196 Michael McCullough and Timothy Smith
although the direction of the associations still indicated that frequent attenders were
living slightly longer lives than were nonattenders. The fact that religious involvement
was related to reduced mortality by so many causes led Hummer and colleagues to pro-
pose that religious involvement might actually be one of the “fundamental causes”
of longevity: Because each of the major causes of death has its own specific etiology,
the so-called effects of religious involvement on mortality must influence mortality
through a variety of casual patterns; thus controlling any single mechanism or cause
of death should not cause the religion-mortality association to disappear.
Is the religion-mortality association a strictly American phenomenon? Perhaps not,
although the data from other places in the world are scant and preliminary. Goldbourt,
Yaari, and Medalie (1993) followed a sample of 10,059 male Israeli government workers
for twenty-three years to examine the predictors of mortality. They assessed religious
orthodoxy using a three-item measure consisting of (a) whether the respondent re-
ceived a religious or secular education; (b) whether the respondent defined himself as
“orthodox,” “traditional,” or “secular”; and (c) how frequently the respondent attended
synagogue. Unadjusted data indicated that each standard unit increase in orthodoxy
was associated with a 16 percent increase in odds of survival through the twenty-three-
year follow-up period. (These data were adjusted for age, but were not adjusted for other
demographic, biomedical, and psychosocial variables.)
Of course, not all investigations of the association of religious involvement and
mortality have revealed favorable associations (e.g., Idler and Kasl 1992; Janoff-Bulman
and Marshall 1982; LoPrinzi et al. 1994; Pargament et al. 2001; Reynolds and Nelson
1981). For example, Koenig et al. (1998) studied whether the use of religion as a source
of coping was a predictor of all-cause mortality in a sample of 1,010 older adult males
who were hospitalized for medical illness. These 1,010 patients were followed for an
average of nine years. At the beginning of their involvement in the study, patients com-
pleted a three-item measure of the extent to which they used their religion to cope with
stress. In both bivariate analyses and multivariate analyses in which the investigators
statistically adjusted for demographic, social, and medical differences among the pa-
tients, those who relied heavily on religion for coping did not live any longer than did
patients who did not rely heavily on religion for coping. Idler and Kasl (1992) reported
similar results from analyses of a sample of basically healthy, community-dwelling older
adults.
Moreover, Pargament et al. (2001) recently reported that in a sample of medically
ill adults people who believed that their illnesses were signs that God had abandoned
them or was punishing them, or who believed that the Devil was creating their illnesses,
hadshorterlives, even after controlling for a variety of demographic, physical health,
and mental health variables.
Conclusions from a Meta-Analytic Review
After conducting an extensive search, for published and unpublished studies relevant
to the topic (using electronic databases, searches through the reference sections of rel-
evant studies, and leads from other investigators), we retrieved forty-two independent
estimates of the association, oreffect sizes, for religious involvement and mortality, in-
corporating data from 125,826 people. We coded these forty-two effect sizes for a variety
of qualities, including (a) how religiousness was measured; (b) percentage of males and