The Spiritual History Scale in Four Dimensions ( SHS-4 ) : Validity and Reliability

The goals of this study were to develop a valid, reliable measure of lifetime religious and spiritual experience and to assess its value in explaining late-life health. Procedures included semi-structured interviews with Duke Aging Center volunteers ( n 5 30), followed by structured interviews of a stratified, random sample of subjects ( n 5 157) from the Established Populations for Epidemiologic Studies of the Elderly at Duke University. Principal components analysis suggested four factors with favorable psychometrics. Health-impaired subjects reported a history of seeking/receiving divine aid (God Helped). At every level of impairment, Lifetime Religious Social Support and current religious attendance were positively correlated. Regardless of current attendance, subjects who reported higher Lifetime Religious Social Support received more instrumental social support. Healthy behaviors were associated with both God Helped and Lifetime Religious Social Support. Cost of Religiousness predicted depressive symptoms and impaired social support. Family History of Religiousness was unrelated to late-life health. Evaluation of the Spiritual History Scale in Four Dimensions (SHS-4) across geographical settings, cultural subgroups, age cohorts, and clinical samples is warranted.

The evidence that religion promotes health or improved coping with health problems in late life continues to grow (Ellison & Levin, 1998;Koenig, Larson, & McCullough, 2000).Although these reports are interesting, the religion and spirituality data upon which they have been based generally lack a historical perspective.Little attention has been paid to religious experience over a lifetime when estimating the attendant risk or protective effects.This study was funded by the Fetzer Foundation (Kalamazoo, MI) to examine the dynamic and multidimensional patterns of religious and spiritual practices and beliefs over the life course and to develop quantitative instrumentation for measuring religious/spiritual life histories in clinical and community samples of elders.

Measuring the Past
When describing the causal web of health outcomes, epidemiologists of religion have focused primarily on estimating effects of current practice and belief (Braam et al., 1998;Ellison, 1995;Gardner, Sangorn, & Slattery, 1995;Idler & Kasl, 1997;Koenig et al., 1997;Krause, 1993;Pargament et al., 1992;Strawbridge, Shema, Cohen, Roberts, & Kaplan, 1998;Williams, Larson, Buckler, Heckmann, & Pyle, 1991) or on changes during late life (Krause, 1998;Markides, Levin, & Ray, 1987).This strategy is consonant with common practice among epidemiologists, who often fail to exploit the multiple dimensions of historical risk for the information they might yield.Typical epidemiological studies estimate exposures as present (vs absent) or ever (vs never).Less frequently, they describe duration or accumulation of exposures over time, for instance, pack-years of smoking (Howard et al., 1998), lifetime frequency of severe sunburn (Elwood & Jopson, 1997), and cumulative days of combat conditions (O'Toole et al., 1996).Social scientists have developed two related strategies to specify long-term risk in stress-crisis research and studies of life course.In the former, health outcomes are observed to be sequelae of distal negative life events, including childhood loss events (Brown, Harris, & Bifulco, 1986;Landerman, George, & Blazer, 1991) and early-life financial decisions (O'Rand & Landerman, 1984).Life course investigators have grouped events and turning points into life course trajectories.The resulting shapes, representing stability and change over the life span, are organized into typologies of cumulative risk, and the differential impact of each type on health outcomes is estimated (Clipp, Pavalko, & Elder, 1992;Elder, 1997).

The Gerontologist Measuring Past Religiousness
Childhood or lifetime religiousness is rarely assessed empirically, and in those studies where it is, its relationship to late-life religiousness, rather than its relationship to late-life health, is the primary focus (Blazer & Palmore, 1976;Zinnbauer et al., 1997).An exception was Hunsberger (1985), who assessed two religious history domains with single items (i.e., parental emphasis on religion during childhood and the agreement between current beliefs and those taught during childhood).The former, but not the latter, demonstrated a weak positive association with global self-rated health.
More usually, one or two items that reference early-life religiousness are combined into a single scale with items referencing more recent religious activities or beliefs, making it difficult to disaggregate early from late effects.For example, two of three items in the Subjective Religiousness Scale of Chatters, Levin, and Taylor (1992) tapped current religious involvement, and the third referred to religious exposure in childhood (i.e., "the importance of religion in your home growing up").The scale was not used to assess the effects of religion on health.In some studies, aggregated scales include items assessing "any" past history, rather than referencing a particular age or stage, for example, "(Has) organized religion...ever hindered or harmed your spiritual well-being more than it had helped?" from Moberg's (1984) Religious Cynicism scale.Neither was the purpose of that study to assess religion's effects on health.Nevertheless, the inclusion of an item with negative valence was noteworthy, because the costly effects of past religious, spiritual, or existential experiences have received less attention in published reports than have neutral or health-promoting effects.
When assessing past religious events, epidemiologists have devoted their primary efforts to describing the distribution of religious conversion experiences in various populations; for example, 50% of White Protestants in a national survey (Dixon, Lowery, & Jones, 1992) and 73% in the North Carolina Piedmont study (Hays et al., 1998) reported such experiences.The InSpirit Religious Experiences Scale (Kass, Friedman, Leserman, Zuttermeister, & Benson, 1991) is a more broadly conceived index of past religious experiences, and scores have been significantly correlated with declines in symptoms of stress.Systematically described religious trajectories are rare.In one exception, Hunsberger (1985) rated the "religiousness" of elders retrospectively by decade across the life span and compared them to trajectories of "happiness," finding little correspondence between the two.
The paucity of measures tapping historical exposure to religion and spirituality limits the conclusions that can be drawn regarding their overall effect on late-life health.On the one hand, late-life health may be primarily affected by late-life religion and spirituality, including its various dimensions and under specific circumstances.If so, investigators could concentrate on measuring more proximal religious exposures and avoid the problems inherent in attempts to mea-sure long-past behaviors, events, and attitudes.On the other hand, late-life health may be additionally affected by religious and spiritual experiences of childhood and early and middle adulthood.If so, investigators would be empowered by the availability of valid measures to tap multiple dimensions of more distal religiousness, thus allowing enhanced specificity for estimating the effects of previous exposures.
We have limited the discussion above to quantitative measures of lifetime exposure to religiousness.Such a discussion omits an important body of nonquantitative work in the tradition of studying religious development.This rich literature is exemplified in Fowler's (1981) Stages of Faith , Coles' (1991) The Spiritual Life of Children , and Wuthnow's (1999) Growing Up Religious.We used these works, with Benson's (1996) studies of loss and growth of faith over the life span and Mandelbaum's (1973) studies of "turnings," to ensure a comprehensive set of probes in the early construct development phase of the current study (see Methods).In the instrument development phase, however, the authors were guided primarily by the material generated by those probes and were not constrained by categories described in previously published work on religious development.
The first aim of this study was to develop and describe a valid and reliable instrument to measure lifetime experience of religion and spirituality for use in samples of elderly people.The second aim was to assess the degree to which the measure added explanatory value to the association between health status and late-life religious practices.

Methods
Construct Development (Phase 1) Design and Sample.-Thestudy was designed in two phases, construct development and instrument development.The design of Phase 1 (construct development) was a case series of subjects ( n ϭ 30) from the Duke University Aging Center Registry, a listing of more than 1,900 community-dwelling elders in the Durham/Chapel Hill/Raleigh area who had expressed a willingness to participate in Duke Aging Center research.Subjects varied by race (33.3% African American & 66.7% White), gender (60% female), age (65-88 years), and religious tradition (13.3% Jewish, 10% Roman Catholic, 76.7% Protestant).
Measures and Analysis.-Basedon an extensive literature review, the authors developed a semistructured interview instrument to assess seven domains: demographic characteristics; religious affiliation across the life span; earliest religious or spiritual memory; characteristics of salient periods of religious or spiritual significance; frequency of 10 religious practices and overall importance of religion or spirituality across the life span; primary positive and negative influences on lifetime religiousness and spirituality; and current beliefs about creation, the human prob- lem and its solution, eschatology, ethics, communality, and the relationship between early spiritual history and late life health.Three of the authors (JH, KM, PB) interviewed subjects face-to-face, generating taped recordings lasting 1.5-3.0hours.Transcripts were analyzed using constant comparative analysis (Qualitative Solutions & Research, 1997;Reinharz & Rowles, 1988)  The sampling frame for the current study included Duke/EPESE survivors who scored 0-1 (errors) on the Short Portable Mental Status Questionnaire (SPMSQ; Pfeiffer, 1975), as measured at their most proximal (1996) interview.These were randomly sampled by religious affiliation, as reported at their baseline (1986) interview.The effective sample ( n ϭ 228) included 1:7 fundamentalist and pietistic Protestants, 1:2 mainline Protestants, and all affiliated and unaffiliated others, including Roman Catholics, Jews, Mormons, Unitarian Universalists, Jehovah's Witnesses, agnostics, and atheists.Of those sampled, 15 were deceased and 6 had moved beyond an expanded 11-county tracking area since 1996.Of the remainder, 75.8% ( n ϭ 157) consented to face-toface interviews in their home by a trained interviewer.A subgroup of these ( n ϭ 20) consented to a reliability study in which the face-to-face interview was repeated approximately 2 weeks later.
We compared subjects and nonrespondents, using Duke/EPESE data.Subjects were not demographically different from nonrespondents by age (median ϭ 80 years), gender (66% female), race (31% African American), marital status (54% married), urban-rural dwelling (61% urban), or income (33% Ͻ $10,000, 35% Ͼ $20,000), but were slightly more likely to have reported some college education.The two groups did not differ with respect to late-life religious practices reported in 1986 or 1996.Compared to nonrespondents, subjects reported similar self-rated health, impairment of basic and instrumental activities of daily living, and one-year history of hospitalization, but slightly fewer physical performance and gross mobility limitations.
Measures.-Assessmentincluded retrospective repeated measures of religious service attendance, religious media use, and devotional activity for each of the first five decades of life, as well as currently (cross-sectional), using religion items from the Duke/ EPESE (Hays et al., 1998).Repeated measures of "the importance of religion" (not available in the Duke/ EPESE) were also assessed for each decade across the life span.Seventy items, many of which were used verbatim from Phase 1 transcripts, measured degree of agreement (or disagreement) with emergent themes.Seven items asked subjects to choose-from a limited selection-the description that best summarized selected domains of their spiritual life history (e.g., continuity of religious affiliation and presence or absence of regrets about their religious or spiritual history).Subjects also reported on 15 parameters of (at most 3) religious or spiritual turning points (e.g., their permanence and scope); four parameters of their earliest religious or spiritual memory (e.g., when and where it occurred); and the most positive and negative personal influences on their spiritual life.In addition to the religion items, selected measures from the Duke/ EPESE were administered, including negative life events in the past year and incident chronic illnesses since the last interview (1996).
Analysis Strategy.-Analysesincluded development of two instruments: (1) a 23-item Spiritual History Scale in four dimensions (SHS-4), based on principal components analysis (PCA) of the agree-disagree items, and (2) typologies of religious/spiritual life trajectories for the four domains of repeated measures.The former is presented here.Following Johnson and Wichern (1992), we compared factor loadings of the 70 items with respect to quantitative criteria and construct coherence, using PCA and maximum likelihood procedures with varimax rotations.Selected items loaded Ͼ 0.40 on meaningful factors across both procedures and were not correlated more than 0.70 with any other item.Three, four, and five factors were tested.The final scale (Appendix) included four conceptually meaningful factors with eigenvalues Ͼ 1.0.We present factor loadings and test-retest reliability for the 23 items in the scale, as well as factor distributions and internal consistency in the total sample and in race and gender subgroups.
Next, we describe bivariate associations between factor scores and other demographic, health status, and organizational and nonorganizational religious measures from 1986 and 1996 Duke/EPESE and from the current study.Because of the skewed distribution of the four SHS factors, all bivariate tests were nonparametric: Spearman's correlation coefficient for continuous variables and Wilcoxon ranksum for differences between means.
As an operational measure of late-life religiousness, we used frequency of church attendance in 1996 for three reasons.First, church attendance has been the most robust religious or spiritual domain to date to correlate with the greatest variety of health outcomes.Second, the health status variables most recently measured in our sample were from 1996.Third, the 1999 church attendance score from the current study could potentially be contaminated by concurrent measures of history of religion and spirituality, the possibility of which was minimized by using a measure of attendance from three years prior.The correlation between church attendance measures from 1996 and from the current study was high ( r ϭ .79,p Ͻ .0001).Associations between health and history of religion and spirituality were adjusted for demographic confounders and weighted to account for the 1986-1996 sequential stratified sampling procedures of the Duke/EPESE; analyses were not reweighted for stratified sampling by affiliation, used in the current study.

Spiritual History Scale Psychometrics
Table 1 presents the factor loadings from the principal components analysis of the 23-item SHS-4 scale.God Helped is an index of past help-seeking (e.g., prayed, trusted, looked for) and instrumental support received (e.g., blessings, guidance, answers, instruction, longevity) from the divine and/or from religious practices.Family History of Religiousness is an index of perceived childhood religious attachments and the religious legacy from the family of origin.Lifetime Religious Social Support is an index of adult religious participation, including indices of religious involvement in midlife and religious hospitality.Cost of Religiousness measures the presence or absence of physical, emotional, and interpersonal losses and difficulties associated with one's past religious life.Test-retest reliability estimates were favorable, with weighted kappas for 20 of the 23 items Ն 0.90 (range: 0.74-1.00).
Table 2 presents the distribution of the four factors in the total sample and in race and gender subgroups.In this population, all groups scored toward the top of the range on three of the four factors and low on Cost of Religiousness.For example, where the potential range of God Helped was 0-40, the mean score for all groups was Ն 34.0, and the interquartile range included the maximum value.The reliability coefficient, Cronbach's alpha, was in most cases Ն 0.70, suggesting that a high proportion of the observed variance was due to true differences among individuals in the sample.The two exceptions to this pattern were differences among African American elders on Lifetime Religious Social Support and Cost of Religiousness.

Spiritual History, Demographics, and Late Life Religiousness
Table 3 presents differences in demographic and religious subgroup scores on each of the four factors.Racial and socioeconomic differences were pronounced.African American elders scored significantly higher on God Helped, Family History of Religiousness, and Lifetime Religious Social Support than White elders.Elders with no college credit were more likely to score high on God Helped and Lifetime Religious Social Support than their more educated counterparts, and those with annual incomes Յ $20,000 were more likely to score high on God Helped than more wealthy subjects.Scores on Cost of Religiousness were higher among married than unmarried elders.Age, gender, and urban-rural differences on all factors were trivial.
Religious affiliation groups differed significantly in their mean scores on the God Helped and Lifetime Re- ligious Social Support scales.On the God Helped subscale, fundamentalist/pietistic subjects scored higher than mainline Protestants, and these higher than Roman Catholic and Jewish subjects, and these higher than other subjects.On the Lifetime Religious Social Support subscale, the fundamentalist/pietistic group reported higher mean scores than the other three groups.Mean scores across affiliation groups did not differ on Family History of Religiousness or Cost of Religiousness.Subjects who scored high on religious service attendance or private devotional activity at previous interviews were more likely to score high on God Helped and Lifetime Religious Social Support, and somewhat more likely to score high on Family History of Religiousness and Cost of Religiousness.
The four spiritual history factors were associated with other religion items measured in the current study.When subjects' lifetime religious affiliation had not changed over their lifetime, when they reported that God was involved in their daily lives, and when they engaged regularly in religious practices, subjects tended to report higher scores on God Helped, Family History of Religiousness, and Lifetime Religious Social Support.Cost of Religiousness scores were highest for subjects who reported that someone had influenced their religious life in a nega-tive way and for those reporting at least one religious or spiritual turning point.

Spiritual History and Health Status
Table 4 presents relationships between 1996 health status variables and 1996 religious service attendance, first unadjusted and then with the addition of each of the four SHS factors.Health status measures included chronic medical conditions; basic and instrumental ADLs; physical performance; gross mobility; self-rated health; depressive symptoms; use of alcohol and tobacco; instrumental aid received and given; and availability of a confidant.In general, the unadjusted tests were significant and positive for physical health and social support, and significant and negative for health behaviors.In other words, subjects who attended church or synagogue were in better health, were less isolated, and drank and smoked somewhat less than non-attendees.Unadjusted relationships of religious service attendance with self-rated health, depressive symptoms, smoking, and instrumental help received did not reach statistical significance.
Net of the often-reported association of service attendance and physical health, God Helped was posi- tively associated with both impaired gross mobility and poor self-rated health.The interpretation of this result is that health deficits are associated with less frequent organizational religion and with a past his-tory of seeking/receiving divine aid.Impaired gross mobility was also positively related with Lifetime Religious Social Support, when controlling for service attendance.In this case, the interaction term between the two variables was significant (p ϭ .0004).At every level of mobility impairment from least (0) to most (3), subjects with the highest Lifetime Religious Social Support scores attended services significantly more often than did subjects with the lowest Lifetime Religious Social Support scores (untabled).The same pattern of interactive effect was evident for impaired instrumental ADLs.The single measure of mental health (depressive symptoms) was negatively but nonsignificantly associated with church attendance.However, Cost of Religiousness explained significant variance in depressive symptoms.Net of church attendance, subjects who endorsed a clinically relevant number of symptoms (Blazer, Burchett, Service, & George, 1991), as well as those in the subthreshold range, reported significantly higher Cost of Religiousness scores than subjects with few or no depressive symptoms (untabled).
Healthy behaviors were associated with both God Helped and Lifetime Religious Social Support, over and above the protective effects of service attendance.In comparison to subjects with the least lifetime exposure to religious social support, odds of alcohol use were significantly lower for subjects with moderate (odds ratio [OR] ϭ .22,confidence levels [CL] ϭ .10,.47)and high exposure (OR ϭ .05,CL ϭ .22,.01).Odds of smoking were similarly low for these comparison groups, respectively (OR ϭ .13,CL ϭ .04,.38;OR ϭ .02,CL ϭ .04,.01).
Subjective social support, measured as the availability of a confidant, was negatively associated with Cost of Religiousness.Regardless of church attendance, subjects with the lowest subjective social support scores had mean Cost of Religiousness scores more than double those of elders with higher subjective social support scores (means: 4.1 vs. 1.6; untabled).Giving instrumental support was strongly associated with late-life service attendance but was unrelated to any dimension of lifetime religious experience.On the other hand, receiving instrumental support was unrelated to late-life service attendance, but was strongly associated with God Helped and History of Religious Social Support.Regardless of current attendance, subjects with more history of religious social support received significantly more instrumental support than did subjects without such history.

Spiritual History and Cumulative Lifetime Religiousness
As an alternative check on the construct validity and efficiency of Lifetime Religious Social Support and God Helped subscales, we compared them to measures of cumulative lifetime worship-years and devotional-years.These were constructed in a similar manner to pack-years of smoking.Using repeated measures of the frequency of religious service attendance for each decade of the life span, we summed the average attendance for each year and divided the total by 52.For example, an 85-year-old person who attended services twice a week between the ages of 5 and 15 years, never between the ages of 15 and 25 years, and once a week since then, would score 80 worship-years.Devotional-years were based on years of daily devotionals.
In this sample, mean worship-years was 67.6 (interquartile range [IQR]: 53-80); mean devotionalyears was 37.4 (IQR: 21-51).Lifetime Religious Social Support and worship-years were moderately correlated (Pearson r ϭ .64);God Helped and devotional-years were also moderately correlated (Pearson r ϭ .49).We substituted these two cumulative lifetime measures in the models described above where Lifetime Religious Social Support and God Helped, respectively, were significant predictors of late-life health.In each instance, the cumulative lifetime measure was a less robust predictor than was either Lifetime Religious Social Support or God Helped.

Discussion
The Spiritual History Scale (SHS-4) is a 23-item, four-dimensional, retrospective summary measure of religious and spiritual practices and attributions over the life course.It was developed in a stratified, random, biracial sample of cognitively intact community-dwelling elders, living in a predominantly Protestant Christian region of the United States.Three subscales (Lifetime Religious Social Support, God Helped, and Cost of Religiousness) added explanatory value to tests of religious attendance effects on physical and mental health, health behaviors, and subjective and instrumental support.Family History of Religiousness, while strongly related to consistency of affiliation over the life course, did not demonstrate significant associations to health status in a population of elders aged 77 years and older.

Reliability and Internal Validity
Traditional criteria for reliability and internal validity were generally achieved.Evaluations of testretest reliability of individual items and internal consistency of subscales were favorable.Selection bias was minimized by use of (a) a randomized sample from an established population for epidemiologic research among elders; (b) extensive data on nonrespondents; and (c) in-home interviews designed to minimize nonparticipation.Thus, the results are most likely to be valid for older-old African American and White persons who are cognitively intact and living in the urban and rural Piedmont area of North Carolina.Information bias was minimized by use of (a) structured instrumentation, based on systematic grounded theory procedures; (b) trained interviewers; and (c) chronologically separate measures of exposure (religious/spiritual history) and outcome (health status).
In spite of efforts to the contrary, misclassification may have occurred.Recall bias could have affected scores on God Helped if chronic illness and functional disability, persisting across the three intervening years, triggered differential recall of divine help- seeking and aid.Similar differential recall might also distort estimates of effect between Lifetime Religious Social Support and availability of a confidant.Such misclassification could have produced an undeterminable exaggeration or underestimation of the effects observed in this study.In general, however, the mixing of exposure and outcome (and the resulting ambiguity in the direction of causal inference between religious/spiritual history and health status) was reduced by using measures of spiritual history exposure from the current study with the prospective measures of health outcomes obtained from a prior interview wave.
Construct validity was addressed in several ways.Over the two phases of the current study, constructs relevant to a religious history (e.g., affiliations, practices, and influences) were assessed by means of multiple operations (including a narrative-based, semistructured interview that informed a structured interview) and multiple methods (including repeated measures of frequency, agree-disagree measures, and multiple-choice measures.)We found widespread evidence of convergent validity across measures within the current study.For example, God Helped scores were highest among subjects who reported regular religious practices and the involvement of the divine in their daily lives.God Helped and Lifetime Religious Social Support were moderately correlated with cumulative lifetime years of devotionals and service attendance, respectively.As well, Cost of Religiousness was most prevalent in the presence of a negative religious influence.Analyses of the correspondence between the SHS-4 subscales and lifetime trajectories of religious practice and salience are currently under way.
SHS-4 scores were also consonant with, and in some cases extended, the findings from other epidemiologic studies from North Carolina and elsewhere.As would be expected, Lifetime Religious Social Support was strongly correlated with 10-year repeated measures of church attendance, and God Helped was moderately correlated with 10-year repeated measures of private devotional activity from the Duke/ EPESE.At the same time, the previous and current measures did not appear to tap equivalent constructs: The correlation coefficients were not so high as to preclude an appropriate divergence between the repeated measures of proximal history and new scales referencing a more comprehensive "lifetime history." Convergence with non-EPESE samples was also evident.First, scores on Family History of Religiousness were dramatically elevated, similar to the frequencies of childhood churchgoing reported in a previous Piedmont North Carolina elderly cohort by Blazer and Palmore (1976).As in their study, we found early childhood and late-life religious participation remarkably uncorrelated.Second, the observed race differential in religious upbringing, lifetime religious participation, and religious help-seeking/receiving also confirms and extends findings on the religiousness of African American elders, described earlier by Krause (1992) and Chatters and colleagues (1992).Third, the relationship of Cost of Religiousness to spiritual turning points and negative influences is reminiscent of Zinnbauer and colleagues' (1997) work on the antecedents of developing a selfconcept as "spiritual but not religious."The availability of this subscale may be particularly useful for testing cohort differences in future studies.
Fourth, significant relationships between religious history and health were in a plausible direction.Given recent findings for church attendance and health behaviors (Koenig, George, Meador, Blazer, & Ford, 1994;Koenig et al., 1997;Musick, Blazer, & Hays, 2000), it was reasonable to expect that cumulative exposure to religious social support would be associated with health behaviors and might be a more significant predictor than late-life organizational religion.Likewise, it was reasonable to expect that lifetime involvement in organizational religion would explain late-life receipt of instrumental support, whereas current church attendance failed to demonstrate such a relationship.Finally, even as late-life church attendance remained inversely related to late-life functional status, as reported by Hays and coworkers (1998), the accumulated experience-or habit, as it were-of involvement in religious organizations appeared to provide motivation to persist in such activity, even in the face of significant functional impairment.These observations suggest intriguing longitudinal hypotheses about psychosocial mechanisms for future tests using these data (under way) and data from other studies.

External Validity
The criteria for external validity pose significant challenges to epidemiologists of religion.Measures have different meanings in different subgroups, and this may be especially the case with measures of religious constructs.The SHS-4 was developed among elders who were born in 1922 or earlier and grew up where the majority faith tradition was Christian.Its psychometric properties may vary when used in younger cohorts, where non-Christian faith traditions represent the majority and Christian adherents the minority, or where the distribution of Christian denominations differs.Nevertheless, the sampling strategy deliberately maximized the range of religious affiliation in a sample whose diversity in other regards (e.g., racial and urban-rural) was an asset and whose health status had been extensively and independently characterized.For elders in many regions of the United States and for the foreseeable future, the SHS-4 provides a rigorously developed instrument for measuring lifetime experience of religion and spirituality.Varying the geographical setting, the cultural subgroup, and the age cohort will be critical strategies for evaluating the broader generalizability of the measure.Equally important will be the extension of these findings to clinical groups characterized by greater levels of stress and more extensive requirements for coping.Finally, the SHS-4 could be used to address the long-standing concerns of sociologists and psy-chologists of religion to describe typical patterns of life-course religious involvement and the nature of spiritual growth and development.
neither agree nor disagree, somewhat disagree, or strongly disagree?

God Helped
God has helped me this far through my life.
(GOD_HELP) Some of the good things that have happened to me were blessings from God. (BLESSING) When I was about 60, I looked for God's guidance in my daily life.(GUIDE_60) Overall, God has answered my prayers.(FAM_PASS)

Cost of Religiousness
At times, my religious life has caused conflict between myself and other people.(CONFLICT) At times, my religious life has caused me stress.
(CAUSESTR) I have suffered physically because of my religion. (SUFFER) (ANSWER) I have trusted God to take care of me through the years.(TRUSTGOD) Through the years I have prayed for my health or for the health of others.(PRAYHLTH) Overall, my religious life has taught me to have a positive attitude.(POSITIVE) Overall, my religious life has helped me to reduce stress.(LOWERSTR) Overall, my religious life has helped me to persevere (go on when life gets hard).(PERSEVER) I have lived this long because God's time for me to die has not yet come.(LIV_TIME) Lifetime Religious Social Support For most of my life, my social life has revolved around the church (synagogue).(SOCIAL_C) For most of my life, I have known many of the people in my church (synagogue) well.(KNOW WELL) An important part of my religious life has been inviting and taking people to my church (synagogue).(INVITE) When I was 50, I was very involved in the church (synagogue).(INVOLV50) Family History of Religiousness When I was a child, I was very involved in the church (synagogue).(INVOLVCH) When I was a child, the church (synagogue) was like family to me.(FAM_CH) When I was a child, religion was a natural part of my life.(NATURAL) When I was a child, my parents left my religion up to me.(LEFT_UP) When I was a young child, I had a religious or spiritual role model.(ROLE_CH) My family passed down their religion to me.

Table 1 . Principal Components Analysis With Varimax Rotation: Factor Loadings of Spiritual History Study Items for Total Sample ( N ϭ 157)
Downloaded from https://academic.oup.com/gerontologist/article-abstract/41/2/239/648330 by guest on 12 April 2019 Note : Items loading on each factor appear in boxes.