Donahue 1985

You might also like

Download as pdf or txt
Download as pdf or txt
You are on page 1of 20

Journal of Personality and Social Psychology Copyright 1985 by the American Psychological Association, Inc.

1985, Vol. 48, No. 2, 400-419 0022-35I4/85/S00.75

Intrinsic and Extrinsic Religiousness: Review and Meta-Analysis

Michael J. Donahue
Brigham Young University

The major findings of this meta-analytic review concerning intrinsic and extrinsic
religiousness are these: (a) Samples consisting of respondents with conservative
theological orientations seem more likely to display a negative correlation between
intrinsic and extrinsic religiousness than do others, (b) Extrinsic religiousness
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.

tends to be positively correlated with negatively evaluated characteristics, and


This document is copyrighted by the American Psychological Association or one of its allied publishers.

uncorrelated with measures of religious belief and commitment, (c) Intrinsic


religiousness tends to be uncorrelated with negatively evaluated characteristics,
and positively correlated with measures of religiousness, (d) A fourfold typology
based on median splits of the two scales is of little use when the dependent
variable is religious in nature, but with various nonreligious variables produces
results that may correspond to findings of curvilinearity observed with other
measures of religiousness. Recommendations concerning the use of the intrinsic
and extrinsic scales in future research are made. The article concludes with a
review of recent conceptual developments by Batson (1976) and Hood (1978).

No approach to religiousness has had McGaw, & Smith, 1981; Hunter, Schmidt, &
greater impact on the empirical psychology Jackson, 1982) to I-E research in an attempt
of religion than Gordon W. Allport's concepts to explain a number of seemingly inconsistent
of intrinsic (7) and extrinsic (E) religiousness findings.
(Meadow & Kahoe, 1984).' Nearly 70 pub-
lished studies have used Allport's Religious Early Conceptual Development
Orientation Scale (ROS), making it one of
Hunt and King (1971) reviewed the early
the most frequently used measures of reli-
giousness. conceptual history of 7 and E. In brief,
Allport distinguished between two types of
Research concerning / and E has been
religious sentiment: Intrinsic religiousness is
reviewed in three current psychology-of-reli-
religion as a meaning-endowing framework
gion textbooks (Batson & Ventis, 1982;
in terms of which all of life is understood; it
Meadow & Kahoe, 1984; Paloutzian, 1983).
is religion as proto-point (Rychlak, 1977).
Although these reviews are useful, they have
Extrinsic religiousness, in contrast, is the
not closely examined a number of relevant
religion of comfort and social convention, a
issues, such as the I-E correlation and the
self-serving, instrumental approach shaped to
I-E interaction. In addition to addressing
suit oneself. Table 1 is a summary of Allport's
these issues, in the present review I seek to
writings on these points.2 7 and E were
apply the techniques of meta-analysis (Glass,
probably best summed up by Allport and
Ross (1967) when they stated that "the ex-

This article is a revision and expansion of a review


that originally appeared in the author's doctoral disser-
1
tation, submitted at Purdue University. It was also pre- Throughout this review, the term religiosity is avoided.
sented at the meeting of the American Psychological The reason is etymological: "Religiosity" connotes an
Association, Anaheim, California, August 1983. affected, artificial, or exaggerated religious interest; "re-
Thanks are due to Richard Gorsuch, Allen Bergin, ligiousness" does not carry that conceptual baggage and
Richard Williams, Joseph Rychlak, Alice Eagly, Raymond is therefore more appropriate in the present context.
2
Paloutzian, Peter Benson, Kay Deaux. Richard Heslin, Table 1 is similar to a table presented by Hunt and
Hariey Bernbach, and an anonymous reviewer for their King (1971). However, Hunt and King examined / and
comments on earlier drafts. E in a predominantly sociological light and omitted such
Requests for reprints should be sent to Michael Don- characteristics as maturity from their analysis. They also
ahue, who is now at Search Institute, 122 West Franklin, omitted Allport's (1961, 1963, 1966a) references, which
Suite 215, Minneapolis, Minnesota 55404. stress the mental health aspects of the concepts.

400
I-E META-ANALYSIS 401

Table 1 ligious (henceforth nonreligious). Although


Concepts Associated with Intrinsic and Extrinsic Allport and Ross (1967) originally developed
Religiousness in Allport's Writings a scoring procedure that excluded the non-
Extrinsic
religious category (on the presumption that
Intrinsic
there would be a no nonreligious respondents
Relates to all of life Compartmentalized in their sample of church members), Hood
(a, b, c, d, f, g, h, j) (a, c, d, h) (1970) proposed classifying the four groups
Unprejudiced; tolerance Prejudiced; exclusionary
on the basis of median splits. This procedure
(a, b, c, h, i) (a, b, c, d, e, h)
Mature (a, d) Immature; dependent; has since been followed by the majority of
comfort; security researchers using the typology.
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.

(a, b, d, f, g, h, i, j)
This document is copyrighted by the American Psychological Association or one of its allied publishers.

Integrative; unifying; Instrumental; utilitarian;


Early Conceptual Critiques
meaning-endowing self-serving (a, c, d, e,
(a, c, d, f, g, h, i) f, g, h , i , j )
Regular church attendance Irregular church As the dates in Table 1 indicate, the con-
(e, g, h) attendance (e, g, h, i) cepts of intrinsic and extrinsic religiousness
Makes for mental health Defense or escape underwent a rather long development before
(f, g) mechanism (d, f, g) they were operationalized. When empirical
studies addressing the issues of / and £ began
Note. Letters in parentheses refer to the following refer-
ences: (a) Allport (1950), (b) AUport (1954), (c) Allport to be published, they were quickly followed
(1959), (d) Allport (1961), (e) Allport (1962), (f) Allport by two articles that, on the basis of rather
(1963), (g) Allport (1964), (h) Allport (1966a), (i) Allport meager evidence, offered a preliminary as-
(1966b), (j) Allport & Ross (1967).
sessment of the concepts.
The first, by Hunt and King (1971), con-
tended that E was a useful concept, but that
trinsically motivated person uses his religion, 7 was not because it was too "metaphysical"
whereas the intrinsically motivated lives his (p. 354). The other review, by Dittes (1971),
religion" (p. 434). which appeared in the same issue of the
As the descriptions in Table 1 indicate, Journal for the Scientific Study of Religion
these two types of religiousness were originally (JSSR), seems to have been based primarily
considered to be ends of a bipolar continuum. on purist, logical positivist views of science
But from the very beginning of the empirical and ad hominems against Allport. Dittes
research, doubt was cast on the appropriate- stated that the intrinsic-extrinsic concept in-
ness of characterizing / and E in this way. cluded a "heavy contraband load of value
Feagin (1964) reported a factor analysis in judgment that simply will not be sloughed
which items from / and E scales loaded on off" (p. 375). Dittes accused AUport of "ne-
separate, orthogonal factors. Allport (1966b) glecting the purity of conceptualization" (p.
himself began to take note of a group of 380) and accused I-E researchers of "persis-
"muddleheads who refuse to conform to our tently neglect[ing] or resisting]" the devel-
neat religious logic" (p. 6). These individuals opment of multidimensional frameworks (p.
agreed with items on both scales despite 381). The latter was manifestly not the case,
Allport's attempt to construct the scales to because in all three of the ROS studies up to
represent polar opposites. that time, Allport and Ross (1967), Feagin
As a result of these findings, Allport ex- (1964), and Hood (1970) had noted that /
panded his original bipolar approach into a and E were separate dimensions.
fourfold typology. Those who agreed with Dittes (1971) also accused Allport of con-
items on the / scale and disagreed with items ceptual agglutination: that is, creating a new
on the E scale he called intrinsics. Those concept out of a constellation of independent
who disagreed with / items and agreed with concepts. This criticism cannot be evaluated
E items he called extrinsics. Those who agreed at a purely conceptual level, because whether
with items on both scales he called indiscrim- a particular group of concepts does form a
inately proreligious (henceforth indiscrimi- syndrome is an empirical issue. One should
nate) and those who disagreed with items on be free to posit such syndromes, so long as
both scales he called indiscriminately antire- one is willing to proceed further and test
402 MICHAEL J. DONAHUE

their validity. At the time of Dittes' review, most quickly to mind, both Glass et al. (1981) and
there was not enough empirical evidence to Hunter et al. (1982) contended that it may not be
appropriate. The procedure adopted and reported here
address the issue.
(consistent with the recommendations of Hunter et al.)
Since the publication of these early reviews, was to weight the untransformed re by their sample size,
a considerable body ofl-E research has been sum them, and divide by the sum of the sample sizes.
produced. The purpose of this review is to Weighted z scores were also computed, and though
always greater, never differed by more than .03 from the
apply those data to the following questions:
means reported here.
1. What is the relation between / and E? A second issue involved the choice of how to calculate
What factors influence the correlation be- or estimate the product-moment correlation between /
tween them?3 and E when only contingency table data were reported.4
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.

2. What are the correlates of the two After consultation with several sources (Carroll, 1961;
This document is copyrighted by the American Psychological Association or one of its allied publishers.

Glass et al., 1981; Guilford, 1965; Nunnally, 1976;


constructs considered individually? What Taylor, 1972), both tetrachoric and phi coefficients were
types of variables do each of the two con- calculated. In light of comments by Guilford (1965, p.
structs form relations with, and what do these 300), tetrachoric correlations were computed only on
relations reveal concerning the nature of / samples of 200 or more, and, when used in computations,
were weighted by 45% of their sample size. Examination
and£?
of the marginal probabilities for the contingency tables
3. What is the utility of the Allport-Hood indicated that the corrections for tetrachoric correlations
fourfold typology? Does it increase predictive recommended by Jenkins (1955; see also Fishman, 1956)
power over the two unipolar constructs? were not required. When phi coefficients were used in
4. Have later conceptual developments calculations, they were weighted by their sample size. In
no case did the mean correlations involving phi coefficients
shed any further light on / and E? differ by more than .04 from those involving tetrachorics.

An Empirical Assessment Results

Method Relations Between I and E

Literature Search One way to address the relation between /


and E is through factor analysis of the items
The data to be presented here are based on a review
of the subscales. All five published reports of
of the literature published in English to the end of 1982.
It was conducted by means of (a) examining all volumes
such factor analyses, Carey (1974), Elifson
of JSSR, the Review of Religious Research, and the (1976), Feagin (1964), Patrick (1979), and
Journal of Psychology and Theology for relevant articles; Vincenzo, Hendrick, and Murray (1976), re-
(b) examining all the articles that the Social Science port similar findings: Items from the two
Citation Index listed as referencing Allport and Ross
subscales tend to load on two separate, or-
(1967) or Feagin (1964); and (c) the "ancestry" method:
examining the reference lists of the articles obtained by thogonal factors with few, if any, cross-load-
the two previous methods. ings. (See also Batson & Ventis, 1982,5 in
The general form of this review is that of a meta- which / and E tend to load on separate,
analysis (Glass et al., 1981; Hunter etal., 1982), in which
orthogonal factors.)
I seek to combine the results of independent studies in
order to make empirical determinations of the issues
generally addressed by literature reviews. In order to
facilitate combining the results of these studies, consid- 3
Because of the bipolar emphasis in Allport's concep-
eration was given only to studies in which / and E were
tualization, / was originally reverse-scored, so that agree-
defined in terms of either Allport and Ross's (1967) ROS,
ment with the items resulted in a low scale score. For
or Feagin's (1964) Intrinsic-Extrinsic (l/E) Religiosity
ease of consideration, the signs of all correlations reported
measure, which consists almost exclusively of a subset of
here have been adjusted so that a higher score indicates
the ROS. Related approaches, such as Allen and Spilka's
agreement for both scales.
(1967) "committed-consensual" dimensions and alter-
"In one case (Joe, McGee, & Dazey, 1977), exact
native conceptualizations of intrinsic (Hoge, 1972) or
reconstruction of the contingency table was based on
extrinsic (Wilson, 1960) religiousness, were not included.
reported degrees of freedom for Is in which the four cells
were compared. In two other cases (Alker & Gawin,
Analyses 1978; Dodrill, Bean, & Bostrom, 1973), information
allowed reconstruction of tables consistent with the data.
The analyses of these data required a number of 'Batson and Ventis (1982) presented 1-E data that
judgment calls (McGrath, Martin, & Kulka, 1982). The was based on a pooling of several studies. Because this
first involved how to calculate mean correlations across included otherwise unpublished data, the pooled data,
studies. Although Fisher's r to z transformation comes rather than the individual studies, have been used here.
l-E META-ANALYSIS 403

Another way to address the nature of the the 34 samples reported here is -.06. This
relation between / and E is to examine the mean, however, is strongly influenced by four
reported correlations between them. In Table studies in which either the Feagin I/E scale
2 the relevant data are presented, arranged or factor-analytically refined versions of the
from the most negative correlation to the ROS were used; three of these studies had
most positive. The mean correlation across rather large sample sizes. For ease of discus-

Table 2
Intrinsic-Extrinsic Correlations for Various Samples
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.

Source Sample
This document is copyrighted by the American Psychological Association or one of its allied publishers.

Dodrill et al., 1973 -.58' evangelical Protestant students 255


Bolt, 1977 -.57 conservative Protestants 62
Strickland & Shaffer, 1971 -.54 college students 24
Shoemaker & Bolt, 1977 -.45 conservative Protestant students 51
Spilka, 1977 -.41 '"affiliated and active Christians" 106
Batson, 1976 -.41 Princeton Seminary students 67
Paloutzian et al., 1978 -.40 adult Sunday school and nonacademic 177
college staff
Kahoe, 1974a -.37 religiously conservative students 333
Alker & Gawin, 1978 -.33" "religiously active" church members, many 101
denominations
Baker & Gorsuch, 1982 -.32 "religious wilderness camping organization" 52
Hood, 1971 -.31"" Baptist psychology students 83
Stewin & Anderson, 1974 -.29 eleventh-grade students 107
McConahay & Hough, 1973 -.24 Protestant graduate seminarians 159
Kahoe & Dunn, 1975 -.22 Catholics, Methodists, and Baptists 70
Allport & Ross, 1967 -.21 many denominations 309
Minion & Spilka, 1976 -.19 Protestant church members 67
Thompson, 1974 -.18" Catholic mothers 532
Spilka, Stout, Minton, & -.16 respondents personally involved in religion 167
Sizemore, 1977
Spilka et al., 1968 -.16 college students 146
Batson & Ventis, 1982 -.14 "undergraduates with at least a moderate 258
interest in religion"
Paloutzian et al., 1978 -.13 psychology students 84
Thompson, 1974 -.11' Catholic fathers 532
Tate& Miller, 1971 -.11" Methodist laity and clergy 175
Hood, 1971 -.08 Baptist students 83
Thompson, 1974 -.08' Catholic adolescents 532
Hoge & Carroll, 1973 -.04- Methodist and Presbyterians, Southern 343
sample
Elifson, 1976 -.01' Southern Baptist women 452
Hood, 1978 .00 psychology students 147
Hoge & Carroll, 1973 .01 At Methodists and Presbyterians, Northern 515
sample
Joeetal., 1977 .06b psychology students 167
Elifson, 1976 .09* Southern Baptist men 562
Patrick, 1979 .llf Buddhists, Baptists, and Congregationalists 91
Hood, 1970 .18 psychlogy students 89
Stewin, 1976 .21 eleventh-grade students 100
Hunt & King, 197 1« ,24f four denominations 1356

Mean r -.06 N 8271


Mean r for ROS studies -.20 Total ROS N 4952

Note. All studies used Allport ROS except as noted: ' Tetrachoric correlation. b Because of sample size, only phi
coefficient calculated.c Not included in the calculation of the mean; same sample as —.08 correlation below.d Used
Feagin //£".e Correlations based on stratified sample weighted to increase representativeness.f Scales refined through
factor analysis. • I-E correlation reported in Hunt & King, 1971; sample and methodology reported in King & Hunt,
1972.
404 MICHAEL J. DONAHUE

sion, only the 28 ROS correlations are con- means and medians are significantly corre-
sidered initially and the remaining studies lated with the I-E correlation across studies,
later. and in a way that is consistent with the
Across the ROS studies, the mean corre- hypothesis presented earlier: The more neg-
lation is —.20, which is interesting because ative the I-E correlation is, the higher is the
of its similarity to the —.21 originally reported mean / score, r(4) = -80, p < .05, and the
by Allport and Ross (1967). But more im- lower is the mean E score, r(4) = -.94, p <
portant is the considerable range of the cor- .01. Or, to put it another way, strongly negative
relations across these studies: from —.58 to I-E correlations are associated with samples
.24. Is this variation due to the operation of that are strongly intrinsic in terms of the
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.

some moderator variable, or is it simply theoretical ranges of the scales (9-45 for I;
This document is copyrighted by the American Psychological Association or one of its allied publishers.

sampling variance? Hunter et al. (1982) pre- 11-55 for E). This characteristic rapidly de-
sented equations to address this question. creases as the correlation declines. Thus it
Applying those equations to the ROS corre- seems quite possible that the population value
lations in Table 2, I found that the corrected of the I-E correlation is close to zero, rep-
(nonsampling) standard deviation for these resenting orthogonal constructs.
data was .15. This explanation is supported by the more
The next question is, What moderator broadly based samples. Hoge and Carroll
variable is involved? One possibility suggests (1973), using the Feagin (1964) //£, obtained
itself on examination of the third column in carefully matched samples of Methodist and
Table 2: the nature of the samples. The four Presbyterian churchgoers in Philadelphia and
studies desribed by the researchers as involv- Atlanta and observed a low I-E correlation,
ing evangelical or conservative religionists as did Thompson (1974) with a sample
have a mean of —.44, which is significantly of Catholics "distributed geographically
(p < .001) higher than the mean of the sam- throughout the United States and across so-
ples that are not so described (—.16). Simi- cioeconomic levels" (p. 473). Thus the I-E
larly, the samples that consist of respondents correlation may say as much about the sample
described as having some sort of religious on which it is based as it does about the
affiliation or interest in religion have a sig- relation between the concepts themselves.
nificantly (p < .001) more negative mean The conditions under which the correlation
I-E correlation (—.23) than those not so de- becomes markedly positive (e.g., King &
scribed (-.10). Hunt, 1972) are currently unclear.
The effect of using predominantly religious
respondents on the I-E correlation can be Correlates of I and E
considered in terms of the fourfold typology. Measures of religiousness. Perhaps the
Conservative, denominational colleges would most important way to determine the types
seem likely to have disproportionate percent- of religiousness that / and E are measuring
ages of intrinsics, because in such environ- is to relate them to other measures of reli-
ments, the importance of an intrinsic, per- giousness. When correlated with measures of
sonal religious orientation is constantly being Glock and Stark's (1966) religious belief scale,
made salient. Individuals without such a or measures of religious orthodoxy (usually
commitment would be less likely to be at- theologically conservative beliefs), / and E
tracted to such a setting. One way to test generally display markedly different relations.
such a hypothesis is to examine the I and E Across six reports (Batson, 1976; Batson &
means across studies. Because the publishing Ventis, 1982; Dodrill et al., 1973; Hoge &
of means is rather uncommon, and there Carroll, 1973; King & Hunt, 1972; Spilka,
have been several different scoring systems Pelligrini, & Dailey, 1968), intrinsic religious-
used, there were only five correlations for ness correlated .39 with such measures,6 and
which comparable means or medians were extrinsic religiousness correlated . 16. Perhaps
available. These are presented in Table 3.
The small sample size requires considerable 6
When Dodrill et al.'s (1973) sample, which consists
caution in interpreting these data. It is none- of evangelical students, is excluded, the mean correlation
theless interesting to note that the / and E between / and orthodoxy rises to .59.
l-E META-ANALYSIS 405

Table 3
Relation Between Intrinsic-Extrinsic Means and Medians and the I-E Correlation

Source r I

Bolt, 1977 -.57 36.18" 21.13


Shoemaker & Boll, 1977 -.45 36.51' 21.86
Thompson, 1974" -.18 34.30 29.30
Thompson, 1974 -.11 28.20 29.70
Thompson, 1974 -.08 29.50 36.70
Hood, 1978b .00 32.00 36.00
Correlations across studies -.75C 0.80" -0.94"
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.
This document is copyrighted by the American Psychological Association or one of its allied publishers.

* Calculated by reversing the scoring reported in the article.


b
Thompson's and Hood's data are medians.
e
Correlation between / and E means and medians shown here.
d
Correlation between I or E means and medians and the size of the I-E correlation.

more relevant, across four studies (Batson, udice across most available measures. E is
1976; Batson & Ventis, 1982; Jackson, 1981; positively correlated with prejudice, but not
Spilka et al., 1968), the correlation with the nearly so strongly as Allport's writings might
respondents' ratings of the importance of have predicted.
religion or religious commitment was .76 for One difficulty with the data reported in
7 and .03 for E. Table 4 is that all are correlations between
Such correlations might, at first glance, pencil-and-paper scales. The study by Batson,
call into question the validity of E as a Naifeh, & Pate (1978) deserves special atten-
measure of religiousness, because it correlates tion, however, because / and E were correlated
so poorly with other measures in the same with the respondent's self-reported willingness
domain. But remember that E measures not to be interviewed by a black or white inter-
so much religiousness per se as an attitude viewer. Although this was still, strictly speak-
toward religion. The items of the E scale call ing, a pencil-and-paper measure, it was one
for agreement with the treatment of religion with presumed behavioral consequences. Al-
as simply one of many influences on life or though the correlations obtained with 7 and
as a source of comfort and social support. It £ were not statistically significant given the
is therefore not surprising to find it basically sample size, they were positive, and indicated
uncorrelated with other measures of both an association between 7 and prejudice in a
belief and commitment. real-world setting.
Prejudice. Because the intrinsic-extrinsic However, Brannon (1970) reported an even
distinction originated in the study of prejudice more direct test of the behavioral correlates
(cf. Paloutzian, 1983), it is not surprising to of 7 and E. In a unique field study of a
find that in a number of studies researchers naturally occurring experiment, he obtained
have addressed their relation with that vari- ROS scores on a white congregation that had
able. In Table 4 the available data is sum- split in two over the issue of membership for
marized.7 The mean correlation across all blacks. Although Brannon unfortunately used
measures of prejudice for / is -.05, and .34 bipolar scoring, his results are relevant none-
for E. Selecting antiblack measures (anti- theless. Those who stayed in the integrated
Negro, antiblack, racial conservatism, and congregation were significantly ( p < .001)
symbolic racism) as a single index across all more intrinsically oriented than those who
studies (in order to preserve the statistical associated themselves with the segregated
independence of the correlations being con-
sidered), a similar relation is found. These
7
Gorsuch and McFarland (1972) also present "corre-
measures correlated -.09 with / and .28 with
lation ratios" that describe the relation between the I
E. Here we find a departure from Airport's and E scales and prejudice, but these are unsigned and
original conceptualization. / is uncorrelated, are not product-moment correlation coefficients, and so
rather than negatively correlated, with prej- are not included here.
406 MICHAEL J. DONAHUE

white congregation. Given the strength of the (1960) dogmatism scale. Across these five
effect and the continuing personal ramifica- studies, the correlation between / and dog-
tions of the decision involved, these findings matism ranges between .22 and .04, and
strongly support Allport's characterization of averages .06. The correlation between E and
intrinsics as unprejudiced.8 dogmatism ranges between .66 and .08, and
Dogmatism. Allport's approach seems to averages .36. In short, the same findings are
imply that / should be related to openmind- observed in this area as were observed con-
edness and E to closemindedness. In five cerning prejudice: Extrinsic religiousness
studies, Hoge and Carroll (1973), Kahoe shows a correlation in the predicted direction,
(1974a), Kahoe and Dunn (1975), Paloutzian, but intrinsic religiousness is basically uncor-
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.

Jackson, and Crandall (1978), and Thompson related with the measure.
This document is copyrighted by the American Psychological Association or one of its allied publishers.

(1974) have correlated / and E with Rokeach's A comment concerning these results seems
relevant. The absence of a correlation between
Table 4 / and dogmatism seems to call into question
Relation Between Intrinsic (\) and Extrinsic (E) the assertion by some that the intrinsic ori-
Religiousness and Various Measures of Prejudice entation is consistent with that of a "com-
pulsive, conforming, and unquestioning true
Source and measure / E believer" (Batson, 1976, p. 32). However,
Allport & Ross, 1967
Kahoe (1975, 1977a) examined the relation
(N = 309) between 7 and a concept related to dogma-
Antiblack nr .26 tism, that of authoritarianism (F). Although
Anti-Jewish nr .21 / was uncorrelated (.03) with F as reported
Anti-other" nr .32
by Kahoe (1974a), Kahoe (1977a) examined
CM! nr .44
Batson, 1976 (N = 42)
the relation between / and Krug's (1961)
Antiblack -.22 .24 dimensions of F. He found / uncorrelated
A nti-Jewish -.18 .34 with the subscales of cynicism, aggression,
Anti-other* -.19 .15 projectivity, and good versus bad people, but
CMI .16 .17
associated with conventionalism (.35) and
Batson eta!., 1978 (N = 51)
Antiblack -.36 .17 superstition and stereotypy (.31). Thus it may
Batson etal., 1978^ = 50) be that / is associated with portions of the
Preference for white over dogmatism concept, rather than the entire
black interviewer .19 .15
syndrome.
Feagin, 1964 (A1 = 286)
Ethnocentrism -.01 .35 Fear of death. According to Allport (1963),
Hoge & Carroll,* 1973, extrinsic religiousness "resembles a neurosis"
Northern sample and "is a defense against anxiety," whereas
(N= 515) intrinsic "makes for health" (p. 194). Because
Anti-Jewish .00 .30
fear of death tends to be correlated with
Antiblack -.15 .35
CMI .03 .49 neurotic preoccupations (Lester, 1967), the
Hoge & Carroll,11 1973, straightforward prediction is that E is posi-
Southern sample tively and 7 negatively correlated with fear of
(A' = 343)
death.
Anti-Jewish -.14 .35
Antiblack -.12 .24 The data in Table 5 give only partial
CMI -.06 .49 support to this analysis. The mean correlation
Kahoe, 1977b(A' = 142) between E and negative approaches toward
Racial conservatism .01 .18
McConahay & Hough, 1976
( f f = 153)
Symbolic racism ' Those who score at the intrinsic end of the bipolar
(antiblack) .24 nr continuum are high on the / scale and low on the £ scale,
Means (Total N = 1,891) -.05 .34 and are thus intrinsics as denned in the fourfold typology.
Those who score at the extrinsic end are extrinsics, and
Note, nr = not reported. CMI = Custodial Mental Illness nonreligious and indiscriminate individuals score in the
Ideology Scale. (Gilbert & Levinson, 1956). midrange (as with undifferentiated and androgynous re-
* Anti-other = anti-Oriental and anti-Hispanic. spondents in Bern's 1974 original, bipolar scoring of
" Used Feagin Intrinsic-Extrinsic Religiosity Measure. androgyny).
I-E META-ANALYSIS 407

death is .27. The mean correlation for /, considered in light of the fact that it correlated
however, is —.06. (Signs of correlations with -.62 with religious orthodoxy. When that
positive orientations toward death, e.g., after- study is excluded from the calculations, the
life of reward, were reversed to compute this mean fear of death correlations are -.17 for
mean.) Selecting only (a) fear of death and / and .30 for E. There is some evidence,
(b) death as pain and loneliness measures as therefore, that continuing research may in-
a single index across all studies produces dicate that fear of death is negatively corre-
—.06 for / and .30 for E. Examination of lated with 7 and positively correlated with E.
Table 5, however, indicates that the mean / Sex differences. The consensus of previous
correlation is due largely to the results of research has consistently been that women
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.

Spilka et al. (1968). Spilka (personal com- are more religious than men; indeed, Argyle
This document is copyrighted by the American Psychological Association or one of its allied publishers.

munication, April 1983) has commented that and Beit-Hallahmi (1975) referred to this as
the scales used in that study were early the "most important of the statistical com-
versions, different from the similarly named parisons" that they made (p. 71). It therefore
scales used in later research. The fact that might be informative to examine how the
afterlife of reward was so strongly negatively intrinsic-extrinsic literature addresses this
correlated with / in this study is perhaps best question.
Of the 67 I-E studies surveyed for the
present review, only 7 addressed the issue of
sex differences on the / and E scales. In 2,
Table 5
Alker and Gawin (1978), and Paloutzian et
Relation Between Intrinsic (\) and Extrinsic (E)
Religiousness and Various Measures
al. (1978) reported that there were no signif-
of Fear of Death icant sex differences on the scales used in
their research, which included / and E. In 3
Source and measure E other studies, Baither and Saltzberg (1978),
Strickland and Shaffer (1971), and Strickland
Bolt, 1977 (N = 62) and Weddell (1972) scored the I-E scale in
Death anxiety -.14 .29
Kahoe & Dunn, 1975 (N = 70)
a bipolar fashion and found that women were
Fear of death -.28 .10 more likely to score toward the intrinsic end.
Magni, 1972' (N = 53) Baither and Saltzberg also found that women
Negative death attitude -.38 .37 scored higher on the 7 scale, with no difference
Fear of death -.09 .27
on E. Thompson (1974) reported that the
Response latency to death-
related picture -.12 .22 mothers in his sample had significantly higher
Minton & Spilka, 1976 (AT = 67) medians for / than did the fathers, although
Death as pain and loneliness" -.15 .34 there was no difference for E. Spilka et al.
Forsaking dependents .03 .35 (1968) reported correlations with sex of -.26
Death as failure -.17 .41
Afterlife of reward .23 -.37
for / and -.15 for E, which presumably
Patrick, 1979 (N = 91) indicated association with femaleness. Only
Positive death outlook .31 .20 the 7 correlation was significant.
Negative death outlook -.23 .28 In short, there is evidence that women
Fear of death -.02 .33
score higher on 7 than men, but there is no
Spilka et al., 1968 (N = 146)
Death as pain and loneliness" .34 .34 evidence of a sex difference for E.
Forsaking dependents .13 .41 Other correlates. Several other patterns of
Death as failure .28 .40 results are present in the research. Intrinsic
Afterlife of reward -.52 .07 religiousness is negatively correlated, and ex-
Spilka et al, 1977 (N = 167)
Death as pain and loneliness -.26 .36
trinsic religiousness positively correlated, with
Forsaking dependents -.13 .31 trait anxiety (Baker & Gorsuch, 1982; Lovekin
Death as failure -.23 .49 & Malony, 1977). Internal locus of control
Afterlife of reward .37 -.07 correlates positively with 7 (Kahoe, 1974a;
Means (Total JV = 656) -.06 .27
Strickland & Shaffer, 1971; cf. Morris &
8
Used Feagin (1964) Intrinsic-Extrinsic Religiosity Mea- Hood, 1981), as does the related measure of
sure. purpose in life (Crandall & Rasmussen, 1975;
b
Mean of two scales later combined by Spilka et al. (1977). cf. Bolt, 1975). Perceived powerlessness is
408 MICHAEL J. DONAHUE

positively correlated with E and uncorrelated Given the powerful manipulation of poten-
with / (Minton & Spilka, 1976; Spilka & tial experimenter effects in these studies, the
Mullin, 1977). Feminism is apparently neg- null results observed seem rather meaningful
atively correlated with / (Kahoe, 1974b; (cf. Greenwald, 1975a). In addition, Stewin
McClain, 1979). and Anderson (1974) reported a nonsignifi-
Batson et al. (1978) found that / correlated cant correlation of —. 17 between 7 and agree-
.36 and E correlated .17 with social desir- ment response set, whereas the corresponding
ability. Batson et al. seemed impressed with correlation for E was .26. In light of these
this finding, noting that when the correlation findings it seems that the burden of proof
between social desirability and / was partialed has now shifted back to demonstrating (or
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.

out of the relation between prejudice and 7, replicating) a relation between social desir-
This document is copyrighted by the American Psychological Association or one of its allied publishers.

"the negative correlation between intrinsic ability and intrinsic religiousness.


religion and prejudice . . . diminished" (p. Summary. Considered as unipolar mea-
31; abstract). They went on to state that "the sures, research indicates that E tends to be
burden of proof now falls on anyone wishing positively correlated with variables such as
to contend that the relationship between in- prejudice and dogmatism, whereas 7, contrary
trinsic religion and reduced prejudice is not to the expectations derived from Airport's
an artifact of social desirability" (p. 40). conceptualization, is uncorrelated with such
More recent research evidence, however, measures. When the variable of interest is
calls Batson et al.'s (1978) conclusions into religious, however, available research indicates
question. Consider, for example, three studies that intrinsic religiousness tends to be posi-
reported by Hunsberger and Ennis (1982). In tively correlated, and extrinsic religiousness
the first, students were administered a battery uncorrelated.
of religion questions, including the ROS, by
an experimenter dressed either as a priest or
The AHport-Hood Fourfold Typology and
a layman. Despite the fact that 82% of the
the I-E Interaction
sample identified themselves as Roman Cath-
olic, there was only one significant difference The treatment of / and E as two separate,
in scores between the two conditions; respon- unipolar dimensions of religiousness was a
dents scored significantly (p < .01) lower on response to the finding of Allport and Ross
the / scale in the priest condition than in the (1967) that the two scales, rather than being
layman condition. There were no significant strongly negatively correlated, and thereby
effects on the 10 remaining scales. In a bipolar, were in fact largely uncorrelated.
second study that also included the ROS, no Those authors suggested, however, an alter-
differences were observed on any religiousness native approach of considering the two di-
scores as a function of whether the group mensions simultaneously in a fourfold clas-
sponsoring the questionnaire was Catholic, sification system of intrinsics, extrinsics, in-
Protestant, or secular. In a third study, par- discnmlnales, and nonreligious, as discussed
ticipants were interviewed by either "Reverend earlier.
Zeir" or "Professor Zeir." A control group This approach has been criticized by Kahoe
filled out the same questionnaire without (1976) on the ground that it deviates
being interviewed. Although the ROS was
unfortunately not included in the interview from normally preferred practice in psychology. . . .
While it may be useful at times to classify people into
format, several other religiousness measures
types on the basis of independent continua, it is concep-
were, including a measure of Christian ortho- tually sounder and more common to conceive and study
doxy (Fullerton & Hunsberger, 1982). The separate traits that vary independently of one another.
social desirability scale (Crowne & Marlowe, . . . Furthermore, while statistical interactions between
1964) was also included. Although there were intrinsic and extrinsic religious orientations might provide
some justification for constructing typologies from reli-
differences on the social desirability scale, gious orientations there was no published evidence of
with questionnaire participants scoring higher such interactions, (pp. 91, 92)
than interview participants, there were no
differences observed on any of the religious- It seems reasonable, then, to examine whether
ness measures. more recent research addressing the interac-
I-E META-ANALYSIS 409

tion between / and E has shed any further than Unitarians. Looking within the Baptist
light on this issue. sample, however, they found intrinsics signif-
When ROS respondents are classified into icantly less prejudiced than either indiscrim-
four groups according to median splits on inates or extrinsics. On dogmatism scores,
the subscales (after the fashion of androgyny they found the indiscriminates most dog-
research; e.g., Spence & Helmreich, 1978), matic, with no difference between extrinsics
and the dependent variable is a religious and intrinsics.
measure, the result is a main effect for /, no When intrinsics are compared with extrin-
main effect for E, and no interaction. This is sics, the intrinsics come out on the healthier
the general pattern of findings reported by sides of such measures as psychological well-
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.

(a) Hood (1970, 1978), concerning a measure being (Alker & Gawin, 1978), internal locus
This document is copyrighted by the American Psychological Association or one of its allied publishers.

of religious experience9 and the tendency for of control, and existential and trait anxiety
individuals to examine books with religious (Sturgeon & Hamley, 1979). McClain (1978)
or nonreligious titles; (b) Dodrill et al. (1973), found intrinsics higher than nonreligious re-
concerning measures of religious belief, prac- spondents on factors labeled Self-Corttrol,
tice, and experience; and (c) Tate and Miller Personal and Social Adequacy, and Stereo-
(1971), concerning religious values on the typed Femininity, and lower on Egocentric
Rokeach Value Survey. Sexuality and Restlessness (see also Wiebe &
When, on the other hand, the measure is Fleck, 1980).
nonreligious, a more interesting pattern These generalizations do not apply to all
emerges. Thompson (1974), for example, us- of the published research. Gibbs and Achter-
ing dogmatism scores, reported a pattern of berg-Lawlis (1978) reported that "there were
intrinsic = nonreligious < extrinsic < indis- no statistical relationships of import obtained
criminate. Joe et al. (1977) reported a one- for religous orientation" (p. 566), when they
way analysis of variance (ANOVA), across the used the ROS fourfold typology in a study of
four groups that indicated that intrinsics were fear of death. Kahoe (1974a) found positive
significantly less likely to devalue a rape associations, in the form of main effects,
victim than the other three groups. Inspection among /, E, and F.
of the t values that Joe et al. reported indicates In sum, for religious variables, intrinsic
that the ordering of the means was intrin- and indiscriminate individuals are generally
sic < nonreligious < extrinsic = indiscrim- indistinguishable, and both score higher than
inate. either the extrinsic or nonreligious individuals,
Allport and Ross (1976), as noted, used with no I-E interaction. There is an interest-
scoring procedures that produced only intrin- ing trend, however, among nonreligious vari-
sic, extrinsic, and indiscriminate categories. ables for intrinsics to score with nonreligiouses
At the level of means, the patterning of racist and for these two groups to score higher than
prejudice scores was the same as what extrinsics and indiscriminates. The differences
Thompson (1974) found for dogmatism among the four groups occasionally indicate
scores. At the level of individual denomina- a pattern of interaction between 7 and E, but
tional groups, however, only one showed that there has yet to be sufficient research using
actual pattern. Although the relative order of the entire fourfold typology and 2 X 2 ANOVA
the intrinsics and indiscriminates never designs to decide whether the variables inter-
changed, in three of the six groups extrinsics act. Thompson (1974), for example, reported
had the highest mean prejudice scores, data that display an interaction between /
whereas in two other groups they had the and E with dogmatism as the dependent
lowest. The interaction between religious type variable. I observed this interaction in a 2 X
and denomination did not, however, attain 2 (High-Low / X High-Low E) ANOVA for
the standard level of statistical significance. the fathers in his sample, but not for the
Strickland and Weddell (1972) used the
same scoring procedures as did Allport and
Ross (1967), but found no effect at all for
'Hood (1972, 1973; Hood & Morris, 1981) reported
religious type on prejudice scores; they re- similar findings in studies comparing only intrinsics and
ported that Baptists were more prejudiced extrinsics.
410 MICHAEL J. DONAHUE

mothers or the adolescents.10 Kahoe (1974a), Indications for Further Research


however, did not observe any interactions,
Joe et al. (1977) may have obtained an inter- Use of the present scales. The most im-
action with derogation of a rape victim as portant change needed in the present I-E
research involves standardization of proce-
the dependent variable, in light of the t values
they report, but they performed a one-way dures to allow greater comparability and rep-
licability of findings. Some of these changes
ANOVA, so the issue could not be further
would be rather minor, such as the abandon-
addressed without reanalysis of their original
data. ment of the discredited (Allport & Ross,
1967; Digenan & Murray, 1975; Hood, 1971)
The importance of the 1-E interaction,
bipolar scoring procedures, which are still
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.

and of the fourfold typology in general, stems


This document is copyrighted by the American Psychological Association or one of its allied publishers.

occasionally used (e.g., Brown & Annis, 1978;


from its relation to findings obtained with
Ernsberger & Manaster, 1981).
other measures of religiousness. Several in-
More important would be the standardiza-
vestigators have obtained curvilinear (in-
tion of the median splits used to form the
verted-U) relations between religiousness and
fourfold typology. Although these cutting
other variables such as prejudice (Gorsuch &
Aleshire, 1974), fear of death (Nelson & points have not often been reported, they
have undoubtedly varied from sample to
Cantrell, 1980), obedience in a Milgram
sample (see Table 3), producing inconsistent
(1974) paradigm (Bock & Warren, 1972),
classifications across studies. The best way to
and even unhappiness and physical and men-
address this issue would be for all researchers
tal illness (Shaver, Lenauer, & Sadd, 1980).
to use the theoretical midpoints of the scales
Moreover, these effects seem quite robust.
(27 for /, 33 for E) to classify the respondents
Gorsuch and Aleshire's findings were based
into the four groups. Such an approach is
on a review and were noted in 20 of 25
not without its drawbacks. For example, when
studies in which the published statistics
tested with the theoretical median criteria,
allowed the assessment of curvilinearity,
some samples of students that I obtained at
whereas Shaver et al. sampled from the mas-
Brigham Young University (and, no doubt,
sive data base of a Redbook survey.
samples at other conservative denominational
Simple correlational procedures with the /
colleges, e.g., that of Dodrill et al., 1973)
and E scales used independently would ob-
were 90% intrinsic. Such circumstances result
scure curvilinearity. But when a pattern of
in the loss of the powerful 2 X 2 ANOVA
intrinsic = nonreligious < extrinsic < indis^
design because of empty nonreligious cells,
criminate is observed, as in the case of
and the attendant loss of the ability to ex-
Thompson's (1974) large, broadly based sam-
amine interactions. However, the advantages
ple, this can be considered replication of
of standardization in terms of generalizability
curvilinearity. Both the intrinsics and the
across samples, and the impetus to individual
nonreligious show low dogmatism, with the
researchers to obtain more heterogeneous
occasionally attending extrinsics and the ap-
samples, would result in a net gain for the
parently confused indiscriminates, who rep-
field.
resent intermediate levels of religious com-
Refinement of the present scales. There
mitment, scoring higher.
are four major concerns with the present
It must be emphasized that this interpre-
scales that may call for further refinement.
tation is tentative. So far there has only been
The first is that Hood (1971) reported that
one published study in which the researcher
the Feagin (1964) short versions of the scales
used the fourfold typology, nonreligious de-
are at least as reliable as Allport's forms. The
pendent variables, and an appropriate 2 X 2
Allport and Feagin / scales were quite highly
between-groups ANOVA: Kahoe, 1974a (see
correlated (.90), which was not unexpected,
also Kahoe, 1975, which did not obtain any
because the Feagin scale consists of six of
significant interactions). The interaction find-
ings concerning other variables presented here
have been based on reanalysis of published 10
These findings concerning Thompson's data, as well
data. But the initial results appear promising, as those discussed earlier concerning Hood (1978) and
and it seems appropriate to call for research Dodrill et al. (1973), are based on ANOVAS that I performed
investigating this possibility. on data presented in the relevant articles.
I-E META-ANALYSIS 411

Allport's nine items. The E scales, however, fications of range restriction in I-E research,
were less well correlated (.61) despite the fact see the following discussion of the quest
that Feagin's scale contains five of Allport's research.
eleven items, plus a sixth that Allport did A fourth refinement of the scales, which
not use. This raises the possibility that one has already been empirically addressed, is
of the two E scales may have greater validity, the development of an "age universal" version
a question that should be empirically ad- of the ROS. Gorsuch and Venable (1983)
dressed. (For further information concerning produced such a scale, which should prove a
scale reliabilities and item-total correlations, powerful tool for those who wish to examine
see Robinson & Shaver, 1973.) a whole range of fascinating developmental
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.

The second issue is the question of response questions in this area.


This document is copyrighted by the American Psychological Association or one of its allied publishers.

set that Kahoe (1974a, 1976) noted. None of


the items on either scale is reverse-keyed, Later Conceptual Developments
leading to concerns over acquiescence effects,
at least in the case of indiscriminates. A In addition to measurement issues, there
revision of the scales that includes some is considerable need for further conceptual
reverse-scored items seems appropriate. development concerning / and E. Allport's
Third, there is need for the development approach, as summarized in Table 1, was left
of a version of the scales that would allow in some disarray when the empirical evidence
easy administration to nonreligious individ- did not support his bipolar orientation. His
uals. My experience is that research partici- initial work on a two-dimensional framework
pants who have no religious orientation report was cut short by his death in 1967. Since
difficulty responding to items concerning how then, there have been two major conceptual
religion functions in their lives. As a result, approaches to / and E. One of these, proposed
they may tend to leave such items blank. The by Hood (1978), is basically an elaboration
standard instructions for the ROS indicate of the original bipolar approach proposed by
that they should do this "if no choice is Allport. The other however, pursued by Bat-
possible," and that these items should be son and his colleagues, constitutes a major
scored as a neutral response (3 on a 5-point challenge to the validity of the entire I-E
scale). This procedure may produce invalid understanding. If the criticisms offered by
I-E scores in populations with a significant Batson are valid, most of the forgoing discus-
nonreligious subgroup. sion becomes moot. It seems reasonable
An alternative view on this issue would be therefore to consider both Batson's and
that there really is no need for such a "non- Hood's conceptual approaches at some length.
religious" version of the ROS, because re-
Batson: Different Ways of Being Religious
searchers of religious orientation should con-
cern themselves primarily with religious in- Batson (1976; Batson & Gray, 1981; Batson
dividuals. For example, Dittes (1969) argued et al., 1978; Batson & Ventis, 1982; Darley
that "active and committed persons are likely & Batson, 1973) proposed an alternative to
to make more subtle distinctions" (p. 610) the intrinsic-extrinsic conceptualization.
in research findings. But if the concern is to Batson's major dissatisfaction with the ap-
examine the relation between religious ori- proach was that he felt that Allport gradually
entation and various variables of social psy- wandered away from the richness of his orig-
chological interest, then limiting consideration inal distinction between mature and imma-
to religious respondents seems to be little ture religion, reducing / to simple "single-
more than embracing range restriction and mindedness":
thereby diluting potential findings. The effect
Single-mindedness and centratity were pan of Allport's
of such range restriction can be considerable. conception of mature religion, but they were not all.
For example, Kahoe and Meadow (1981) . . . First, Allport suggested that the mature religious
report that "for a state university sample sentiment is integrative. . . . Second, mature religion
involves a readiness to doubt and to be self-critical: "The
intrinsic religion and [religious] observance
mature religious sentiment is ordinarily fashioned in the
correlated .48 (significant beyond the .001 workshop of doubt...." Third, there is an emphasis on
level), but for a Baptist college sample the incompleteness and tentativeness. . . . (Batson & Ventis,
correlation was .04" (p. 11). For other rami- 1982, p. 149)
412 MICHAEL J. DONAHUE

In an attempt to address these perceived (Darley & Batson, 1973). With a sample of
shortcomings, Batson constructed the Reli- college students (Batson et al., 1978), he
gious Life Inventory (RLI; Darley & Batson, found prejudice negatively correlated with
1973) which consists of three subscales: ex- both / and Q. After he applied controls for
ternal, which measures the importance of social desirability, however, only the Q cor-
others (churches, clergy, parents) in one's relation remained significant. Batson and
religious development; internal, which mea- Gray (1981) found that Q formed a significant
sures one's need for religion, and the inter- positive correlation with willingness to help
actional or guest (Q) scale. Batson presents an individual who stated she needed help,
the Q scale as an alternative to the / scale; and a negative relation when the individual
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.

indeed, in the first published reference to the stated she preferred to solve the problems by
This document is copyrighted by the American Psychological Association or one of its allied publishers.

Q scale (Darley & Batson, 1973) he called it herself. The / scale formed no significant
"the intrinsic scale" of the RLI. This scale is relations in either of the latter two studies,
designed to capture the religious sentiment and only the correlation between E and not
of those who, though helping was significant. Last, Batson and Ven-
tis (1982) presented data (also published by
not necessarily aligned with any formal religious insti-
tution or creed, . . . are continually raising ultimate
Batson & Raynor-Prince, 1983) indicating
"whys," both about the existing social structure and that / and E are uncorrelated, and Q is
about the structure of life itself. While it may seem positively correlated, with complexity of
strange to call such an individual religious, there is thoughts about religious concerns. As a result
actually a long history of such a view. It goes at least as
of these findings, Batson and his colleagues
far back in Western thought as the Hebrew prophets and
much further back in Eastern religions. (Batson, 1976, presented a picture of the quest orientation
P. 32) as being less dogmatic, less prejudiced, and
more responsive to the true needs of others
In its final version, the Q scale consists of six than intrinsic religiousness.
items emphasizing doubt ("Questions are far It initially appears that Batson did a better
more central to my religious experience than job at operationalizing Allport's concepts than
are answers") and personal development ("My Allport himself. But this alternative approach,
religious development has emerged out of my and the research associated with it, has a
growing sense of personal identity"). number of difficulties, both conceptual and
In research involving this scale, respondents methodological.
are administered the ROS, the RLI, and a
doctrinal orthodoxy scale based on that pre- Conceptual Difficulties
sented by Glock and Stark (1966). Scores on
each of these scales are then standardized Three difficulties with Q arise from the
and intercorrelated. The resulting intercor- "conceptual genealogy" that Batson claims
relation matrix is then factor analyzed. Across for his approach. As we have seen, he con-
the five Batson studies cited earlier, these tended that his approach is consistent with
analyses have consistently produced three (a) prior religious tradition, (b) Allport's writ-
factors: a means factor with a primary loading ings, and (c) an underlying current in the
for E; an end factor with primary loadings nature of religion. Each of these are examined
for /, external, internal, and orthodoxy; in turn.
and a quest factor defined almost exclusively Religious tradition. In the quotation pre-
byQ. sented earlier, Batson contended that individ-
Batson regularly presented the correlations uals who have a quest orientation are not
of the dependent variable with both the factor necessarily identified with any particular tra-
scores and the individual scales. The results dition but are constantly raising "whys" about
for the means, end, and quest factors have the social structure and life. He stated that
been virtually the same as those for the E, I, this is reflected both in the writings of the
and Q scales, respectively. In a sample of Hebrew prophets and in Eastern religions.
seminarians, Batson (1976) found no consis- Because the reference to Eastern religions is
tent relation between any of the scales and vague, it cannot be addressed by reference to
either prejudice or willingness to aid an in- any particular set of writings or commentar-
dividual in a "Good Samaritan" paradigm ies. There seems, however, to be little evidence
/-£ META-ANALYS1S 413

to support Batson's contention that the He- mature religionist are, according to Allport,
brew prophets would be examples of a quest simply the fires in which belief is tempered.
orientation to religion, at least as he opera- Nature of religiousness. All of what has
tionalized it. The prophets were "necessarily been said to now does not indicate that Q is
aligned" with a formal creed; they were pas- not a measure of religiousness, only that it
sionate servants of Yahweh. They were less seems unlikely that this orientation is actually
likely to "raise ultimate 'whys' " about the supported by some of the sources that Batson
social order than to deliver "what fors," con- offered. But there is an additional conceptual
demning idolatry, injustice, and reliance on difficulty with his approach, in that the con-
military might. They did not spend much of struct validity of his measure has never been
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.

their time questing for an answer; answers empirically demonstrated. The Q scale has
This document is copyrighted by the American Psychological Association or one of its allied publishers.

were central to their message. And perhaps consistently failed to correlate with any mea-
most to the point is that they did not present sure of religiousness. Depending on one's
their message as part of their personal devel- predilection, this might be considered evi-
opment, but because they understood them- dence for discriminant validity or a lack of
selves to be messengers of God (nabi'im). convergent validity (Campbell & Fiske, 1959);
Their words reflected their direct experience but until some group of individuals reasonably
of the divine (Heschel, 1962). This is a far identifiable as religious can be demonstrated
cry from the identity-crisis picture that the to have higher Q scores than another group,
Q scale presents. it seems invalid to call this a measure of
Allport's writings. Do Allport's writings religiousness.
in fact support the picture of intrinsic religion What, then, is Q measuring? An exami-
that Batson contended they do? Recall that nation of the items of the Q scale, such as
Batson criticized the 7 scale as being more those quoted earlier, indicates that it might
concerned with religion as master motive best be characterized as an agnosticism scale,
than was originally the case in Allport's writ- as suggested by Harley Bernbach (personal
ings. But consider the following quote, and communication, July 1981). Certainly persons
especially the emphasis that Allport (1950) with mature and differentiated religious ori-
himself chose to place in it: entations might agree with such items, but
I venture to assert that the most important of all distinc-
so might iconoclasts who sophomorically and
tions between the immature and the mature religious reflexively respond "why" to every answer
sentiment lies in this basic difference in their dynamic given. Means, end, and quest are certainly
characters. . . . Mature religion is less of a servant, and three separate, orthogonal, replicable dimen-
more of a master, in the economy of the life. No longer
sions, but they may not all be dimensions of
goaded and steered exclusively by impulse, fear, wish, it
tends rather to control and to direct these motives toward religiousness.
a goal that is no longer determined by mere self-interest,
(p. 72) Methodological Difficulties
What then has happended to the "workshop Even if one allows Q to be a dimension of
of doubt" in the face of this master motive? religiousness by defining religiousness as an
Perhaps this is best explained by the entire existential concern (which is the approach
paragraph in which that expression occurs: taken by Batson & Ventis, 1982), the research
We may then say that the mature religious sentiment is
methodology that Batson used would require
ordinarily fashioned in the workshop of doubt. Though further caution. Three concerns relate to
it has known intimately "the dark night of the soul" it sampling. First, the sample size for his studies
has decided that theoretical skepticism is not incompatible has averaged about 50, which is rather meager
with practical absolutism. While it knows all the grounds
when one of the thrusts of the discussion
for skepticism, it serenely affirms its wager. In so doing,
it finds that the successive acts of commitment, with
their beneficent consequences, slowly strengthen the faith
and cause the moments of doubt gradually to disappear,
" Pagination for these quotations come from the 13th
(p. 83)"
paperback printing (1974) of The Individual and His
Religion: A Psychological Interpretation (pp. 72, 83) by
Thus, unlike the individual operating G. W. Allport, (1950). Copyright 1950 by Macmillan
within a quest orientation, for whom doubts Publishing Co., Inc., renewed 1978 by Robert B. Allport.
are almost a master motive, doubts for the Both quotations adapted by permission.
414 MICHAEL J. DONAHUE

sections has been an acceptance of null hy- means is basically E, end is basically / and
potheses concerning / (cf. Greenwald, 1975b). its intercorrelates, and quest is almost exclu-
Second, much of his research is done with sively Q. Why, then, constantly submit the
college students. Although accusations against scales to sophisticated analysis procedures
the repesentativeness of "sophomore psychol- when it appears that the individual scales are
ogy" are too well known to require reference, adequate and much more replicable measures
the inappropriateness of that particular pop- of the three constructs?
ulation to determine the validity of a religion The larger concern over Batson's analyses
scale involving existential questioning is at is the use of correlational procedures. It was
least worth comment. Researchers addressing noted earlier that a number of variables have
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.

religious development have consistently noted been found to form curvilinear relations with
This document is copyrighted by the American Psychological Association or one of its allied publishers.

the large degree of religious questioning and religiousness. The consistent use of correla-
disaffection that occurs during the college tional analyses in the Q research obscures
years in comparison with the time periods such relations. Perhaps the most intriguing
preceding and following it (see Strommen, possibility is that Q and / might be curvilin-
1971). The correlations between religious early related, but this must await further
questioning and other variables in college research.
students may not be generalizable to other All of this having been said, it is necessary
age groups. to clarify that Batson has, in fact, made an
Third, Batson and his colleagues consis- important contribution to psychology of re-
tently described their samples as being com- ligion through his quest concept. The necessity
posed of individuals with "at least a moderate for constant spiritual questing and growth is
interest in religion." Batson and Raynor- central to what is best in all religious traditions
Prince (1983) operationalized this criterion and has been the hallmark of such twentieth
by stating that "interest was assessed by century religious luminaries as Thomas Mer-
subjects' reponses to the question, 'How in- ton (Furlong, 1980) and others. This ideal of
terested in religion are you?' (1 = not at all, continuing growth is addressed only tangen-
9 = extremely). Only those marking a 4 or tially in the present / scale. But although
higher were included in the sample" (p. 42). questioning is central to such growth (as
The problem with such a procedure is that noted by Allport in the "doubt" quote),
Gorsuch and McFarland (1972) demonstrated calling such questioning doubt may be an
that / is highly correlated with such single- inappropriate label. This initial version of the
item measures of religiousness. By truncating Q scale has raised a central, unaddressed
the distribution of religious interest, Batson issue in psychology of religion, and has pro-
no doubt truncated the range of / scores in duced provocative data. The next step will
his sample. At the same time, because both involve refining the construct and using an-
E and Q are largely uncorrelated with /, he alytical methodologies that allow for the ex-
was not truncating their range. His results amination of the effects of interaction between
were thus influenced by the effects of range Q, I, and E, and curvilinear relations among
restriction, lowering the resulting correlations. them or with the dependent variables being
This may explain why he so rarely reported studied.
significant correlations between 7 and his
dependent variables. Hood: Managing the Religious Sentiment
Two other shortcomings relate to the ana-
lytical method applied in Batson's research. Hood's (1970, 1971, 1973, 1978; Hood &
The first is the consistent use of factor scores. Morris, 1981; Morris & Hood, 1981) pro-
Such scores rely on the intercorrelations grammatic I-E research has led to an ap-
among the variables, which change from proach that is more directly related to the
sample to sample, thus reducing the repiica- original, bipolar understanding. According to
bility of the research. This is all the more this approach, one must manage the intrinsic
curious considering that examination of the and extrinsic orientations by accepting reli-
intercorrelation matrices and factor loadings gion as either part of life (extrinsic) or as the
that Batson has published indicates that meaning of life (intrinsic). Hood contends
I-E META-ANALYSIS 415

that both the indiscriminates and the nonre- Table 6


ligious are experiencing difficulty either in Incidence of Stress, Chi-Square Values, and
integrating one of these two approaches or in Stress Probabilities, From Hood (1978)
affirming a nonreligious stand.
Incidence Stress
As one test of this approach, Hood (1978) Group N of stress X2 probability
reported data concerning the likelihood that
an individual would display stress while dis- Indiscriminate 46 31 4.89 .674'
cussing whether he or she had had a religious Nonreligious 31 18 0.52 .581'
Extrinsic 19 8 12.41 .205"
experience. In analyzing these data, he applied Intrinsic 31 3 18.58 .097"
chi-square analyses to the incidence of stress
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.

for each of the four groups individually (see Note. Probabilities with different superscripts are signifi-
This document is copyrighted by the American Psychological Association or one of its allied publishers.

Table 6). The difficulty with such an analysis cantly different (z > 3.2). From "The Usefulness of the
Indiscriminately Pro and Anti Categories of Religious
is that the base-rate probability for displaying
Orientation" by R. W. Hood, Jr., 1978, Journal for the
stress while discussing religious experiences Scientific Study of Religion, 17, p. 427. Copyright by the
is assumed to be .50. In the absence of Society for the Scientific Study of Religion, Adapted by
previous data concerning this issue, and in permission.
light of the fact that Hood's major concern
is a comparison between the four groups, a cases, these individuals marked Salvation number one
more appropriate analysis might be to test and then failed to complete the other rankings, (p. 364)

the differences between the probabilities of Such findings support not only the label
showing stress for each of the four groups. I indiscriminate applied to such respondents,
performed these tests, and they indicated that but the larger picture of value confusion that
the indiscriminate and nonreligious respon- both Hood and Allport associate with this
dents are alike and they differ from the group.
intrinsic and extrinsic groups in displaying
stress in such settings. Interestingly, this find- Conclusions
ing is also more consistent with Hood's (1978)
hypothesis. Intrinsic religiousness serves as an excellent
Consistent with Hood's finding of greater measure of religious commitment, as distinct
stress for indiscriminates in religious contexts from religious belief, church membership,
is Vincenzo et al.'s (1976) finding that re- liberal-conservative theological orientation,
spondents who scored high on either an /- or and related measures (see Wimberley, 1974).
.E-related dimension were less likely to have Its lack of doctrinal content and open-ended
reported seeing "the fear-provoking religiously definition of religion makes it usable with
oriented movie—The Exorcist." Presumably, virtually any Christian denomination, and
those who score high on both factors (the perhaps even with non-Christian religions
indiscriminates) would be least likely to have (Patrick, 1979). It correlates well with other
seen it. measures of religiousness, and even such
On a more anecdotal level, Tate and Miller behaviors as preference for religious over
(1971) had some interesting experiences while nonreligious books (Hood, 1978). But by
administering the Rokeach Value Survey itself it seems to correlate with little besides
(Rokeach, 1973) to each of the four groups. other measures of religiousness. Aside from
This instrument requires respondents to rank- internal locus of control (Kahoe, 1974a),
order 18 "instrumental values" (e.g., ambi- purpose in life (Crandall & Rasmussen, 1975),
tious, forgiving, intellectual, self-controlled) and lack of anxiety (Baker & Gorsuch, 1982),
and 18 "terminal values" (e.g., comfortable it has not yet been found to be predictive of
life, freedom, salvation, true friendship) in much that is of interest to personality-social
terms of their importance in the respondent's psychologists, although maybe this is because
life. They found that two intrinsics it has not yet been widely used as a measure
of religiousness in such research. Given the
indicated they could not complete the instrument because
they could not separate the values from each other. By opportunity, it would probably be a strong
contrast, 12 [indiscriminates] returned their value mea- (negative) predictor of such variables as pre-
sures either incompletely or improperly marked. In several marital intercourse (Davidson & Leslie, 1977)
416 MICHAEL J. DONAHUE

and initial drug abuse (Gorsuch & Butler, Allen, R. O., & Spilka, B. (1967). Committed and
1976). consensual religion: A specification of religion-prejudice
relationships. Journal for the Scientific Study of Reli-
Extrinsic religiousness on the other hand,
gion, 6, 191-206.
does a good job of measuring the sort of Allport, G. W. (1950). The individual and his religion: A
religion that gives religion a bad name. It is psychological interpretation. New York: Macmillan.
positively correlated with prejudice, dogma- Allport, G. W. (1954). The nature of prejudice. Cambridge,
MA: Addison-Wesley.
tism (Hoge & Carroll, 1973), trait anxiety
Allport, G. W. (1959). Religion and prejudice. Crane
(Baker & Gorsuch, 1982), and fear of death Review, 2, 1-10.
(Minton & Spilka, 1976) and is apparently Allport, G. W. (1961). Pattern and growth in personality.
uncorrelated with altruism (Batson & Gray, New York: Holt, Rineharl & Winston.
1981; Benson et al., 1980). Allport, G. W. (1962). Prejudice: Is it societal or personal?
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.

Journal of Social Issues, 18(2), 120-134.


This document is copyrighted by the American Psychological Association or one of its allied publishers.

Considering these two measures simulta- Allport, G. W. (1963). Behavioral science, religion, and
neously produces considerably greater ex- mental health. Journal of Religion and Health, 2, 187-
planatory power, however. Such variables as 197.
prejudice (Allport & Ross, 1967), dogmatism Allport, G. W. (1964). Mental health: A generic attitude.
Journal of Religion and Health, 4, 7-21.
(Thompson, 1974), derogation of an innocent
Allport, G. W. (1966a). The religious context of prejudice.
victim (Joe et al., 1977), and a wide array of Journal for the Scientific Study of Religion, 5, 447-
personality characteristics (McClain, 1978; 457.
Wiebe & Fleck, 1980) have been shown to Allport, G. W. (1966b). Traits revisited. American Psy-
display strong and meaningful relations when chologist, 21, 1-10.
Allport, G. W., & Ross, J. M. (1967). Personal religious
considered in the context of either the entire orientation and prejudice. Journal of Personality and
fourfold typology or two or more of its cells. Social Psychology, 5, 432-443.
Such an approach also casts considerable Argyle, M., & Beit-Hallahmi, B. (1975). The social
light on the results obtained in the unipolar psychology of religion. Boston: Routledge & Kegan
Paul.
research. Intrinsic religiousness fails to cor-
Baither, R. C, & Saltzberg, L. (1978). Relationship
relate with nonreligious variables because it between religious attitude and rational thinking. Psy-
confounds two distinct orientations. People chological Reports, 43, 853-854.
scoring high on / may either be intrinsic (low Baker, M., & Gorsuch, R. (1982). Trait anxiety and
in prejudice and dogmatism), or display the intrinsic-extrinsic religiousness. Journal for the Sci-
entific Study of Religion, 21, 119-122.
unsettled orientation of the indiscriminate Batson, C. D. (1976). Religion as prosocial: Agent or
(high on both characteristics). Those scoring double agent? Journal for the Scientific Study of Reli-
low on / may be either nonreligious (low in gion, 15, 29-45.
dogmatism; no data for prejudice) or extrinsic Batson, C. D., & Gray, R. A. (1981). Religious orientation
and helping behavior: Responding to one's own or to
(second highest on prejudice and dogmatism).
the victim's needs? Journal of Personality and Social
It is not surprising, therefore, that a statistic Psychology, 40, 511-520.
with which one attempts to measure a linear Batson, C. D., Naifeh, S. J., & Pate, S. (1978). Social
relation between 7 and prejudice or dogma- desirability, religious orientation, and racial prejudice.
Journal for the Scientific Study of Religion, 17, 31-
tism shows little or no relation. The fourfold
41.
approach also explains the findings observed Batson, C. D., & Raynor-Prince, L. (1983). Religious
with E: High scorers are either extrinsic or orientation and complexity of thought about existential
indiscriminate and low scorers are either concerns. Journal for the Scientific Study of Religion,
intrinsic or nonreligious, hence the positive 22, 38-50.
Batson, C. D., & Ventis, W. L. (1982). The religious
correlation with prejudice and dogmatism.
experience: A social-psychological perspective. New
Although, as was noted earlier, this research York: Oxford University Press.
is at an initial stage, the findings currently Bern, S. L. (1974). The measurement of psychological
available bode well for the potential of the /- androgyny. Journal of Consulting and Clinical Psy-
E framework as a powerful explanatory tool chology, 42, 155-162.
Benson, P. L., Dehority, J., Carman, L., Hanson, E.,
in personality-social psychology. Hochschwender, M., Lebold, C., Rohr, R., & Sullivan,
J. (1980). Intrapersonal correlates of nonspontaneous
References helping behavior. Journal of Social Psychology, 110,
87-95.
Alker, H. A., & Gawin, F. (1978). On the intrapsychic Bock, D. C., & Warren, N. C. (1972). Religious belief as
specificity of happiness. Journal of Personality, 46, a factor in obedience to destructive commands. Review
311-322. of Religious Research, 13, 185-191.
I-E META-ANALYSIS 417

Bolt, M. (1975). Purpose in Life and religious orientation. values and death anxiety: Implications for counseling
Journal of Psychology and Theology. 3, 116-118. with terminal cancer patients. Journal of Counseling
Boh, M. (1977). Religious orientation and death fears. Psychology, 25, 563-569.
Review of Religious Research. 19. 73-76. Gilbert, D. C., & Levinson, D. J. (1956). Ideology,
Brannon, R. C. L. (1970, April). Gimme that old-time personality, and institutional policy in the mental
racism. Psychology Today. 42-44. hospital. Journal of Abnormal and Social Psychology,
Brown, D. M., & Annis, L. (1978). Moral development 53, 263-271.
level and religious behavior. Psychological Reports, 43, Glass, G. V., McGaw, B., & Smith, M. L. (1981). Mela-
1230. analysis in social research. Beverly Hills, CA: Sage.
Campbell, D. T., & Fiske, D. W. (1959). Convergent and Clock, C. Y., & Stark, R. (1966). Christian beliefs and
discriminant validation by the multitrait-multimethod anti-Semitism. New York: Harper & Row.
matrix. Psychological Bulletin, 56. 81-105. Gorsuch, R. L., & Aleshire, D. (1974). Christian faith
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.

Carey, R. G. (1974). Emotional adjustment in terminal and ethnic prejudice: A review and interpretation of
This document is copyrighted by the American Psychological Association or one of its allied publishers.

patients: A quantitative approach. Journal of Counseling research. Journal for the Scientific Study of Religion,
Psychology, 21, 433-439. 13, 281-307.
Carroll, J. B. (1961). The nature of the data, or how to Gorsuch, R. L., & Butler, M. C. (1976). Initial drug
choose a correlation coefficient. Psychometrika, 26, abuse: A review of predisposing social psychological
347-372. factors. Psychological Bulletin, S3, 120-137.
Crandall, J. E., & Rasmussen, R. D. (1975). Purpose in Gorsuch, R. L., & McFarland, S. G. (1972). Single vs.
Life as related to specific values. Journal of Clinical multiple-item scales for measuring religious values.
Psychology. 31, 483-485. Journal for the Scientific Study of Religion, 11,
Crowne, D. P., & Marlowe, D. (1964). The approval 53-64.
motive: Studies in evaluative dependence. New York: Gorsuch, R. L., & Venable, G. D. (1983). Development
Wiley. of an "age universal" I-E scale. Journal for the Scientific
Darley, J. M., & Batson, C. D. (1973). "From Jerusalem Study of Religion, 22, 181-187.
to Jericho": A study of situational and dispositional Greenwald, A. G. (1975a). Consequences of prejudice
variables in helping behavior. Journal of Personality against the null hypothesis. Psychological Bulletin, 82,
and Social Psychology, 27, 100-108. 1-20.
Davidson, J. K., & Leslie, G. R. (1977). Premarital Greenwald, A. G. (1975b). Does the Good Samaritan
sexual intercourse: An application of axiomatic theory parable increase helping? A comment on Darley and
construction. Journal of Marriage and the Family, 39, Battson's no-effect conclusion. Journal of Personality
15-25. and Social Psychology, 32, 578-583.
Digenan, M. A., & Murray, J. B. (1975). Religious beliefs, Guilford, J. P. (1965). Fundamental statistics inpsychology
religious commitment, and prejudice. Journal of Social and education (4th ed.). New York: McGraw-Hill.
Psychology, 97, 147-148. Heschel, A. J. (1962). The prophets: Pan II. New York:
Dittes, J. E. (1969). Psychology of religion. In G. Lindsey Harper & Row.
& E. Aronson (Eds.). The handbook of social psychology Hoge, D. R. (1972). A validated intrinsic religious mo-
(2nd ed.): Vol. 5. Applied social psychology (pp. 602- tivation scale. Journal for the Scientific Study of Reli-
659). Reading, MA: Addison-Wesley. gion. II, 369-376.
Dittes, J. E. (1971). Typing the typologies: Some parallels Hoge, D. R., & Carroll, J. W. (1973). Religiosity and
in the career of church-sect and extrinsic-intrinsic. prejudice in Northern and Southern churches. Journal
Journal for the Scientific Study of Religion, 10, 375- for the Scientific Study of Religion, 12, 181-197.
383. Hood, R. W., Jr. (1970). Religious orientation and the
Dodrill, C., Bean, P., & Bostrom, S. (1973). The assess- report of religious experience. Journal for the Scientific
ment of religiosity in evangelical college students and Study of Religion, 9, 285-291.
its relationship to prior family religious involvement. Hood, R. W., Jr. (1971). A comparison of the Allport
Journal of Psychology and Theology, 1, 52-57. and Feagin scoring procedures for intrinsic/extrinsic
Elifson, K. W. (1976). Religious behavior among urban religious orientation. Journal for the Scientific Study
Southern Baptists: A causal inquiry. Sociological Anal- of Religion. 10, 370-374.
ysis. 37, 32-44. Hood, R. W., Jr. (1972). Normative and motivational
Ernsberger, D. J., & Manaster, G. J. (1981). Moral determinants of reported religious experience in two
development, intrinsic/extrinsic religious orientation Baptist samples. Review of Religious Research, 13,
and denominational teachings. Genetic Psychology 192-196.
Monographs, 104, 23-41. Hood, R. W., Jr. (1973). Religious orientation and the
Feagin, J. R. (1964). Prejudice and relegious [sic] types: experience of transcendence. Journal for the Scientific
A focused study of Southern fundamentalists. Journal Study of Religion, 12, 441-448.
for the Scientific Study of Religion. 4. 3-13. Hood, R. W., Jr. (1978). The usefulness of the indiscrim-
Fishman, J. A. (1956). A note on Jenkins' "Improved inately pro and anti categories of religious orientation.
method for tetrachoric r." Psychometrika, 21, 305. Journal for the Scientific Study of Religion, 17. 419-
Fullerton, J. T, & Hunsberger, B. (1982). A unidimen- 431.
sional measure of Christian orthodoxy. Journal for the Hood, R. W., Jr., & Morris, R. J. (1981). Sensory
Scientific Study of Religion, 21. 317-326. isolation and the differential elicitation of religious
Furlong, M. (1980). Merton: A biography. San Francisco: imagery in intrinsic and extrinxic [sic] persons. Journal
Harper & Row. for the Scientific Study of Religion. 20, 261-273.
Gibbs, H. W., & Achterberg-Lawlis, J. (1978). Spiritual Hunsberger, B. & Ennis, J. (1982). Experimenter effects
418 MICHAEL J. DONAHUE

in studies of religious attitudes. Journal for the Scientific McClain, E. W. (1979). Religious orientation the key to
Study of Religion, 21, 131-137. psychodynamic differences between feminists and non-
Hunt, R. A., & King, M. B. (1971). The intrinsic- feminists. Journal for the Scientific Study of Religion,
extrinsic concept: A review and evaluation. Journal 18, 40-45.
for the Scientific Study of Religion, 10, 339-356. McConahay, J. B., & Hough, J. C., Jr. (1973). Love- and
Hunter, 3. E., Schmidt, F. L., & Jackson, G. B. (1982). guilt-oriented dimensions of Christian belief. Journal
Studying organizations: Innovations in metkodobgy: for the Scientific Study of Religion, 12, 53-64.
Vol. 4. Mela-analysis: Cumulating research findings McConahay, J. B., & Hough, J. C., Jr. (1976). Symbolic
across studies. Beverly Hills, CA: Sage. racism. Journal of Social Issues, 32, 23-45.
Jackson, S. E. (1981). Measurement of commitment to McGrath, J. E, Martin, J., & Kulka, R. A. (1982).
role identities. Journal of Personality and Social Psy- Studying organizations: Innovations in methodology:
chology, 40, 138-146. Vol. 2. Judgment calls in research. Beverly Hills, CA:
Jenkins, W. L. (1955). An improved method for tetra- Sage.
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.

choric r. Psychometrika, 20, 253-258. Meadow, M. J., & Kahoe, R. D. (1984). Psychology of
This document is copyrighted by the American Psychological Association or one of its allied publishers.

Joe, V. C, McGee, S. J., & Dazey, D. (1977). Religiousness religion: Religion in individual lives. New York: Harper
and devaluation of a rape victim. Journal of Clinical & Row.
Psychology, 33, 64. Milgram, S. (1974). Obedience to authority: An experi-
Kahoe, R. D. (1974a). Personality and achievement mental view. New York: Harper & Row.
correlates of intrinsic and extrinsic religious orientation. Minton, B., & Spilka, B. (1976). Perspectives on death
Journal of Personality and Social Psychology, 29, 812- in relation to powerlessness and form of personal
818. religion. Omega, 7, 261-267.
Kahoe, R. D. (1974b). The psychology and theology of Morris, R. J, & Hood, R. W., Jr. (1981). The generaliz-
sexism. Journal of Psychology and Theology, 2, 284- ability and specificity of intrinsic/extrinsic orientation.
290. Review of Religious Research, 22, 245-254.
Kahoe, R. D. (1975). Authoritarianism and religion: Nelson, L. D., & Cantrell, C. H. (1980). Religiosity and
Relationships of F-scale items to intrinsic and extrinsic death anxiety: A multi-dimensional analysis. Review
religious orientations. JSAS Catalog of Selected Doc- of Religious Research, 21, 148-157.
uments in Psychology, 5, 284-285. (Manuscript No. Nunnally, J. C. (1976). Psychometric theory (2nd ed.).
1020) New York: McGraw-Hill.
Kahoe, R. D. (1976). Comment on Thompson's "Open- Paloutzian, R. F. (1983). Invitation to the psychology of
mindness and indiscriminate antireligious orientation." religion. Glenview, IL: Scott, Foresman.
Journal for the Scientific Study of Religion, 15, 91- Paloutzian, R. F., Jackson, S. L., & Crandall, J. E. (1978).
93. Conversion experience, belief system, and personal and
Kahoe, R. D. (19/7a). Intrinsic religion and authoritar- ethical attitudes. Journal of Psychology and Theology,
ianism: A differentiated relationship. Journal for the 6, 266-275.
Scientific Study of Religion, 16. 179-183. Patrick, J. W. (1979). Personal faith and the fear of death
Kahoe, R. D. (1977b). Religious conservatism in a quasi- among divergent religious populations. Journal for the
longitudinal perspective. Journal of Psychology and Scientific Study of Religion, IS, 298-305.
Theology, 5, 40-47. Robinson, J. P., & Shaver, P. R. (1973). Measures of
Kahoe, R. D., & Dunn, R. F. (1975). The fear of death social psychological attitudes (rev. ed.). Ann Arbor,
and religious attitudes and behavior. Journal for the MI: Institute for Social Research.
Scientific Study of Religion, 14, 379-382. Rokeach, M. (1960). The open and closed mind: Inves-
Kahoe, R. D., & Meadow, M. J. (1981). A developmental tigations into the nature of belief systems and personality
perspective on religious orientation dimensions. Journal systems. New \fork: Basic Books.
of Religion and Health, 20, 8-17. Rokeach, M. (1973). The nature of human values. New
King, M. B., & Hunt, R. A. (1972). Measuring the York: The Free Press.
religious variable: Replication. Journal for the Scientific Rychlak, J. F. (1977). The psychology of rigorous human-
Study of Religion, 11, 240-251. ism. New York: Wiley.
Krug, R. E. (1961). An analysis of the F scale: I. Item Shaver, P., Lenauer, M., & Sadd, S. (1980). Religiousness,
factor analysis. Journal of Social Psychology, 53, 285- conversion, and subjective well-being: The "healthy-
291. minded" religion of modern American women. Amer-
Lester, D. (1967). Experimental and correlational studies ican Journal of Psychiatry, 137, 1563-1568.
of the fear of death. Psychological Bulletin, 67, 27-36. Shoemaker, A., & Bolt, M. (1977). The Rokeach Value
Lovekin, A., & Malony, H. N. (1977). Religious glossolalia: Survey and perceived Christian values. Journal of
A longitudinal study of personality changes. Journal Psychology and Theology, 5, 139-142.
for the Scientific Study of Religion, 16, 383-393. Spence, J. T, & Helmreich, R. L. (1978). Masculinity
Magni, K. G. (1972). The fear of death: An exploratory and femininity: Their psychological dimensions, cor-
study of its nature and its correlates. In A. Godin relates, and antecedents. Austin: University of Texas
(Ed.). Death and presence: The psychology of death Press.
and the afterlife (pp. 125-138). Brussels: Lumen Vitae Spilka, B. (1977). Utilitarianism and personal faith.
Press. Journal of Psychology and Theology, 5, 226-233.
McClain, E. W. (1978). Personality differences between Spilka, B., & Mullin, M. (1977). Personal religion and
intrinsically religious and nonreligious students: A psychological schemata: A research approach to a
factor analytic study. Journal of Personality Assessment, theological psychology of religion. Character Potential,
42, 159-166. 8, 57-66.
1-E META-ANALYSIS 419

Spilka, B., Pelligrini, R. J., & Dailey, K. (1968). Religion, Tate, E. D., & Miller, G. R. (1971). Differences in value
American values & death perspectives. Sociological systems of persons with varying religious orientations.
Symposium. 1, 57-66. Journal for the Scientific Study of Religion, 10, 357-
Spilka, B., Stout, L., Minton, B., & Sizemore, D. (1977). 365.
Death and personal faith: A psychometric investigation. Taylor, P. A. (1972). An introduction to statistical methods.
Journal for the Scientific Study of Religion, 16, 169- Itasca, IL: F. E. Peacock.
178. Thompson, A. D. (1974). Open-mindedness and indis-
Stewin, L. L. (1976). Integrative complexity: Structure crimination [sic) antireligious orientation. Journal for
and correlates. Alberta Journal of Educational Research, the Scientific Study of Religion, 13. 471-477.
22, 226-236. Vincenzo, J., Hendrick, C, & Murray, E. J. (1976). The
Stewin, L., & Anderson, C. C. (1974). Cognitive com- relationship between religious beliefs and attending the
plexity as a determinant of information processing. fear-provoking religiously oriented movie—"The Ex-
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.

Alberta Journal of Educational Research, 20, 233- orcist." Omega, 7, 137-143.


This document is copyrighted by the American Psychological Association or one of its allied publishers.

243. Wiebe, K. F., & Fleck, J. R. (1980). Personality correlates


Strickland, B. R., & Shaffer, S. (1971). I-E, I-E, & F. of intrinsic, extrinsic, and nonreligious orientations.
Journal for the Scientific Study of Religion, 10, 366- Journal of Psychology. 105, 181-187.
369. Wilson, W. C. (1960). Extrinsic religious values and
Strickland, B. R., & Weddell, S. C. (1972). Religious prejudice. Journal of Abnormal and Social Psychology,
orientation, racial prejudice, and dogmatism: A study 60, 286-288.
of Baptists and Unitarians. Journal for the Scientific Wimberley, R. C. (1974). Toward the measurement of
Study of Religion, 11. 395-399. commitment strength. Sociological Analysis, 15, 211-
Strommen, M. P. (Ed.) (1971). Research on religious 215.
development: A comprehensive handbook. New York:
Hawthorne Books.
Sturgeon, R. S., & Hamley, R. W. (1979). Religiosity and Received April 5, 1983
anxiety. Journal of Social Psychology, 108, 137-138. Revision received August 8, 1983

Correction to Swingle

In the article "Temporal Measures of Vocalization: Some Methodological Con-


siderations," by Paul G. Swingle (Journal of Personality and Social Psychology,
1984, Vol. 47, No. 6, pp. 1263-1280), the copyright notice was inadvertently
omitted. The notice that should have appeared on the first page of this article is
as follows:
Journal of Personality and Social Psychology, 1984, Vol. 47, No. 6, 1263-1280
Copyright 1984 by the American Psychological Association, Inc.

You might also like