Religiosity is generally associated with positive health outcomes in adult populations (Lee & Newberg, 2005; Oman & Thoresen, 2005; Powell, Shahabi, & Thoresen, 2003; Seeman, Dubin, & Seeman, 2003; Seybold & Hill, 2001). For example, church attendance is related to a decreased risk of cardiovascular disease and cancer, lower mortality rates, better blood pressure, and increased immune functioning. Despite the relation between religiosity and health, other research has shown that certain religious populations are at an increased risk for obesity--a known predictor of multiple adverse health outcomes (Koenig, 2012). Protestants have shown an increased risk for obesity, which some have attributed, at least in part, to the prominent role food plays in Protestant church life (Dodor, 2012; Feinstein, Liu, Ning, Fitchett, & Lloyd-Jones, 2010; Ferraro, 1998, Sack, 2001).

Health-related quality of life (HRQOL), which focuses on individuals' assessment of subjective perceptions of their physical and psychosocial well-being, has been increasingly acknowledged as an important health outcome measure (WHOQOL Group, 1993). Findings from both patient and university populations support a positive association between religiosity and HRQOL (Corrigan, McCorkle, Schell, & Kidder, 2003; Ferriss, 2002; Poston & Turnbull, 2004). For example, the degree to which adult patients with advanced cancer considered themselves to be religious was associated with better overall HRQOL (Vallurupalli et al., 2012). In domestic and international university students in New Zealand, Hsien-Chuan Hsu et al. (2009) found religiosity was associated with greater psychological and social quality of life.

Obesity is a risk factor for diminished HRQOL. In a meta-analysis, Ul-Haq and colleagues (2013a) found obese adults manifest diminished physical quality of life compared to normal weight adults (weighted mean difference [WMD] = -3.73; 95% confidence interval = -5.54 to -1.92) (1). Adults with class III obesity demonstrated significantly lower mental quality of life than normal weight adults (WMD = -1.75; 95% confidence interval= -3.33 to -0.16). A similar pattern was observed in a separate meta-analysis conducted in youth and young adults up to age 19 (Ul-Haq, Mackay, Fenwick, & Pell, 2013b). Significantly lower physical (WMD = -11.93; 95% confidence interval= -15.13 to -8.74) and psychosocial (WMD = -999; 95% confidence interval= -1398 to -6.01) quality of life were self-reported in youth and young adults who were obese. Greater deficits in health-related quality of life were observed with increasing BMI categories.

Analytic Models

There are a number of ways that scholars have posited how religious practice can influence people's reaction to stress (Nelson, 2009; Pargament, 1997). Concerning health outcomes, there are five common analytic models (Chatters, 2000; See Figure 1 for simplified versions of these models.) The suppressor model (2) postulates that the presence of a stressor leads individuals to increase their religious activities (e.g., prayer, service attendance), which then functions to reduce (suppress) the deleterious effects of stress on health. In the distress-deterrent (counterbalancing) model, stress and religion have independent, opposite influences on health. In the prevention model, religion exerts both direct and indirect protective effects on health. Religious involvement benefits health indirectly by lessening the stressor. In the moderator model, the stressor interacts with religion such that the stressor's deleterious effect on health varies by degree of religious involvement (i.e., religion has a buffering effect by weakening the effect of the stressor on health). Last, there is the health effects model, which is similar to suppressor model, but the direction of the stressor's relation to religion is the opposite. With this model, the stressor prevents certain types of religious activity.

Current Study

To date, there has been limited research on the relations between religiosity, HRQOL, and BMI. The purpose of the present study was to examine how religiosity relates to health-related quality of life using BMI as a stressor. We fit and compared the five common analytic models used to explain how religious practice can influence people's reaction to stress.

Method

Participants

Participants were 264 undergraduate students from a private, religiously-affiliated university in the United States. Demographic information for the sample is presented in Table 1. Given the small correlations between sex and the main study outcome variables (Table 2), the decision was made to combine the 225 females and 38 males (one participant had missing sex information) into one sample for the statistical analyses.

Measures

All respondents completed HRQOL, religiosity, and demographic instruments. In addition, we measured respondents' height and weight.

Health-related quality of life. We measured HRQOL using the fourth edition of the Pediatric Quality of Life Inventory Young Adult Scales[TM] (PedsQL 4.0[TM]; Varni & Limbers, 2009). The PedsQL 4.0 is a self-report measure for ages 18 to 25 years. It includes 23 items across four domains: (a) Physical Functioning (8 items), (b) Emotional Functioning (5 items), (c) Social Functioning (5 items), and (d) Work/School Functioning (5 items). Respondents indicate how much of a problem each item's statement has been for them in the past month using a 5-point scale (0 = never a problem; 1 = almost never a problem; 2 = sometimes a problem; 3 = often a problem; 4 = almost always a problem). After scoring the items, they are transformed to a 0-100 scale with higher scores indicating better HRQOL.

Religiosity. To measure religiosity, we adapted items from the in-home interview of the National Longitudinal Study of Adolescent Health (Harris et al., 2009). The item stems were: (a) How important is religious faith in shaping how you live your daily life (0=Extremely important, 1=Very, 2=Somewhat, 3=Not very, or 4=Not important at all)? and (b) How often do you attend religious services not counting wedding, baptisms, and funerals (0=Never, 1= A few times a year, 2=Many times a year, 3= Once a month, 4= 23 times a month, 5=Once a week, 6=More than once a week)? Participants also responded to the question, "Which religion best describes you?" (Buddhist, Christian Catholic, Christian Protestant, Eastern Orthodox, Hindu, Islamic, Jewish, Muslim, Nation of Islam, Native American, Orthodox (Eastern), Pagan or Wiccan, or Other--).

Body Mass Index. We calculated respondents' body mass index (BMI) from their height and weight measures, computed as weight in kilograms divided by height in meters squared.

Procedures

We recruited participants from a private university in the southwest part of America. Although the university is religiously affiliated, students are not required to endorse any particular religion for admittance. Undergraduate students in introductory psychology courses participated in this research project as part of their course requirements. Students from a variety of majors take these courses. We collected data for this study in person through groups of 15-20 individuals.

Before respondents started the study, we obtained their written informed consent. After providing consent, participants completed the measures in the following order: demographic questionnaire, religiosity questionnaire, and PedsQL[TM] 4.0. Upon completing the paper-and-pencil measures, a trained research assistant escorted respondents individually to a separate room and weighed and measured them. The university's Institutional Review Board (IRB) approved the study procedures before data collection began.

Statistical Analyses

We fit path models representing each of the Models A-E in Figure 1 (Loehlin & Beaujean, 2016). Models A and E are structurally the same, only differing by the direction of path coefficients. We used each PedsQL scale score (Physical Functioning, Emotional Functioning, Social Functioning, School/Work Functioning) as a separate outcome variable within the same model. After determining the optimal model, we then fit the same set of models but included demographic variables (i.e., gender, age, race/ethnicity, religion). All statistical analyses were conducted in R using the lavaan and psych packages (Revelle, 2017; Rosseel, 2012; R Development Core Team, 2017). Syntax for all analyses are available on the Open Science Framework at osf.io/hegmv.

Model fit. Since the models in Figure 1 are not nested or over-identified, we used two information-criterion measures to compare the models' fit: Akaike's corrected information criterion (AICc) and the sample-size adjusted Bayesian information criterion (aBIC). Both measures seek to find the simplest model that can describe the data well, although they use somewhat different criteria to make this determination (Burnham & Anderson, 2004).

Individual AIC and BIC values are not directly interpretable because they contain arbitrary constants and are affected by sample size. Thus, for both fit measures the typical interpretation is that smaller values indicate a more favored model among the models fit. When between-model differences in AICc/aBIC values are small, however, the acceptance of a single model may lead to a false sense of confidence. Consequently, we calculated the evidence weights for both in favor of the best model for both indices (Wagenmakers & Farrell, 2004). These weights can be interpreted as the probability that a given model is the best model, given the data and the set of models evaluated.

Results

Missing Data

Four respondents (2% of sample) had missing data on either gender (n=1), race (n=1), or BMI (n=2). To handle these missing values, we used full information methods when estimating the parameters (Enders, 2011).

Religiosity

The two religiosity items were strongly correlated (Pearson and polychoric correlations of .73). Instead of using the variables separately, we created a single religiosity variable. We did this by conducting an item factor analysis and then extracting scores from the first factor. We created the factor scores using the method described by ten Berge, Krijnen, Wansbeek, and Shapiro (1999). The correlation between the factor scores and first factor was .92, providing some validity evidence for the scores (Grice, 2001).

Descriptive Statistics

The bottom part of Table 1 presents descriptive statistics for BMI, religiosity, and PedsQL scores. Correlations among the variables are provided in Table 2.

All variables except BMI appear to be univariate normal. We examined multivariate normality using Mardia's tests of multivariate skew and kurtosis. Both values ([b1.sub.p] = 999, [b2.sub.p] = 59.48) as well as a Q-Q plot indicate they are likely not multivariate normal. Consequently, we used a robust estimator for our analyses (MLR), which provides somewhat more accurate standard error estimates with non-normal data.

Models

The top part of Table 3 presents the AICc and aBIC results from fitting the models without any covariates. The distress-deterrent model appears to fit the best, indicating the stressor (BMI) and religiosity have independent relations to all four measures of HRQOL.

The bottom part of Table 3 presents the AICc and aBIC results from fitting the models with the covariates for gender, race, age, and religion included in the model. For all models, the covariates were used to predict all four HRQOL outcomes. Again, the distress-deterrent model appears to fit the best, indicating that after controlling for gender, race, age, and religion, BMI and religiosity are still independently related to all four HRQOL outcomes.

Table 4 contains the path coefficients for the distress-deterrent model with covariates. Across all HRQOL outcomes except physical, BMi and religiosity have opposite relations (BMI negative and religiosity positive). This indicates that higher levels of BMI are associated with poorer perceived emotional, social, and work/school functioning, while higher levels of religiosity are associated with better functioning in the same domains. For physical HRQOL, BMI and religiosity both have negative relations. This indicates that higher levels of BMI and religiosity are associated with poorer perceived physical functioning.

Nonetheless, these relations should be interpreted with some caution. While the distress-deterrent model was the best-fitting model of those examined, BMI and religiosity explain a relatively minimal amount of the variability in all of the PedsQL variables ([R.sup.2] ranging from .01-.05). Moreover, while it appears that BMI may have a stronger influence on perceived functioning than religiosity, confidence intervals for the variables overlap. Thus, we cannot rule out that the two variables have the same relations with the outcome variables.

Discussion

The purpose of the present study was to examine how religiosity relates to health-related quality of life using BMI status as a stressor. We examined five common models for how religiosity and stress relate to health outcomes. In the present study, the distress-deterrent model demonstrated the best overall fit. Thus, stress (i.e., BMI) and religiosity had independent and opposite influences on health-related quality of life. For emotional, social, and work/school functioning, BMI had a negative relation while the relation to religiosity was positive.

Our findings are consistent with a study in older adults with chronic illness that found support for the distress-deterrent model (Boswell, Kahana, & Dilworth-Anderson, 2006). In their path analysis, greater spirituality was associated with better well-being; however, spirituality did not have a significant or positive association with chronic stress. Thus, spirituality was not responsive to stress but was still beneficial to well-being (Boswell et al., 2006). Similarly, in the current study greater religiosity was associated with better perceived emotional, social, and work/school functioning. Religiosity was not associated with BMI status as these variables had independent and opposite influences on emotional, social, and work/school domains of the PedsQL. As such, religiosity was not responsive to the stressor of having an elevated BMI status but was still beneficial to emotional, social, and work/school quality of life.

For physical functioning, however, the relation was negative for both BMI and religiosity. While the finding that elevated BMI was associated with worse physical functioning is consistent with previous literature (Nader et al., 2006), the finding that greater religiosity was associated with worse physical functioning (but not emotional, social, or work/school functioning) is notable. This finding may support the idea that components of Protestant Christian church life contribute to a more sedentary lifestyle. For example, young adults who spend more time attending religious services (which are generally sedentary in nature) may be less likely to engage in physical activity. Although we did not measure levels of physical activity in the present study, in his systematic review Koenig (2012) found the majority (68%) of studies that assessed religiosity and physical activity in adults reported a significant positive correlation. The measurement of physical activity in future studies that assess the relations between religiosity, HRQOL, and BMI status would be valuable to further elucidate mechanisms underlying the negative association between religiosity and physical functioning found in the present study.

Overall, we found religion and BMI explained a relatively small amount of variability in our model. One explanation for the low amount of variability explained by BMI is the age of our sample. Since the deleterious effects of overweight/obesity tend to be more pronounced over time, it is possible that in our sample of young adults elevated weight status was not a major stressor. It will be important for future research to replicate our findings in samples of middle and older-aged adults where it would be anticipated that overweight/obese would be associated with greater health problems (e.g., hypertension, cardiovascular disease, stroke).

With regard to the small amount of variability explained by religiosity, it is possible that effects of religion on HRQOL occur indirectly through an in-between factor that was not accounted for in our model. For example, Stolzfus & Farkas (2012) found the association between daily stress and alcohol use was moderated by positive religious coping among female university students (Stoltzfus & Farkas, 2012). We did not explicitly measure religious coping in the present study. One of the criticisms of the religiosity-health research has been lack of a standard measure of religiosity (Hill & Hood, 1999). Studies in this area have generally utilized measures of religious attendance/religious media practice, religious salience, frequency of prayer, religious identification, or religious consolation (Cline & Ferraro, 2006; Kim, Sobal, & Wethington, 2003). In the present study, the two religiosity items (How important is religious faith in shaping how you live your daily life, How often do you attend religious services) were strongly correlated. Thus, instead of using the variables separately, we created a single religiosity variable. While our single religiosity variable focused on importance of religious faith in shaping daily life and attendance at religious services, it will be important for future research to replicate our findings using other indicators of religiosity such as religious coping, religious media practice, and frequency of prayer.

Limitations

There were a number of limitations in the present study. The sample was made up of predominantly Caucasian, young adult females who self-identified as Protestant Christian. Thus, our findings may not generalize to other populations. The cross-sectional and correlational design of our study limited us from inferring causation. In addition, we did not have information on the socioeconomic status of our participants, and as a result, we were not able to control for this variable in our statistical analysis.

Conclusions

In conclusion, the distress-deterrent model demonstrated the best overall fit in the present study. Among our sample of young adults, BMI status and religiosity had independent (and opposite) influences on emotional, social, and work/school health-related quality of life. The current study may serve as a model for future research that aims to understand how religious practice can influence people's reaction to stress using a comparative model study design.

Notes

(1) The weighted mean difference (WMD) is the average value after pooling results of individual studies (i.e., each study's contribution to the mean difference is weighted by sample size). This effect size is appropriate when all the studies in a meta-analysis use the same instrument, which was the case in the Ul-Haq et al. meta-analyses.

(2) The term suppressor is used differently in these models than how it is typically defined statistically.

References

Boswell, G. H., Kahana, E., & Dilworth-Anderson, P. (2006). Spirituality and healthy lifestyle behaviors: stress counter-balancing effects on the well-being of older adults. Journal of Religion and Health, 45, 587-602.

Burnham, K. P., & Anderson, D. R. (2004). Multi-model inference: Understanding AIC and BIC in model selection. Sociological Methods & Research, 33, 261-304.

Chatters, L. M. (2000). Religion and health: Public health research and practice. Annual Review of Public Health, 21, 335-367.

Cline, K. C., & Ferraro, K. F. (2006). Does religion increase the prevalence and incidence of obesity in adulthood? Journal for the Scientific Study of Religion, 45, 269-281.

Corrigan, P., McCorkle, B., Schell, B., & Kidder, K. (2003). Religion and spirituality in the lives of people with serious mental illness. Community Mental Health Journal, 39, 487-499.

Dodor, B. (2012). The impact of religiosity on health behaviors and obesity among African Americans. Journal of Human Behavior in the Social Environment, 22, 451-462.

Enders, C. K. (2011). Analyzing longitudinal data with missing values. Rehabilitation Psychoogy, 56, 267-288.

Feinstein, M., Liu, K., Ning, H., Fitchett, H., & Lloyd-Jones, D. M. (2010). Burden of cardiovascular risk factors, subclinical atherosclerosis, and incident cardiovascular events across dimensions of religiosity: The multi-ethnic study of atherosclerosis. Circulation, 121, 659-666.

Ferraro, K. F. (1998). Firm believers? Religion, body weight, and well being. Review of Religious Research, 39, 222-244.

Ferriss, A. (2002). Religion and the quality of life. Journal of Happiness Studies, 3, 199-215.

Grice, J. W. (2001). Computing and evaluating factor scores. Psychological Methods, 6, 430-450.

Harris, K. M., Halpern, C. T., Whitsel, E., Hussey, J., Tabor, J., Entzel, P., & Udry, J. R. (2009). The National Longitudinal Study of Adolescent Health: Research Design. Retrieved from http://www.cpc.unc.edu/projects/addhealth/design

Hill, P. C., & Hood, R. W. (Eds.). (1999). Measures of religiosity. Birmingham, AL: Religious Education Press.

Hsien-Chuan Hsu, P., Krageloh, C. U., Shepherd, D., & Billington, R. (2009). Religion/spirituality and quality of life of international tertiary students in New Zealand: An exploratory study. Mental Health, Religion, and Culture, 12, 385-399.

Kim, K. H., Sobal, J., & Wethington, E. (2003). Religion and body weight. International Journal of Obesity and Related Metabolic Disorders, 27, 469-477.

Koenig, H. G. (2012). Religion, spirituality, and health: The research and clinical implications. International Scholarly Research Notices, 1-33. http://dx.doi.org/10.5402/2012/278730.

Lee, B. Y., & Newberg, A. B. (2005). Religion and health: A review and critical analysis. Zygon, 40, 443-468.

Loehlin, J. C., & Beaujean, A. A. (2016). Latent variable models: An introduction to factor, path, and structural equation analysis (5th ed.). New York, NY: Routledge.

Nader, P. R., O' Brien, M., Houts, R., Bradley, R., Belsky, J., Crosnoe, R., & Susman, E. J. (2006). Identifying risk for obesity in early childhood. Pediatrics, 118, e594-e601.

Nelson, J. M. (2009). Psychology, religion, and spirituality. New York, NY: Springer.

Oman, D., & Thoresen, C. E. (2005). Handbook of the psychology of religion and spirituality. New York, NY: The Guilford Press.

Pargament, K. I. (1997). The psychology of religion and coping: Theory, research, practice. New York, NY: Guilford Press.

Poston, D., & Turnbull, A. P (2004). Role of spirituality and religion in family quality of life for families of children with disabilities. Education and Training in Developmental Disabilities, 39, 95-108.

Powell, L. H., Shahabi, L., & Thoresen, C. E. (2003). Religion and spirituality: Linkages to physical health. American Psychologist, 58, 36-52.

Revelle, W. (2017). psych: Procedures for psychological, psychometric, and personality research (Version 1.7.8) [Computer program]. Evanston, IL: Northwestern university.

Rosseel, Y. (2012). lavaan: An R package for structural equation modeling. Journal of Statistical Software, 48, 1-36.

R Development Core Team. (2017). R: A language and environment for statistical computing (Version 3-2.3) [Computer program]. Vienna, Austria: R Foundation for Statistical Computing.

Sack, D. (2001). Whitebread Protestants: Food and religion in American culture. New York, NY: Palgrave.

Seeman, T. E., Dubin, L. F., & Seeman, M. (2003). Religiosity/spirituality and health: A critical review of the evidence for biological pathways. American Psychologist, 58, 53-63.

Seybold, K. S., & Hill, P. C. (2001). The role of religion and spirituality in mental and physical health. Current Directions in Psychological Science, 10, 21-24.

Stoltzfus, K. M., & Farkas, K. J. (2012). Alcohol use, daily hassles, and religious coping among students at a religiously affiliated college. Substance Use and Misuse, 47, 1134-1142.

ten Berge, J. M. F., Krijnen, W. P., Wansbeek, T., & Shapiro, A. M. (1999). Some new results on correlation-preserving factor scores prediction methods. Linear Algebra and its Applications, 289, 311-318.

Ul-Haq, Z., Mackay, D. F., Fenwick, E., & Pell, J. P. (2013a). Meta-analysis of the association between body mass index and health-related quality of life among adults, assessed by the SF-36. Obesity, 21, E322-E327.

Ul-Haq, Z., Mackay, D. F., Fenwick, E., & Pell, J. P. (2013b). Meta-analysis of the association between body mass index and health-related quality of life among children and adolescents, assessed using the pediatric quality of life inventory index. Journal of Pediatrics, 162, 280-286.

Vallurupalli, M., Lauderdale, K., Balboni, M. J., Phelps, A. C., Block, S. D., Ng, A. K., & Balboni, T. A. (2012). The role of spirituality and religious coping in the quality of life of patients with advanced cancer receiving palliative radiation therapy. The Journal of Supportive Oncology, 10, 81-87.

Varni, J. W., & Limbers, C. A. (2009). The PedsQL 4.0 Generic Core Scales young adult version: Feasibility, reliability and validity in a university student population. Journal of Health Psychology, 14, 611-622

Wagenmakers, E.-J., & Farrell, S. (2004). AIC model selection using Akaike weights. Psychonomic Bulletin & Review, 11, 192-196.

WHOQOL Group, W. (1993). Study protocol for the World Health Organization project to develop a Quality of Life assessment instrument. Quality of Life Research, 2, 153-159.

Christine Limbers is an Associate Professor in the Department of Psychology and Neuroscience at Baylor University. Her research focuses on the assessment and validation of neurocognitive and patient-reported outcomes in youth and young adults.

Ashley Teasdale is a Clinical Psychology Doctoral student at Baylor University. Her research interests focus on health outcomes and coping mechanisms used by families who have a member with a chronic health condition.

A. Alexander Beaujean is an Associate Professor in the Department of Psychology and Neuroscience at Baylor University. His research focuses on the application of quantitative methods to study human variability, especially as it relates to cognitive ability and health outcomes.

Christine A. Limbers

Ashley Teasdale

A. Alexander Beaujean

Baylor University

Address correspondence to Christine A. Limbers, Ph.D., Associate Professor, Department of Psychology and Neuroscience, Baylor University, One Bear Place #97334, Waco, TX 76798; Christine_Limbers@baylor.edu.

Caption: Figure 1. Alternative models for how religion relates to stress and health

Table 1
Sample Demographic Information (n=264).

Variable                                             n/Mean (%/SD)

Age (years)                                          19.11 (1.04)
  Range                                              17-23
Female                                               225 (85.23)
  Missing (%)                                        1 (0.38)
Race/Ethnicity
  Caucasian                                          158 (59.85)
  Black                                              37 (14.02)
  Hispanic                                           32 (12.11)
  Asian                                              22 (8.35)
  Multi-racial                                       9 (3.41)
  Other                                              5 (1.89)
  Missing                                            1 (0.38)
How important is religious faith in shaping your
daily life?
  Extremely important                                87 (33.00)
  Very                                               78 (29.55)
  Somewhat                                           63 (23.86)
  Not very                                           29 (10.98)
  Not important at all                               7 (2.65)
How often do you attend religious services?
  Never                                              22 (8.33)
  A few times a year                                 33 (12.50)
  Many times a year                                  17 (6.44)
  Once a month                                       17 (6.44)
  2-3 times a month                                  43 (16.29)
  Once a week                                        71 (26.89)
  More than once a week                              61 (23.11)
What religion best describes you?
  Protestant Christian                               180 (68.18)
  Catholic Christian                                 58 (21.97)
  Hindu                                              6 (2.27)
  Muslim/Islamic                                     4 (1.52)
  Buddhist                                           3 (1.10)
  Agnostic/Atheist/Non-Religious                     5 (1.89)
  Deist                                              1 (0.38)
Other (Not specified)                                7 (2.70)
BMI                                                  23.16 (4.14)
  Range                                              15.30-45.20
  Missing                                            2 (0.76)
BMI Category
  Underweight                                        13 (4.92)
  Normal                                             186 (70.45)
  Overweight                                         47 (17.80)
  Obese                                              16 (6.06)
PedsQL Scale
  Physical Functioning                               83.18 (13.72)
  Emotional Functioning                              69.49 (19.04)
  Social Functioning                                 88.84 (13.52)
  Work/School Functioning                            76.22 (16.96)
Religiosity (factor score)                           0.00 (0.92)

Table 2
Zero-order correlations

                              1.       2.       3.       4.       5.

1        Age                 1.00
2        BMI                 0.07     1.00
3        Sex                -0.01     0.16     1.00
4        Physical            0.01    -0.14     0.14     1.00
         Functioning
5        Emotional          -0.03    -0.15     0.14     0.45     1.00
         Functioning
6        Social              0.05    -0.17    -0.02     0.52     0.54
         Functioning
7        Work/School         0.03    -0.10     0.06     0.45     0.47
         Functioning
8        Reliniositv        -0.05     0.02    -0.14     0.03     0.09
9        Black               0.00     0.05    -0.03     0.01     0.00
10       Asian              -0.02    -0.04     0.35    -0.05     0.04
11       Hispanic            0.02     0.13     0.06    -0.02    -0.13
12       Mixed              -0.10     0.07     0.08     0.03     0.00
13       Other              -0.04    -0.12    -0.06     0.04     0.12
14       Catholic           -0.03     0.13     0.06     0.07     0.03
15       Non-Religious       0.04     0.04     0.08    -0.14    -0.08
16       Non-Christian       0.01    -0.16     0.09     0.03     0.05
17       Religiosity x      -0.03     0.09    -0.05     0.00     0.02
         BMI interaction
SDs                          1.04     4.14     0.35    13.69    19.00
Means                       19.11    23.16     1.14    83.18    69.49

                              6.       7.       8.       9.      10.

1        Age
2        BMI
3        Sex
4        Physical
         Functioning
5        Emotional
         Functioning
6        Social              1.00
         Functioning
7        Work/School         0.42     1.00
         Functioning
8        Reliniositv         0.14     0.05     1.00
9        Black               0.06     0.10     0.03     1.00
10       Asian              -0.09     0.03    -0.05    -0.12     1.00
11       Hispanic           -0.06    -0.07    -0.19    -0.15    -0.11
12       Mixed               0.02     0.03     0.01    -0.08    -0.06
13       Other               0.07     0.09    -0.07    -0.06    -0.04
14       Catholic            0.05     0.00    -0.12    -0.08    -0.03
15       Non-Religious      -0.15    -0.09    -0.45    -0.04     0.12
16       Non-Christian       0.03     0.03    -0.16    -0.09     0.31
17       Religiosity x       0.10     0.09     0.60     0.05     0.03
         BMI interaction
SDs                         13.49    16.93     0.92     0.35     0.28
Means                       88.85    76.22     0.00     0.14     0.28

                             11.      12.      13.

1        Age
2        BMI
3        Sex
4        Physical
         Functioning
5        Emotional
         Functioning
6        Social
         Functioning
7        Work/School
         Functioning
8        Reliniositv
9        Black
10       Asian
11       Hispanic            1.00
12       Mixed              -0.07     1.00
13       Other              -0.05    -0.03     1.00
14       Catholic            0.34    -0.05    -0.07
15       Non-Religious       0.13     0.05    -0.03
16       Non-Christian      -0.03    -0.04     0.35
17       Religiosity x      -0.07    -0.04    -0.10
         BMI interaction
SDs                          0.33     0.18     0.14
Means                        0.12     0.03     0.02

Note. Correlations calculated using full information maximum
likelihood. Race variables were dummy-coded with Caucasian as the
reference group. Religion variables were dummy-coded with Protestant
as the reference group.

Table 3
Model Fit Results.

Model                  AICc     AICc Weight   aBIC     aBIC Weight

Without Covariates

Suppressor/Health      10747    0.05          10752    0.03
Effects
Distress-Deterrent     10740    0.91          10745    0.95
Prevention             10747    0.05          10752    0.03
Moderator              12146    <0.01         12157    <0.01

With Covariates

Suppressor/Health      11547    <0.01         11534    <0.01
Effects
Distress-Deterrent     11421    1.00          11438    1.00
Prevention             11476    <0.01         11463    <0.01
Moderator              12841    <0.01         12828    <0.01

Note. AICc: Corrected Akaike's information criterion, aBIC: Sample-
size adjusted Bayesian information criterion. AICc and aBIC weights
are defined in Equation 1.

[mathematical expression not reproducible], (1)

where [phi] is the information criterion index (i.e., AICc, aBIC)
and [[DELTA].sub.i] is the difference in [phi]
between model i and the "best model" (i.e., model with lowest value
[phi]). [[DELTA].sub.i][phi] is defined as

[[DELTA].sub.i][phi] = [[phi].sub.i] - min([phi]), (2)

where min([phi]) is the lowest information criterion value of
all the models fit.

Table 4
Path Coefficients for Distress-Deterrent Model

Predictor            Unstandardized      Standard
                    Path Coefficient      Error

Outcome: Physical Functioning

BMI                      -0.48             0.22
Religiosity              -0.33             1.00

Race (a)
  Black                   0.91             2.55
  Asian                  -1.61             3.58
  Hispanic               -0.55             2.76
  Multi-racial            4.11             3.40
  Other                   2.50             3.75
Religion (b)
  Catholic                2.55             2.40
  Non-Religious          -8.57             4.73
  Non-Christian           0.56             3.43
Age                       0.48             0.78

Outcome: Emotional Functioning

BMI                      -0.65             0.24
Religiosity               1.96             1.55

Race (a)
  Black                   0.28             3.45
  Asian                   1.91             4.42
  Hispanic               -6.66             4.22
  Multi-racial            1.05             4.21
  Other                  15.51             5.87
Religion (b)
  Catholic                4.95             3.35
  Non-Religious          -0.10             5.14
  Non-Christian           0.68             6.94
Age                      <0.01             1.10

Outcome: Social Functioning

BMI                      -0.58             0.23
Religiosity               1.95             1.08

Race (a)
  Black                   2.39             2.42
  Asian                  -4.44             3.65
  Hispanic               -1.64             3.09
  Multi-racial            3.60             3.20
  Other                   4.01             4.51

Religion (b)
  Catholic                3.77             2.46
  Non-Religious          -3.27             4.03
  Non-Christian           3.55             3.35
Age                    1.04 0.69

Outcome: School/Work Functioning

BMI                      -0.40             0.25
Religiosity               0.44             1.34
Race (a)
  Black                   5.44             3.20
  Asian                   3.72             3.71
  Hispanic               -1.33             3.82
  Multi-racial            5.44             3.80
  Other                  12.41             5.98
Religion (b)
  Catholic                1.41             3.13
  Non-Religious          -5.77             4.25
  Non-Christian          -2.28             8.44
Age                       0.96             1.04

Predictor             Standardized      [R.sup.2]
                    Path Coefficient

Outcome: Physical Functioning              0.05

BMI                      -0.15
Religiosity              -0.02

Race (a)
  Black                   0.02
  Asian                  -0.03
  Hispanic               -0.01
  Multi-racial            0.06
  Other                   0.03
Religion (b)
  Catholic                0.08
  Non-Religious          -0.14
  Non-Christian           0.01
Age                       0.04

Outcome: Emotional Functioning             0.06

BMI                      -0.14
Religiosity               0.10

Race (a)
  Black                   0.01
  Asian                   0.03
  Hispanic               -0.11
  Multi-racial            0.01
  Other                   0.11
Religion (b)
  Catholic                0.11
  Non-Religious          <0.01
  Non-Christian           0.01
Age                      <0.01

Outcome: Social Functioning                0.09

BMI                      -0.18
Religiosity               0.13

Race (a)
  Black                   0.06
  Asian                  -0.09
  Hispanic               -0.04
  Multi-racial            0.05
  Other                   0.04

Religion (b)
  Catholic                0.12
  Non-Religious          -0.05
  Non-Christian           0.06
Age                       0.08

Outcome: School/Work Functioning           0.04

BMI                      -0.10
Religiosity               0.02
Race (a)
  Black                   0.11
  Asian                   0.06
  Hispanic               -0.03
  Multi-racial            0.06
  Other                   0.10
Religion (b)
  Catholic                0.04
  Non-Religious          -0.07
  Non-Christian          -0.03
Age                       0.06

Note. Non-religious: Atheists, Agnostics, Deists, and those that did
not specify a religion. Non-Christian: Hindu, Muslim, and Buddhist.

Intercepts for the models were 84.966 (Physical), 83.498 (Emotional),
81.529 (Social), 66.065 (School)

(a): Dummy-coded variable with Caucasian as the reference group.

(b): Dummy-coded variable with Protestant as the reference group.