Brochures

A MULTIDIMENSIONAL MEASURE OF RELIGIOUS INVOLVEMENT FOR AFRICAN AMERICANS

Description
A MULTIDIMENSIONAL MEASURE OF RELIGIOUS INVOLVEMENT FOR AFRICAN AMERICANS
Categories
Published
of 17
All materials on our website are shared by users. If you have any questions about copyright issues, please report us to resolve them. We are always happy to assist you.
Related Documents
Share
Transcript
  A MULTIDIMENSIONAL MEASURE OF RELIGIOUS INVOLVEMENT FOR AFRICAN AMERICANS Jeffrey S Levin astern Virginia Medical School Robert o s e p h Ta y z   zy   r University of Michigan Linda M. Chatters University of Michigan This article describes the confirmation and validation of a multidimensional measure of religious involvement using data from the National Survey of Black Americans N = 2,107). This model was developed through a multistep strategy of confirmatory factor analysis and structural-equation modeling. First, a three-dimensional factor structure com- prising organizational, nonorganizational, zyxw nd subjective religiosity was confirmed for twelve religious indicators. This measurement model was found to exhibit excellent over- all fit; it compared favorably to alternative models; and all hypothesized factor loadings were strong and statistically significant. Second, several constructs identified by prior re- search as correlates of religious involvement (gender, age, education, region, and urbanic- ity) were found to exhibit significant associations with one or more dimensions of the model. All analyses were conducted in LISREL 8.03, using maximum-likelihood estima- tion and a strategy of split-sample replication. zyxw INTRODUCTION With the advent of the National Survey of Black Americans NSBA), there has been a matura- tion of empirical social research on religious involvement among black adults. Many previous studies of the religious life of African Americans have been 1) anecdotal in nature, 2) de- rived from small nonprobability samples e.g., Heisel and Faulkner 1982), or (3) based on ecologic-level data e.g., Lincoln and Mamiya 1990). Furthermore, studies have often failed to control for the effects of key constructs such as socioeconomic status e.g.. Greeley 1979). Analyses using data from the NSBA have included examinations of the sociohistorical role of black churches Taylor, Thornton, and Chatters 1987), the phenomenon of switching between *Direct all correspondence to Jeff Levin. Department of Family and Community Medicine, Eastern Virginia Medical School, P.O. Box 1980. Norfolk, VA 23501. The Sociologid Quarterly, Volume 36, Number 1, pages 157-173. Copyright zyxwvutsr   995 by Ihe Midwest Sodologid Society. All rights of reproduction in any form requested ISSN 0038-0253.  158 zyxwvuts HE zyxwv OCIOLOGICAL QUARTERLY Vol. 36/No. 1/1995 z religious denominations Ellison and Sherkat 1990; Sherkat and Ellison 1991), correlates of religious involvement among black adults Chatters and Taylor 1989; Taylor 1988a; Taylor 1988b) and older blacks Taylor 1986; Taylor and Chatters 1991). the role of church members in informal social support networks Taylor and Chatters 1986; Taylor and Chatters 1988), and the impact of religious involvement on outcomes such as health and well-being Ellison and Gay 1990; Levin, Chatters, and Taylor forthcoming). In general, these studies have found that most African Americans consider the church to be a beneficial institution, are Baptists, do not engage in denominational switching, and display high levels of religious participation. Despite this apparent progress in substantive areas of black religious research, certain more fundamental issues remain largely unexplored, most notably the conceptualization and mea- surement of religious involvement. Without the use of confirmed and validated religious measures, the reliability of findings bearing on the determinants and outcomes of religiosity may be threatened. In other substantive areas of sociology, the development and confirmation of multidimensional measures has established internally consistent, content-valid instruments for a variety of psychosocial constructs, notably health and well-being. These constructs are thus amenable to sophisticated modeling in multifactorial models in which researchers can account for errors of both measurement and theoretical specification. Within the sociology of religion, investigation of the internal structure of religious beliefs, attitudes, and practices has represented a central focus of empirical research Gorsuch 1984; Cornwall, Albrecht, Cunningham, and Pitcher 1986; Hilty and Stockman 1986). Little of this work, however, has been based on samples of African Americans e.g., Jacobson, Heaton, and Dennis 1990), despite a rich tradition of sociological writing that emphasizes the uniqueness of the black experience in religion due to circumstances of life in the African diaspora Lin- coln and Mamiya 1990). While considerable research exists on the psychometric properties of various multidimensional indices primarily among whites, most research on the impact of religion in the lives of African Americans has relied upon single-item measures or on scales not validated for use with this population. Consequently, researchers are required to assume that these indices represent useful measures of black religious involvement. This article seeks to address this deficiency using the National Survey of Black Americans sample. First, a brief outline will be provided of key conceptual and operational develop- ments bearing on the multidimensionality of religious involvement. Second, based on prior work in sociology and psychology, a three-dimensional measurement model will be developed using confirmatory factor analysis. This model will be compared to alternative one- and two- dimensional models. Finally, the proposed model will be further validated by examining the effects of key sociodemographic correlates of its dimensions. Previous research has identified several such variables as predictive of religious involvement. MEASUREMENT OF RELIGIOSITY The conceptualization and measurement of religious constructs have been key issues in the sociology of religion for over thirty years Gorsuch 1984; Robinson and Shaver 1969). Yet the development of reliable and content-valid measures has not led to their unanimous adop- tion by researchers in other fields whose studies require collecting data on religion. It is perhaps presumed that one indicator is as good as another or that religious involvement is a unidimensional construct. As Talcott Parsons once noted, however, “It is an elementary mis- take to assume that everything bearing the semantic label of ‘religion’ can be dealt with as one phenomenon” Parsons 1961, p. 9). Religion is a complex and variegated human phenome-  A zyxwvutsr ultidimensional Measure zyxwvu 59 z non, and the often used term “religiosity,” for example, seems to cover considerable ground: behaviors, attitudes, values, beliefs, feelings, experiences. At best, this represents “a diverse range of variables only loosely arrayed under the rubric of ‘religion’ zyx   Dittes 1971, p. 79). The notion of a single social-psychological construct termed religiosity is highly improbable. Rather than a single, unidimensional construct, personal religiosity may be better conceived of as a domain of human life. The assessment of different aspects of church or synagogue partic- ipation or being religious taps different constructs and requires a variety of measures. This differentiation of personal religiosity into component parts has its srcins in both soci- ological and psychological writing on religion. In zyxw ociology zyx f Religion, Joachim Wach 1944) proposed that religious expression occurred in three dimensions: “theoretical,” repre- senting the strength of doctrinal adherence; “cultic,” involving the depth of devotional prac- tice and worship; and “sociological,” reflecting the extent of fellowship activities. Gordon W. Allport 1958) later described two basic types of religious orientations, which he called “insti- tutionalized“ and “interiorized.” The former refers to those individuals whose engagement of religion is manifest primarily in public, organizationally-based religious behavior; the latter denotes those who engage religion at a deeper, more personal level. Throughout the 1960s, considerable empirical efforts were made to refine the conceptual- ization of religious involvement and produce valid multidimensional measures. Much of this work has been criticized for confounding measures of religious involvement and measures of religious knowledge or specific doctrines see Himmelfarb 1975). Early research sought to delineate the dimensionality of religious expression, and numerous typologies were proposed. As described by Harold S. Himmelfarb 1975). models with four Lenski 1961), five Glock and Stark 1965), nine King 1967), ten King and Hunt 1972), and eleven King and Hunt 1969) dimensions were proposed, covering such divergent constructs as religious devotion, doctrine, experience, affiliation, ideology, affect, ethics, and others. In his review of these models, Himmelfarb 1975) concluded that across all typologies there seemed to be a funda- mental distinction between “behavioral” and “ideational” religiosity. Concurrently, Allport’s distinction between institutionalized and interiorized religion led to the conceptualization of “extrinsic” and “intrinsic” religiosity see Kahoe 1985), and to efforts at developing useful measures of these constructs Donahue 1985). This distinction was made not so much to differentiate types or categories of religious measures as in the distinction between behaviors and ideations), but to denote substantively different ways in which relig- iousness is expressed Hood 1985). Extrinsic religiosity was defined as a more public and instrumental way of being religious, and people who were more extrinsically religious were believed to be more prejudiced and authoritarian. Conversely, intrinsic religiosity was defined as a more private and expressive way of being religious, and predominantly intrinsic individu- als were believed to be more tolerant, reflective, and mature Allport 1950; Hood 1985). While prior research has encouraged distinctions between religious behaviors and religious ideations, and between intrinsic and extrinsic religiosity, these conceptual advances have not resulted in widespread changes in the way religious involvement is measured in empirical social research. Specifically, the incorporation of validated scales into national social surveys has been limited see Carroll and Roozen 1973). Single-item religious measures still predominate Gorsuch and McFarland 1972), especially in substantive research examining the effects of religion on mental health, particularly in terms of depression, anxiety, suicide, drug and alcohol abuse, psychiatric care utilization, life satisfaction, and self-esteem Gartner, Lar- son, and Allen 1991). Multidimensional scales, when utilized, are often proposed solely on  160 zyxwvut HE SOCIOLOGICAL QUARTERLY Vol. 36/No. zy /1995 z the basis of exploratory analyses and without statistical confirmation or accounting for the ubiquitous errors of measurement implicit in multi-item psychosocial constructs. Further, the indiscriminate mixing of religious behaviors, beliefs, and attitudes in single dimensions or in unidimensional instruments remains common Levin and Tobin forthcoming). Recent work in the sociology of aging has begun to address these conceptual distinctions by developing useful multidimensional instruments of religious involvement. This has proven to be a fruitful mid- dle ground between focusing solely on specific beliefs or doctrines and postulating a dozen or more dimensions in order to span the entire content of religious expression. In the tradition of Allport (1950; 1958), Charles H. Mindel and C. Edwin Vaughan (1978) differentiated between “organizational” and “nonorganizational“ religiosity-distinguishing between public, institutional forms of religious involvement e.g., religious attendance) and more private, informal types of involvement e.g., prayer, watching religious TV programs). Neal Krause and Thanh Van Tran (1989) subjected this dichotomous specification to confir- matory factor analysis using data on older adults from the National Survey of Black Ameri- cans NSBA) and found an excellent overall fit. In both studies, nonorganizational religiosity included variables measuring both private religious practices and subjective religious beliefs or affects e.g., “How religious would you say you are?”). These types of variables seem to pertain, respectively, to Himmelfarb’s (1975) behavioral and ideational categories, Building on the work of Mindel and Vaughan (1978), Stephen C. Ainlay and D. Randall Smith (1984) separated nonorganizational religious involvement into a dimension measuring private religious behaviors and another more attitudinal dimension. Through confirmatory factor analysis, this model exhibited excellent overall fit and an internal structure invariant over several older age groups. This study succeeded in developing a multidimensional mea- sure of religious involvement that distinguished both between religious behaviors and atti- tudes and between public and private forms of religious involvement. This tradition of research is also notable for its specific focus on religious involvement per se in contrast to earlier work that sought to identify and measure the dozen or so possible dimensions of per- sonal religious expression. The empirical generalizability of this three-dimensional schema has not yet been verified. Ainlay and Smith’s (1984) analyses were limited by having been conducted on a sample of older Mennonites. In a recent study Chatters, Levin, and Taylor 1992), Ainlay and Smith’s (1984) three-dimensional latent factor structure was applied to religion variables in a subsam- ple of older adults from the NSBA. Whereas Krause and Tran (1989) confirmed a seven-item, two-dimensional scale for these data, Linda M. Chatters, Jeffrey S. Levin, and Robert Joseph Taylor (1992) were able to confirm a twelve-item, three-dimensional scale with factors termed “organizational,” “nonorganizational,” and “subjective” religiosity. This latter study was lim- ited to a subsample of older adults aged fifty-five and above N zyx   446), and these respondents were poorer, less educated, and more religious, on average, than the overall NSBA sample see Levin and Taylor 1993; Chatters and Taylor 1989). Therefore, one cannot safely infer that this three-dimensional model is generalizable to all black adults. The present study proposes to confirm this three-dimensional measure of religious involve- ment using the entire NSBA sample. This model will be compared to alternative one- and two-dimensional models and will be regressed on several sociodemographic variables be- lieved to be associated with measures of one or more of the component dimensions of reli- gious involvement. The product of these analyses, it is hoped, will be a theory-based, internally consistent, valid multidimensional factor model of religious involvement applicable  A zyxwvutsr ultidimensional Measure zyxwvu 61 z to the overall African-American population of the United States. It is also hoped that this model will serve as the basis for similar confirmatory scale development in other populations. METHODS The Study The National Survey of Black Americans is a nationally representative cross-sectional sur- vey of the adult, African-American population zyxw 1> zyx 8 years old) living in the continental U.S. The NSBA’s sample was drawn via multistage-area-probability sampling based on the 1970 United States Census for further details, see Jackson 1991). The NSBA is part of an ongoing series of research studies conducted by the Program for Research on Black Americans at the University of Michigan’s Institute for Social Research. At total of 2,107 respondents were interviewed in 1979 and 1980; the sample had an overall response rate of 68 percent. For the present analyses, there are 1,978 respondents available after listwise deletion of missing values. In this sample, the mean age of respondents is 42.9 years, 62.2 percent are female, 41.3 percent are married, 57.7 percent are employed, 53.7 percent live in the South and 78.9 percent in urban areas, and there is an average of 10.9 years of schooling and 8-1O,OOO in annual family income. As recommended for analyses of this type Liang, Levin, and Krause 1989), this sample was randomly split into two subsamples N, zyxwvuts   981, N2 = 997) for purposes of replication, in order to minimize the possibility of findings being due to chance Finifter 1972). All stages of data analysis were first conducted using only Sample 1; Sample 2 was reserved solely for replication of final results. In report- ing findings in the tables, however, results of all analyses are presented for both subsamples for purposes of comparison. Variables and Model Specification The Proposed Model Figure 1 represents the proposed model of three intercorrelated religious dimensions, along with seven exogenous constructs believed to predict one or more of these dimensions. This diagram is designed according to the conventional notation of the LISREL methodology used to conduct these analyses. In this model, the 11s etas) represent the latent endogenous con- structs underlying the observed endogenous indicators Ys), and the ’ (ksis) represent the latent exogenous constructs underlying the exogenous indicators Xs). The hs lambda-y) and z s lambda-x) coefficients are the endogenous and exogenous factor loadings, respectively, and the zyxwvu s epsilons) and 6s deltas) are their respective measurement-error terms. Five cor- related measurement errors 6,) are posited in 11, on the basis of their confirmation in the model of older black adults Chatters et al. 1992). Finally, the ys gammas) represent the regression coefficients linking endogenous and exogenous constructs, and the 5s zetas) and ys psis), respectively, are the residual-error and correlated-residual-error terms in the endog- enous constructs. The three latent endogenous constructs in the proposed model are termed organizational 11,). nonorganizational qz), and subjective (q3) eligiosity, and they are proposed to underly the twelve religious indicators included in the NSBA dataset. The NSBA does not contain any other variables assessing religious beliefs, values, or experiences, so secondary analyses are necessarily limited to the more narrowly conceived area of religious involvement. The NSBA items are typical of the types of religious indicators that have been used for many years
Search
Similar documents
View more...
Related Search
We Need Your Support
Thank you for visiting our website and your interest in our free products and services. We are nonprofit website to share and download documents. To the running of this website, we need your help to support us.

Thanks to everyone for your continued support.

No, Thanks