Empirical Support for a Model of Well-Being, Meaning in Life, Importance of Religion, and Transcendent Experiences

The structural equation goodness of fit measures indicate the chain model fits the data well. For the analysis using the maximum likelihood method without measurement error, the goodness of fit chi-square test is not significant (chi-square=5.43, 3 df, N=182, p=.14), the Comparative Fit Index is .978, Bentler and Bonett's Non-normed Fit Index is .955, and the Goodness of Fit Index is .985. (For these indexes, values near 1.0 indicate a good fit and .90 is generally considered the minimum acceptable value.). The goodness of fit results were slightly better for the analyses with measurement error and with weighted least squares. The results for the individual paths are summarized separately below.

To give a reasonable estimate of the effects of measurement error, the results for a reliability of .65 for the single-item measures are presented. The analyses with different reliabilities showed that as reliability decreased, the magnitude of the significant (non-zero) path coefficients monotonically increased and the t values (significance levels) of the path coefficients decreased slightly. On the other hand, the magnitude of nonsignificant (near zero) path coefficients and the associated t values tended to drift near zero as reliability decreased. The magnitudes of the path coefficients were very similar with the weighted least squares method as with the maximum likelihood method. The t values tended to be slightly lower with weighted least squares. For simplicity, only the results for the maximum likelihood method are reported here.

Meaning in Life and Well-Being

Consistent with the chain model, the path coefficients between meaning in life and well-being are .49 and .63 without and with measurement error, respectively. As shown in Table 1, the t values of 6.81 and 5.11 from the full model are well above the value 2.0 that indicates a nonzero path coefficient. These t values test whether the relationship between the two variables is significant after adjusting for the other variables in the model.

These path coefficients are at the low end and middle of the range of correlations typically found in other studies (see Table 1). The correlation coefficient for this study is .49, which is at the low end of the range of typical values. This result is not surprising because the single-item meaning in life measure used in this study should have a lower reliability than the scales used in the majority of the other studies. The path coefficient adjusted for measurement error is .63, which is in the middle of the range of typical values. The fact that the meaning in life measure used here does not have the overlap with life satisfaction and depression found in the longer scales probably also contributes to the lower correlations in this study.
1

TABLE 1.
Paths Expected to be Significant with the Causal Chain Path Model

Other studies typically found correlations of .50 to .75 between meaning in life and well-being. Zika and Chamberlain (1987, 1992b) investigated the correlation between three different meaning in life scales and the three components of well-being. The four samples were from New Zealand and included 194 mothers, 150 elders, 160 students, and 120 randomly selected community adults. The majority of the various correlations between meaning in life and well-being components were in the range of .50 to .75. Harlow, Newcomb and Bentler (1986) in a study of 722 young adults found correlations of .64 for females and .65 for males between the Purpose in Life test and a factor consisting of positive affect, negative affect, impaired motivation, and impaired relationships. In a sample of 560 randomly selected residents of Akron, Ohio, Poloma and Pendleton (1990) found that a 2-item meaning in life scale correlated .54 with a 4-item life satisfaction scale, .44 with a one-item happiness measure, and -.25 with a four-item negative affect scale. In general, these studies found that meaning in life has a higher correlation with positive affect and life satisfaction than with (lack of) negative affect. Also, as expected, the correlations tended to be higher with more reliable measures.

The path coefficients between importance of religion and meaning in life are .39 and .58 without and with measurement error respectively (see Table 1). Consistent with the chain model, the values are very significantly different from zero.

These two values are at the upper end and above the range of correlations typically found in other studies. The .39 correlation for this study is at the upper end of the range of typical values rather than at the lower end as would be expected based on the reliability of the single-item measures used here. The higher values in the present study may reflect the fact that our importance of religion question differed from the questions normally used and was specifically linked to purpose in life. The characteristics of the selected sample in this study may also be a factor in the higher correlation.

Other studies typically found correlations of .25 to .40. With 318 randomly selected residents of Memphis, Peterson and Roy (1985) found that a three-item measure of importance of religion correlated .25 with a three-item meaning and purpose scale. Paloutzian, Jackson, and Crandall (1978) report that the correlations between a standard intrinsic religiosity scale and the Purpose in Life test were .34 for 84 college students and .37 for 177 adults. Crandall and Rasmussen (1975) also found that the correlation between the Purpose in Life scale and a standard intrinsic religiosity scale was .31 for 71 college students. Chamberlain and Zika (1988) report that for the sample of 188 New Zealand mothers, a measure related to intrinsic religiosity correlated r=.27 with the Purpose in Life scale, and .34 and .25 with two other meaning in life scales. For a sample of 822 church members, King and Hunt (1975) found that an 8-item subscale (salience:cognitive) that included 3 items from intrinsic religiosity scales correlated .41 with a 5-item positive meaning in life subscale and -.25 with a 4-item negative (lack of) meaning in life subscale.

Consistent with the chain model, the path coefficients between transcendent experiences and importance of religion are very significantly different from zero. As shown in Table 1, the path coefficients are .39 and .60 without and with measurement error, respectively. These values are in the low end and upper end of the range of typical correlations found in other studies, which is consistent with the expected lower reliability of the single-item measures in the present study.

A reasonable estimate for the relevant range of correlations typically found in other studies is .35 to .60, however, the results vary widely due to the lack of standard measures for transcendent or mystical experiences. Hay and Morisy (1978) found in a national Survey in Great Britain that two questions about mystical or religious experiences correlated .37 and .40 with a question about the importance of the spiritual side of life. In a series of studies with college students, Hood (1970, 1971, 1972, 1973, 1975, 1978) found correlations between intrinsic religiosity and mystical or other intense religious experiences of .51, .56, .50, .61, .81, and .25. Hood used varying methods for measuring mystical experiences in these studies, and, in two studies, intrinsic religiosity was combined (confounded) with extrinsic religiosity (Hood, 1972, 1973). In a sample of 83 medical outpatients participating in a meditation program, Kass, et al, (1991) reported a correlation of .69 between scores on a standard intrinsic religiosity scale and scores on the INSPIRIT scale for spiritual experiences. The INSPIRIT scale covers certain beliefs and practices in addition to spiritual experiences. Similarly, VandeCreek, Ayres and Bassham (1995) reported a correlation of .61 between a standard intrinsic religiosity scale and the INSPIRIT scale for 247 hospital cancer patients and 124 family members.

Importance of Religion and Well-Being

Consistent with the chain model, the direct path coefficients between importance of religion and well-being are very close to zero for both the models with and without measurement error. These path coefficients and t values test whether importance of religion is related to well-being after adjustment for the mediating role of meaning in life. As shown in Table 2, the t values do not approach significance.

Chamberlain and Zika (1992b) reported similar results for samples of New Zealand mothers, elders, and church members. Using religiosity measures that overlapped with intrinsic religiosity, they concluded that meaning in life mediated the effect of religiosity on well-being. Consistent with the mediation or chain model, they reported that (a) the association between religiosity and the components of well-being became nonsignificant when adjusted for scores on the Purpose in Life scale, and (b) the relationships between meaning in life and the well-being components remained significant after adjusting for religiosity. Unfortunately, their report does not give quantitative information and the results for part of the data, reported in Chamberlain and Zika (1988), provide mixed support for their conclusion.³

In the present study, the observed correlation between importance of religion and well-being is .23 (p<.002) and the correlation predicted by the chain model is .19 (see Table 2). These correlations are another way of showing that virtually all of the relationship between well-being and importance of religion can be explained by mediation by meaning in life. To get an indication of how consistent the present data are with other studies, the predicted correlation was also estimated using the midpoints of the ranges of typical correlations shown in Table 1. The predicted correlation is the product of the correlation between importance of religion and meaning in life and the correlation between meaning in life and well-being. This predicted correlation is .20, which is very close to the .19 value estimated with the path coefficients from the present data.

The direct correlations between importance of religion and well-being found in other studies are typically about .15 to .20 for mixed age groups and higher for elders. In a large national survey, Bortner and Hultsch (1970) found that a multi-item life satisfaction measure correlated .17 with a basic importance of religion question. For another large national sample, Hadaway (1978) reported that "importance of having a strong religious faith" correlated .16 with a 9-item life satisfaction scale and .10 with one life satisfaction question. In a sample of 836 older adults, Koenig, Kvale and Ferrel (1988) found a geriatric morale or well-being scale correlated .24 with a standard intrinsic religiosity scale. In a sample of 85 persons aged 65-85, Hunsberger (1985) found that overall happiness correlated .30 with the importance of religious beliefs. McIntosh, Silver and Wortman (1993) in a study of 124 parents who had lost a child to sudden death syndrome found that the correlation between an importance of religion question and a multi-item well-being scale was .18 at 3 weeks after the loss and .05 at 18 months after the loss. In a study of 102 retired blacks, Jackson, Bacon and Peterson (1977-78) found intrinsic religiosity correlated .06 (not significant) with life satisfaction, however, the relationship became significant when adjusted for other covariates (ß=.283). Because these studies generally used one-item measures for at least one of the variables, the correlations are probably on the low side.

Consistent with the chain model, the direct path coefficients between transcendent experiences and well-being are not significantly different from zero for analysis with and without measurement errors. These tests evaluate whether transcendent experiences are related to well-being after adjustment for the mediating roles of importance of religion and meaning in life.

The observed correlation of .08 is very close to the .07 correlation predicted by the chain model but is not significantly different from zero. For comparison, the predicted correlation using the midpoints of the typical correlations in other studies is .10. Given the very low correlation predicted by the model, large sample sizes would be needed to obtain significant evidence that the correlation is not zero.

As predicted by the model, two large surveys found evidence for very low but statistically significant correlations between mystical experiences and well-being. In a U.S. national survey, Greeley (1975:60-62) found a correlation of about .09 (p<.001) between a well-being scale and reports that mystical experiences occurred "often" (correlation estimated from the Yule's Q value reported by Greeley). Mystical experiences were also correlated separately to a slightly lesser degree with positive affect, life satisfaction, and low negative affect. Greeley (1975:79) also reported that the significant correlation with well-being could be attributed entirely to the respondents with "authentic" mystical experiences. In a national survey in Great Britain, Hay and Morisy (1978) found that the correlation between the same mystical experience question and the same well-being scale used by Greeley was .05 (p<.01), and that another religious experience question correlated .03 (p<.05) with the well-being scale.

The present data do not fully resolve the issue of whether an intervening role for importance of religion is the only connection between transcendent experiences and meaning in life. As shown in Table 2, the analysis that adjusts for measurement error is consistent with the chain model and shows no evidence for a direct connection. This result is probably more accurate than the analysis that assumes no measurement error, which found an equivocal t value of 1.94 for the direct path coefficient. Equivalently, the observed correlation of .27 is noticeably higher than the predicted correlations of .15 using path coefficients from the present study or .15 using the midpoints of typical correlations in other studies. We found no other studies that reported correlations between transcendent experiences and meaning in life.

Given the lack of data and the uncertainties of structural equation methods, it is reasonable to conclude at present that most, but possibly not all, of the connection between transcendent experiences and meaning in life appears to be mediated by importance of religion.

DISCUSSION AND CONCLUSIONS

The available data are very consistent with the model that the importance of religion to a person affects the person's sense of meaning in life, which in turn affects the person's subjective well-being. The present data and the results of other studies are consistent with this model.

This mediating role for meaning in life explains virtually all of the correlation between intrinsic religiosity⁴ and well-being and suggests that other mechanisms for intrinsic religiosity to affect well-being have a minor role. The relatively low positive correlation between well-being and intrinsic religiosity scores reflects that facts that, for the population as a whole, (a) importance of religion is one of several factors that affect meaning in life, and (b) meaning in life is one of several factors that affect well-being.

This model does not imply that intrinsic religiosity is the only aspect of religion that affects well-being. However, other religious dimensions that affect well-being and/or meaning in life should be relatively independent of (not correlated with) intrinsic religiosity.⁵

The available data are also consistent with the model that transcendent or mystical experiences affect the importance of religion, which in turn, affects meaning in life and well-being. Thus, in this causal chain model, importance of religion mediates the impacts of transcendent experiences. This model predicts that the net correlation between transcendent experiences and well-being will be positive, but very small. The available data are very consistent with this prediction. Here too, transcendent experiences appear to be one of several factors that affect the importance of religion to a person. In terms of population statistics, the impacts of transcendent experiences become increasingly diluted at each step in the chain. Of course, the effects for individuals may vary greatly from the overall population averages.

The high likelihood of reciprocal causation is the greatest uncertainty with this model. The likely reciprocal relationship between mystical experiences and religious commitment is the clearest example. As suggested by Greeley (1976:141), a "cycle of reinforcement" between mystical experiences and religious orientation seems plausible. Such a cycle would manifest as a positive correlation in cross sectional studies like those reviewed here.

Traumatic events possibly may induce a partial or complete reversal of the causal chain. Baumeister (1991:232-268) concludes that a health crisis or traumatic loss tends to cause a loss of meaning in life. This theory implies that a severe decrease in well-being can affect a person's sense of meaning in life. Further, Baumeister (1991:232) suggests that "suffering stimulates the needs for meaning" because "people analyze and question their sufferings far more than their joys." A crisis or severe loss leads many people to re-evaluate their world view and/or life priorities, which in turn, leads to an increase in the importance of religious beliefs for a portion of these people. Numerous studies have found that a crisis leads to increased importance of religious beliefs for some people (e.g., Hall, 1986; Koenig, 1994:428-430; Lehman, et al., 1993; Reed, 1987). These and other studies also show that for a smaller portion of people, the loss of meaning from a traumatic event apparently propigates back another step on the chain and causes a reduction or loss of religious commitment. At this point, the fact that religious or mystical experiences are sometimes associated with or triggered by a crisis or despair (Gallup and Castelli, 1989:68; Hardy, 1979:28) becomes particularly intriguing; however, to our knowledge, the preceeding and subsequent relationships with religious commitment have not been investigated for these cases.

Compensating reciprocal causation or feedback is implied with these responses to traumatic events. A loss of meaning in life induces an increase in importance of religion, which then increases or restores meaning in life. This compensating (rather than reinforcing) feedback would result in misleadingly low correlations between the variables in cross sectional studies and may be a reason that the correlations between importance of religion and meaning in life are lower than might be expected (typical correlations of .25 to .40).

On the other hand, a strong sense of meaning and purpose can make compensating feedback unnecessary because the person is resilient to a crisis or loss. Baumeister (1991:233) notes that people are willing to endure pain and misfortune if they believe there is a meaning or purpose for it. In fact, people voluntarily undertake experiences that are unpleasant or even traumatic if there is a reason. Athletic training is a mild example and war is an extreme example. When people's sense of meaning and purpose withstands a traumatic experience, the normal positive correlation between meaning and well-being may become dissociated. In time of war, people who are fighting for a cause may have simultaneously a high sense of purpose and low well-being.

What is noteworthy, and the main point of this paper, is that the available data and literature suggest that causal effects in either direction remain on the proposed chain path. The direction of causation may vary; but meaning in life appears to mediate between well-being and intrinsic religiosity with either causal direction. The chain model identifies a sequential path, but cannot prescribe which direction the causes flow on the path or that they always flow in one direction. This conclusion is consistent with the limited ability of path analysis to verify the direction of causation in cross sectional studies. Similarly, given the uncertainties in the current state of methodology, it would be premature to conclude that mediation by meaning in life is absolutely the only connection between intrinsic religiosity and well-being. However, the available data suggest that mediation by meaning in life dominates the relationship between intrinsic religiosity and well-being. In addition, the evidence for the chain model is less convincing for the transcendent experiences end of the model because of the lack of standard methods for investigating these experiences.

These ideas about the process that links intrinsic religiosity and well-being may be of value to those who help people with their spiritual needs. In addition, this review leads to three points in particular regarding future research:

1. Methods that measure various types and degrees of transcendent and religious experiences and corresponding distinctions in terminology need to be developed.

2. Given the central role of meaning in life, research is needed to better understand what factors lead to a strong sense of meaning in life and the relationship between transcendent meanings and more mundane goals.

3. We believe it is likely that meaning in life plays an important role in the trait aspect of well-being. If this is true, meaning in life may offer a means for stable, positive shifts of well-being. Consistent with this hypothesis, Koenig (1994:431-437) provides evidence that religious conversions can produce long-term, increased well-being. These ideas merit further investigation, particularly because there is a major gap in understanding factors that can induce long-term positive shifts in well-being.