanty m the assoctatton between religion and psychological distress

We also explored the extent to which vanattons exist by race, gender and educattonal level m the assoctatton between the rehgton measures and psychological distress Specifically, for each of these soctodemographtc vanables, we regressed Time 2 (1969) Gunn scores on the two religion vanables, all of the soclodemographlc vanables, and the relevant muthphcatlve term for the interaction between each rehgion measure and the soclodemographlc correlate under consideration In these analyses (not shown), none of the interaction tests were slgmficant

In cross-sectional studies researchers frequently assume that the reported level of rehglous mvolve- ment IS a stable characteristic of the respondent In contrast, religious behavior may be a fairly transient phenomenon Lmdenthal et al [ 131, for example, noted that when faced with stress, respondents reported a decline m rehgtous attendance The fact that we are workmg with panel data allows us to explore the nature of changes m rehgtous attendance between 1967 and 1969 and the consequences that these changes could have for mental health status Ftrst, we noted that attendance levels were relattvely stable over the course of 2 years Table 1 reveals that the correlation between rehgtous attendance at Time 1 (1967) and Time 2 (1969) was 0 54

Second, we divided our sample mto subgroups based on the combmatlon of the level of rehglous attendance reported at Ttme 1 (1967) and Time 2

1260 DAVID R WILLIAMS et al

(1969) At each tzme point, all respondents were classified into one of three categories. high attenders (persons who attended rehgzous services once a week or more), moderate attenders (zndzvzduals who attended once a month to two or three times a month) and low attenders (those who never attended as well as those who attended a few times a year or less) Respondents were then assigned to one of five categones based on their 1967 and 1969 attend- ance. The stub/y hzgh group (n = 216) consists of persons who were hzgh attenders at both time points The newly hzgh (n = 70) are hzgh attenders m 1969 who were either moderate or low attenders m 1967 The declznzng attendance group (n = 99) 1s com- pnsed of hzgh attenders at Tzme I who were moderate or low attenders at Time 2 The moderate group (n = 152) consists of persons who were moderate attenders at both time pomts, as well as those who fluctuated from the moderate to low level or vice versa between the two data collectzon points Finally, the stably low (n = 149) were low attenders at both time points

Table 3 presents the results of analyses that exam- med the relatzonshzp between attendance patterns and psychologzcal distress. We anticipated that those who reported conszstently high levels of attendance and those who increased then attendance would have lower levels of psycholo@cal distress than persons with consistently low attendance levels The first model m Table 3 indicates that the stably high, the newly hzgh and the declzmng attendance group all had szgmficantly higher scores on the Gurm scale (that zs, less psychologzcal distress) than the stably low attendance group Thus, a high level of rehgzous attendance m 1967 or m 1969, irrespective of their attendance level at the other tzme point, zs predictive of psychologzcal well-being However, szmzlar to the findings m Table 2, these assoczatzons do not remam szgnzficant when adJusted for Tzme 1 (1967) distress scores

Relzgzon, stress and mental health

We have noted that rehgzon does not directly enhance the psychologzcal well-being of zts adherents

Table 3 Analyses of the assOclauon between Time 2 (1969) Gunn scores and attendance at Time I

(1967) combmed wth Time 2 (1969)’

I II Independent vanables (Si (A)

Rehgmus attendance a Stably high 1 63. 0 23

(0 73) (0 67) b Newly high 2 06’ 0 58

(0 99) (0 90) c Dechnmg I 90. 041

(0 88) (0 ‘30) d Moderate 0 68 001

(0 79) (0 71) e Stably low (omnted)

Time I Gunn 0 40” (0 03)

Constant RZ

l P < 0 05. l *P < 0 01, 2-taded tests ‘Both models mclude controls for age, education,

manta1 status, gender and race b = Unstandardized regressIon coefficients

We now turn to examme the buffenng hypothesis Can rehgzon protect mdzvzduals from at least some of the negative consequences of stress’ Table 4 presents four models that explored the assoczatzons among rehgzon, stress and psychologzcal distress The use of the Time 1 measures of religion m these analyses excludes the posszbzhty that any modifying effects that we observe are due to changes m rehgzous mvolvement resulting from stress The first model shows the assoczatzon of the two stress measures and the two rehgzon measures to the Time 2 (1969) Gunn scale, controlhng for the soczodemographzc vanables The second model adds adJustment for the Time 1 (1967) Gunn score, and models three and four tests for mteractzons between relzgzous attendance and hfe events, and attendance and health problems, respectively

Table 4 shows that both hfe events and health problems are szgnzficantly inversely associated with scores on the Gunn scale As expected, stress IS posrtzvely related to psychologzcal distress Model II indicates that the coefficients for stress are reduced

Table 4 Analyses of the assoclatlon between Time 2 (1969) Gunn scores. Time I (I 967) measures of rehgon and mdlcators of stress’

Independent vanables

I b

(SE)

II

(Sk

111 b

(SE) Attendance

Afiihatlon

Life events (LE)

Health problems (HP)

Gurm 1967

0 62’ (0 28)

-084 (0 66)

- I 77’ (0 23)

-2 69’. (0 28)

Attendance x LE

0 II (0 26)

-083 (0 59)

-1 53” (0 21)

-I 85.. (0 26) 0 39’9

(0 03)

-023 (0 30)

-087 (0 59)

-2 26” (0 40)

-1 88. (0 26) 0 39’.

(0 03) 0 43’

-0 32 (0 32)

-065 (0 60)

-I 55” (0 21)

- 2 70” (0 44) 0 40.’

(0 03)

(021) Attendance x HP 0 52.

(0 23) Constant 70 11 44 53 45 4 453 R’ 0 251 0 399 0 403 0404

l P < 0 05, l *P < 0 01, 2-taded tests ‘All models m&de controls for age, education, manta1 status, gender and race b = unstandardued regresston coefficients

Rehgon and psychologxal distress m a commumty sample 1261

but remam slgnrficant when controlled for Time 1 (1967) psychologtcal distress Model II also reveals that the relationship between attendance and distress IS reduced to non-sigmficance when controlled for Tl distress Models three and four reveal that both of the multtphcative terms for interactions between stress and rehgious attendance are significant The interaction terms capture operant religious effects that would go unnoted otherwise Moreover, the sign is positive for both mteraction coefficients This pattern of results reflects classic buffenng effects That is, at low levels of religious attendance, stress IS associated with increased levels of psychological distress However, as the level of religious attendance increases, the adverse consequences of stress are reduced Surular analyses for the associatton between religious affiliation and the stress measures were not significant

In sum, consistent with other research [7], we find that our measure of rehgious behavior (religious attendance) is more consequential for health status than our mdicator of rehgious affiliation The affiha- tion measure is unrelated to psychological distress In contrast, although rehgious attendance does not directly reduce psychological distress, it does buffer the impact of stressful life events and physical health complamts on psychological well-being

DISCUSSION

The findings reported here underscore the import- ance of giving more systematic research attention to the consequences of rehgious beliefs and behavior for health and well-being. National surveys reveal the contmumg importance of public and pnvate rehgtous mvolement m contemporary Amencan hfe [29] Our results indicate that rehgion may be a potent coping strategy that facihtates adjustment to the stress of hfe Further exploration of this issue merits serious and sustained research attention

One compellmg reason to replicate the analyses reported here is the possibihty that they may reflect period or cohort effects The data utilized m this study are over 20 years old It is possible that the findings documented here are true only for that earlier time period and would not apply today In a comprehensive review of the literature on rehgious involvement and sublective well-bemg, Wetter et al. [30] found a stronger relationship between religion and SubJective well-being m earlier studies than m more recent ones

Our use of longitudmal survey data is clearly an improvement over merely studying cross-sectional associations but analyses of two wave panel data are not without serious hnntanons [31] For example, the inclusion of Time 1 health status adjusts for baseline differences among respondents m the levels of health However, if health status at Time 2 IS also affected by other unmeasured causes, the Time 1 health status indicator IS an inadequate proxy for the mynad factors that are not mcluded m the prediction equation The presence of measurement error is another serious hmitation Errors of measure- ment can create spurious covanance among the variables in the regression models Theoretically- grounded research that utilizes multiple indicators

of religion and that employs structural equation modeling procedures [32] can begm to address these lirmtations

This paper also illustrates some of the cnucal shortcommgs m current research on religion and mental health Rehgious attendance and religious affiliation are the only measures of religious commit- ment that we utihzed These are two of the most commonly used measures m research on religion [l] In contrast, religious mvolvement is a complex multi- dimensional phenomenon [33-351 Kmg and Hunt [33], for example, have tdentified more than a dozen different ways of being rehgious, and have developed and tested scales to measure each component &ml- larly, Levm and associates [2.6,20,36] have pro- posed numerous theoretically mformed mechamsms by which religion can affect health status that clearly constttute the most fruitful extant starting ground for empirical mvestigations of the effects of religion on health. The advancement of our understandmg of the nature of the association between religion and health, is contingent on efforts to comprehensively assess religion, and identify the cnttcal dimensions of reh- gious commitment that are linked to health status

Research efforts of thts kmd are necessary to understand even the results presented here We reported that religious attendance buffers or moder- ates the relationship between stress and health How- ever, we are unable to tell if this effect 1s hnked to anything mtnnsmally rehgtous Although we employ controls for formal education m all of the analyses, it IS still possible that the attendance measure is a proxy for some aspect of social status Sociologists have long noted that religious partici- pation IS frequently a badge of socioeconomic status, secular m character, and of no greater rehgious sigmficance than participation m other community orgamzations [37] And there is abundant evidence that participation m formal and informal social groups, rehgious and non-rehgious, can promote health, reduce stress and buffer the effects of stress on health [22] Moreover, besides social class, rehgious attendance may be confounded with functtonal health [36]

It follows that a simple measure of the frequency of religious attendance does not adequately cap- ture public religious participation A comprehensive assessment of public rehgious mvolvement must include attendance at rehgious meetings other than the main weekly worship service, financial support of religious organizations, and holdmg leadership and volunteer positions m rehgious groups [35]. Researchers must then seek to identify how these public aspects of rehgious mvolvement relate to pn- vate dimensions of religious beliefs and behavior and how they combme to affect levels of health and well-being.

Acknowledgements-An earlier version of this paper was presented at the Ntnety-Seventh Annual Meettngs of the Amertcan Psychologtcal Assoctatton, New Orleans, August, 1989 We wish to thank Jerome K Myers for permIssIon to use the data and the anonymous renewers for very helpful comments on an earlier version of this paper. The research was supported, m part, by grant Rl l-8812285, from the National Science Foundation

1262 DAVID R WILLIAMS er al

1

2

3

4

5

6

7

8

9

10

11

12

13

14

15

16

17

18

19

20

21

22

REFERENCES

Larson D , Pattlson M , Blazer D , Omran A and Kaplan B Research on rehgous vanables m four major psychlatnc Journals Am J Psychlat 143, 329-334, 1986 Levm J S and Vanderpool H Y Is frequent rehgous attendance really conducive to better health? toward an epidemiology of rehgon Sot Scr Med 24, 589-600, 1987 Jarvls G K and Northcott H C Rehglon and dlffer- ences m morbidity and mortality Sot- .%I Med 25, 813-824, 1987 Schlller P L and Levm J S Is ther a reh@ous factor m health care utlhzatlon7 a review Sot Scl Med 27, 1369-1379, 1988 Bergm A E Rehgloslty and mental health a crmcal reevaluation and meta-analysis Prof Psycho1 14, 170-184, 1983 Levm J S and Schlller P L Is there a rehgous factor m health? J Rehg Hlth 26, 9-36, 1987 Gartner J , Larson D B , Allen G D and Gartner A F Religious commitment and psychopathology a review of the empmcal literature J Psycho1 Theo1 In press Gunn G , Veroff J and Feld S Amerwans View thezr Mental Health Basic Books, New York, 1960 Veroff J , Douvan E and Kulka R The Inner Amerzcan Basic Books, New York, 1981 Idler E L Reh@ous mvolvement and the health of the elderly some hypotheses and an mlhal test Sot Forces 66, 226-238, 1987 Petersen L R and Roy A Rehglowty, anxiety and meaning and purpose rehgon’s consequences for psychologlcal well-bemg Rev Rehg Res 27, 49-63, 1985 Hannay D R Rehglon and health Sot Scl Med 14A, 683-685. 1980 Lmdenthal J J , Myers J K , Pepper M and Stern M S Mental statuss and rehglous behavior J Scl Study Rehg 9, 143-149, 1970 Brown D R and Gary L E Stressful hfe events, social support networks, and the physical and mental health of urban black adults J Hum Stress 13, 165-174 Carr L G and Hauser W J Class, reh@ous par- ticlpatlon and psychlatnc symptomology Znt J Sot Psychtat 27, 133-142 Clearly P D and Houts P S The psychological impact of the three mile island mcldent J Hum Stress 10, 28-34, 1984 Markides K S , Levm J S and Ray L A Rehglon, agmg, and life satlsfactlon an eight-year, three-wave lonmtudmal studv The Geront 27. 660-665. 1987 Kaplan B and Blazer D G Rehg&n m the’stress and adaptation paradigm In Soctophyslology of Social Re- lattonshlps (Edited by Barchas P ) In press Martin J E and Carlson C R Spmtual dimensions of health psychology In Behavtor Therapy and Relrgron (EdIted bv Miller W R and Martm J E ). DD 57-l 10 Sage, Palb Alto, CA, 1988

I . .

Levm J S Religious factors m aging, adlustment, and health a theoretlcal overview J Rehg Agrng 4, 133-146, 1987 Elliott G R and Elsdorfer C Stress and Human Health Spnnger, New York, 1982 Wllhams D R and House J S Stress, social sup- port, control and copmg a social epldemlologlcal view In An Introduction to Health Promotron Research (EdIted by KIckbusch I and Badura B ) World Health Organization In press

23

24

25

26

27

28

29

30

31

32

33

34

35

36

37

Krause N and Van Tran T. Stress and rehgous mvolve- ment among older blacks J Geront 44, 54-513, 1987 Myers J K, Lmdenthal J J , Pepper M P and Oitrander D R Life events and men&l status a long- tudmal studv J Hlth Sot Behav 13. 398-406. 1972 Myers J K, Lmdenthal J J and Pepper M ‘P Life events, social integration and psychlatnc symptomatol- ogy J Hlth Sot Behav 16, 421-427, 1975 Kessler R C and Cleary P D Social class and psycho- logcal distress Am Sot Rev 45, 463478, 1980 SAS Institute SAS User’s Gwde Statrstlcs SAS mstl- tute, Gary, NC, 1985 Shaver P , Lenauer M and Sadd S Religiousness, conversion, and subjective well-being Am J Psychrar 137, 1563-1568, 1980 Gallup Report Relrglon m America-50 Years Pnnce- ton reh@ous Research Center Prmceton, NJ Witter R A, Stock W A, Okun M A and Hanng M J Rehglon and SubJective well-being m adulthood a quantitative synthesis Rev Relrg Res 26, 332-342. 1985 Kessler R C , Greenberg D F Lmear Panel Analysis Models of Quantltatlve Change Academic Press, New York, 1981 Joreskog K G and Sorbom D LISREL Esttmarron of Linear Structural Equation Systems by Maximum Llkebhood Methods International Educational Services, Chicago, 1976 Kmg M B and Hunt R A Measunng the rehgous vanable national replication J Scz Study Rehg 14, 13-22, 1975 Mmdel C H and Vaughn C E A multldlmenslonal approach to reltaosity as disengagement J Geronr 33, 103-108, 1978 Amlay S C and Smith D R Agng and rehgous partlclpatlon J Geronf 39, 357-363, 1984 Levm J S and Vanderpool H Y Is rehglon thera- peutically significant for hypertension? Sot Scl Med 29, 69-78, 1989 Goode E Social class and church participation Am J Sot 72, 102-111, 1966

APPENDIX

Life events

Measures of Stress

The 28 undesirable events are (1) failed school or trammg program, (2) problems m school, (3) moved to a worse nelghborhood, (4) wldowed, (5) divorced, (6) separated, (7) trouble with m-laws, (8) serious physical illness, (9) serious injury or accident, (10) death of a loved one, (11) stdlblrth, (12) frequent minor illness, (13) mental illness, (14) death of a pet, (15) demoted or changed to a less responsible Job, (16) laid off temporarily, (17) busmess faded, (18) trouble with boss, (19) out of work for over a month, (20) fired, (21) financial status a lot worse than usual, (22) foreclosure of mortgage or loan, (23) appearance m court, (24) deten- tion m Jail, (25) arrested, (26) law suit or legal action, (27) loss of dnver’s license, and (28) change m relations with neighbor, friend and relative such as serious or maJor disagreement

Health problems

The 16 health problems are (I) eye trouble, (2) ear trouble, (3) sinus trouble, (4) throat trouble, (5) bronchitis, (6) pneumoma, (7) tuberculosis, (8) bolls and abscesses, (9) diabetes, (10) kidney trouble, (11) bodily injury, (12-14) operations, (15) cancer or tumors, and (16) tooth trouble, excluding routme prophylaxis