RELIGIOUS ORIENTATION SCALE

Variable:

The Religious Orientation Scale (ROS) is based on Allport’s early (1950) conceptual work where he characterized the so-called mature religious sentiment as: “1) well-differentiated [complex and critically embraced]; 2) dynamic in character in spite of its derivative nature [motivational in and of itself]; 3) productive of a consistent morality [shapes personal ethical code]; 4) comprehensive [applies to all areas of life]; 5) integral [capable of assimilating new in­ formation]; and 6) fundamentally heuristic [tentatively, though not lightly, held]” (pp. 64-65, interpretive brackets inserted). The “immature” religious sentiment, it was as­sumed, embodied the opposite of these characteristics. Details of the immature-ma­ture distinction are largely absent from later discussions (Allport, 1966; Allport & Ross, 1967) of the relabeled extrinsic and intrinsic religious orientations; however, it is unclear whether this omission represented a narrow­ ng of Allport’s thinking or merely an indi­ cation that the earlier proposed differences had become implicit assumptions. Still ex­plicit, however, was that religious orienta­tion (or sentiment) is a motivational con­ struct: “Instrumental versus ultimate,” “peripheral versus central,” and “servant versus master” all capture the essence of the differential role that Allport assumed reli­gion to occupy within the individual’s life depending on whether he or she is extrinsi­cally or intrinsically oriented, respectively. More formally, extrinsic religious orienta­tion refers to a flagrantly utilitarian motiva­tion underlying religious behaviors: The in­ dividual endorses religious beliefs and attitudes or engages in religious acts only to the extent that they might aid in achieving mundane goals, such as feeling comforted and protected or acquiring social status and approval. In contrast, intrinsic religious ori­entation refers to motivation arising from goals set forth by the religious tradition it­ self, and is thus assumed to have another,” nonmundane, even self-denying qual­ity: Religion is regarded as a “master motive … [whereas] other needs, strong as they may be, are regarded as of less ultimate significance” (Allport & Ross, 1967, p. 434, brackets inserted). Based on this distinction, many subsequent researchers have adopted the convenient, albeit simplistic, conceptual shorthand, initiated by Allport and Ross themselves, of referring to the extrinsic-in­trinsic distinction as “using” versus “living” one’s religion.

Description:

Subscales. Although the ROS represents Allport’s capstone effort to operationalize the extrinsic and intrinsic religious orienta­tions, there were at least two earlier efforts to tap these constructs. Specifically, Wilson (1960), with assistance from Allport, con­ structed a 15-item, forced-choice measure of extrinsic (but not intrinsic) “religious val­ues.” Several years later, Feagin (1964) presented 21 items from whence he derived both a 6-item Extrinsic and a 6-item Intrin­sic Scale. (All but 1 of Feagin’s original 21 items subsequently appeared in the ROS.) Neither of these earlier efforts has received the empirical attention that the ROS has, however.

Within the ROS itself, the Extrinsic (sub)scale assesses an individual’s degree of acknowledgment of the peripheral role that religion plays in his or her life, as well as the degree to which he or she frankly admits to religious involvement in order to secure solace and/or social approval. That is, the items appear to operationalize straightfor­wardly the key elements of extrinsic orien­tation as Allport (1966; Allport & Ross, 1967) understood them.

Sampled from a variety of religious atti­tudes, behaviors, and intentions, Intrinsic (sub)scale items at first glance seem less conceptually focused than do Extrinsic items. To the extent that intrinsic orientation involves enshrining religion as the “master motive” of one’s life as Allport and Ross (1967) suggested, however, the items make considerable sense, for they all reflect the no-nonsense fervency of commitment that such a master motive might evoke (at least as it might be articulated within a traditional Christian context; see “Practical Considera­tions”).

Scoring. Extrinsic and Intrinsic ROS items are best treated as composing distinct scales, owing to the absence of a straightfor­ward inverse relationship between the two orientations (see “Validity”). Thus, given the 5-point (1 = strongly disagree to 5 = strongly agree) response format used in the original report (Allport & Ross, 1967), sep­arate summation of the respective scale items yields score ranges of 11-55 and 9-45 for the Extrinsic and Intrinsic Scales. A 9- point response format is preferred by some researchers (e.g., Batson, 1976). Regardless of the specific response format used, it is recommended that means are scaled to the response format rather than reported as un­ scaled totals, for the former allows for meaningful comparisons between Extrinsic and Intrinsic scale scores within a given sample.

Another scoring issue linked to validity is whether and how individuals should be assigned religious-orientation-type labels based on their Extrinsic and Intrinsic scores (see, e.g., Burris, 1994; Hood, 1978; but also see Batson et al., 1993). This issue was initially confronted by Allport and Ross (1967) when Extrinsic and Intrinsic scores in their samples appeared to be linearly in­ dependent rather than inversely related as expected. In order to account primarily for those individuals who simultaneously tended toward agreement on both the Ex­trinsic and Intrinsic Scales-the so-called indiscriminately proreligious-Allport and Ross treated these respondents as a group, comparing them on pertinent dependent variables with those who tended toward agreement on only one of the two scales (“Extrinsics” or “Intrinsics”), and with those who tended toward disagreement on both scales (the “indiscriminately proreli­gious”). This later evolved into a median­ split approach to classification, in which the four groups are created based on whether individuals score above or below the respec­tive Extrinsic and Intrinsic medians for that sample. The chief advantage of the median­ split approach is that it assures a relatively equal representation of respondents in each of the four groups regardless of sample characteristics. This is also the chief disad­vantage: Extrinsic and Intrinsic score distributions-and thus, their medians-vary as a function of faith tradition (see Burris, Jack­ son, Tarpley, & Smith, 1996, Study 3), so labeled groups (e.g., Intrinsics) may not be comparable across samples. For this reason, the more conceptually meaningful practice of splitting the sample at the scales’ theoret­ical mid- or neutral-point (e.g., 5 on a 1-9 scale) is preferred. Whatever the procedure, Burris (1994) has recommended that typing be employed only when adequately theoreti­cally justified.

Spin-offs. Suggested revisions and re­placements for the ROS scales have been numerous, although they have generally arisen based on one of two types of criti­cisms: the “pure empirical” and the “con­ ceptual-empirical.” Pure empirical criti­cisms focus on one or more of the ROS’s perceived psychometric inadequacies, e.g., the absence of a strongly inverse Extrinsic/ Intrinsic correlation (Hoge, 1972; see also this volume), low interitem correlations and/or multidimensionality of especially the Extrinsic Scale (Genia, 1993; Gorsuch & McPherson, 1989; see also this volume; Kirkpatrick 1989), or excessively abstruse item wording (Gorsuch & Venable, 1983; see also this volume). Conceptual-empirical criticisms are based more on what the ROS does and does not measure than on how well it measures. Typically, Allport’s writings on immature/extrinsic and mature/intrinsic reli­gion are compared against the content of ROS items, discrepancies are noted, and new scales are proposed to fill the presumed conceptual gaps (e.g., Dudley & Cruise, 1990). Of these, the most empirically pro­ lific has been Batson ‘s (1976; Batson, Schoenrade, & Ventis, 1993) means-end ap­proach. It will thus be discussed in some de­ tail.

According to Batson (1976), Allport’s (1950) depiction of the “mature” religious sentiment actually confounds two forms of religious orientation that are conceptually and empirically distinct. The Intrinsic scale of the ROS taps only the “religion as master motive” theme that pervades Allport’s dis­cussion of the intrinsic orientation in later writings, Batson claimed: Unmeasured by either the Intrinsic or the Extrinsic Scale is the motivation to grapple with existential questions, to view religious doubts as posi­tive, and to remain open to religious change that peppers earlier discussions of the ma­ture sentiment. Batson and colleagues thus developed the Interactional, or Quest, Scale to tap these hitherto unmeasured themes (Batson, 1976; Batson & Schoenrade, 1991; see also this volume), and three additional scales intended to capture additional aspects of the extrinsic and intrinsic orientations that seemed implicit but unmeasured in the two constructs. Specifically, based on his assumption that the “master motive” quality of the in­trinsic orientation may be an outgrowth of needs for certainty, strength, and direction that express themselves outwardly (in part) through wholehearted endorsement of insti­tutionally approved religious doctrines, Bat­ son (1976) constructed the Internal and Doctrinal Orthodoxy scales. The former es­sentially measures an individual’s “need to believe” (in religion); the latter measures an individual’s degree of endorsement of a number of traditional Christian beliefs. Both scales were predicted and found to be mod­erately to strongly positively correlated with the Intrinsic scale. The remaining, External, scale measures the degree to which an indi­vidual’s religion is affected by influential others such as peers, family members, and lay or professional religious workers. Bat­ son (1976) initially predicted that this scale would be positively correlated with the Ex­trinsic scale; it has, however, almost invari­antly correlated positively with the Intrinsic Scale instead, leading Batson et al. (1993, p. 169) to concede that the initial assumption has “proved wrong.”

The six scales (all of which use a 9-point response format) in combination are in­ tended to measure three dimensions of reli­gious orientation that Batson assumes (and statistically forces) to be independent: The devout, doctrinaire End dimension (assessed by the Intrinsic, Internal, External, and Doc­trinal Orthodoxy Scales), the utilitarian Means dimension (assessed primarily by the Extrinsic Scale), and the existentially toned Quest dimension (assessed primarily by the Quest Scale). The statistical procedures for deriving Means and End (as well as Quest) scores are sufficiently complex as to be be­ yond the scope of this volume; see Batson et al. (1993) for details.

Practical Considerations:

Although Allport used the term “religion” rather generically in his theoretical works, he was undoubt­edly influenced by his cultural, familial, and personal ties to a North American Protestant articulation of Christianity in his construction of the ROS (see Wulff, 1997). Hence, items including references to church and Bible study, for example, restrict the ROS’s interpretability primarily to respondents with a Christian background. Modifications involving less sectarian wording, e.g., from “church” to “religious gathering,” that do not alter the items’ meaning substantially are therefore recommended.

Finally, as noted under “Description,” the ROS has been criticized for the relatively high reading level of its items. Although this probably presents no problem in most adult samples, it is a legitimate concern when working with special adult popula­tions or with children. An “age universal” version of the ROS has been developed for such instances (Gorsuch & Venable, 1983; see also this volume).

Norms/Standardization:

Allport and Ross’s (1967) sample consisted of 309 members of six different churches/denominations (Roman Catholic, Lutheran, Nazarene, Pres­ by terian, Methodist, Baptist) scattered across the eastern United States, a sample that was claimed to be “in no sense representative” (p. 436). Unfortunately, Allport and Ross reported Intrinsic and Extrinsic scale means for neither each subsample nor the total sample. Donahue (1985a, p. 419) speculated that “smaller, more sect-like [re­ligious] groups, would be expected to have higher Intrinsic and lower Extrinsic scores than larger denominations due to their more stringent membership requirements.” Offer­ing some support for this speculation, Burris et al. (1996, Study 3) found higher Intrinsic scores among conservative Protestant groups (e.g., Baptist, Pentecostal, Mormon) than among either liberal Protestants (e.g., Episcopal, Methodist, Presbyterian) or Catholics in a mid-westem U.S. university sample. Conservative Protestants also aver­ aged lower on the Extrinsic scale compared to Catholics but not compared to liberal Protestants.

Reliability:

Internal consistencies reported for the ROS Intrinsic scale range from ade­ quate to excellent, with Cronbach’s alphas most typically in the mid .80s (e.g., Don­ahue, 1985a). Internal consistencies re­ ported for the ROS Extrinsic scale are in­ variably lower, with Cronbach ‘s alphas most typically in the low .70s (e.g., Don­ahue, 1985a). Burris and Tarpley (in press, footnote 3) reported two-week test-retest re­ liabilities of .84 and .78 for the Intrinsic and Extrinsic scales, respectively (N = 61). The lower reliabilities associated with the Ex­trinsic scale, although subject to criticism, can be attributed-at least in part-to the scale’s tapping of multiple manifestations of the extrinsic orientation (e.g., Kirkpatrick, 1989). Whether the trade-off of psychomet­ric potency for conceptual breadth is justifiable remains open to debate.

Validity:

Evaluating the validity of the ROS Intrinsic and Extrinsic scales is, unfortu­nately, not a simple task, given the subtle shifts in emphasis in Allport’s writings over time, and given the value-ladenness of the religious orientation constructs. The first issue affects evaluations of the scales’ struc­tural properties; the second raises questions as to what standards or markers should be considered relevant to validity.

Structure. As earlier noted, Allport’s dis­cussion (if not his conceptualization) of the intrinsic orientation appeared to narrow over time to a primary emphasis on the ori­entation’s “master motive” character. Thus it could be argued that the adequacy of the Intrinsic scale as a measure of intrinsic ori­entation depends on whether one focuses on Allport’s early or later writings. The Intrinsic scale’s combination of relatively high in­ ternal consistency and breadth of item con­ tent would seem to support its validity as a measure of “religion as a master motive.” Indeed, from this perspective, that intrinsic items scatter across a number of factors when factor-analyzed with other items tap­ ping traditional religious attitudes and activ­ities is not as problematic as Hunt and King (197 l) claimed. Rather, the scattering can be interpreted as suggesting a common un­derlying construct, i.e., centralized religious motivation. On the other hand, the Intrinsic scale does not appear to be adequate as a comprehensive measure of “religious matu­rity” a la Allport (1950), as Batson (1976) has demonstrated.

In contrast, there is much less uncertainty as to whether the Extrinsic scale gets at the intended orientation. Although the scale’s multidimensionality has been criticized on both conceptual and empirical grounds (e.g., Kirkpatrick, 1989) that Extrinsic items load on separable factors related to comfort-seek­ing, status-seeking, and the admission of re­ligion’s tangentiality conforms precisely to Allport’s (1966; Allport & Ross, 1967) con­ceptualization. Thus, from a structural-con­ tent standpoint, the validity of the Extrinsic scale, like the validity of the Intrinsic scale, is very much a matter of perspective.

The same can be said from a structural­ relational standpoint. As was earlier noted, Allport and Ross (1967) clearly expected a strong, inversely linear relationship between their Extrinsic and Intrinsic itemsconsis­ tent with an hypothesized bipolar religious orientation dimension-rather than the near­ zero linear correlation between the two scales that they found. In fact, the expected negative relationship has been obtained but has been restricted to theologically conserv­ative samples (Donahue, 1985b), leading the majority of researchers (e.g., Batson, 1976) to conclude that Allport was wrong, i.e., that the extrinsic and intrinsic orienta­tions are not opposites but are independent. Burris (l 994) demonstrated, however, that the frequently observed near-zero linear re­lationship masks a substantial nonlinear re­lationship. Specifically, the Intrinsic and Extrinsic scales were shown to be inversely curvilinearly related such that, below the In­ trinsic midpoint, the Intrinsic/Extrinsic cor­ relation is positive, suggesting rejection of both forms of religious motivation, or sim­ ple irreligiosity. Above the Intrinsic mid­ point, however, the Intrinsic/Extrinsic corre­ lation is negative, suggesting that, as reported intrinsic orientation increases, re­ ported extrinsic orientation decreases. (A similar overall relationship has been ob­ served for the Intrinsic and Quest scales­ see Burris, 1994). This pattern thus offers at least partial support of Allport and Ross’s bipolarity assumption, but no support for the independence assumption.

Standards. The value-ladenness of the extrinsic/intrinsic distinction is most evident when determining what should serve as standards or markers of the ROS’s validity. Allport and Ross (1967) provided no data relevant to the validity of the Extrinsic and Intrinsic scales, apparently assuming that the differential relationship observed be­ tween measures of ethnocentrism and these two scales was adequate. Their assumption was clear: As the old “mature” and “imma­ture” labels implied, the intrinsic orientation was seen as the embodiment of “good” or “true” religion, whereas the extrinsic orien­tation was the embodiment of “bad” or “false” religion; “true” religion cannot, or at least should not, foster intolerance. Al­ though perhaps quite compelling as a theo­ logical prescription, this assumption-that prejudice or its absence should ipso facto serve as a criterion for the religious orienta­tion scales’ validity-warrants clearer psy­chological (i.e., conceptual) justification. Much subsequent research utilizing the ROS and related measures has sidestepped issues of validity, however, by relying-as did Allport-upon an implicit “extrinsic is bad, intrinsic is good” heuristic. As a conse­quence, research that challenges the moral fiber and purity of motives assumed to be associated with the intrinsic orientation is often met with sharp criticism and accusations of bias.

because of its implicit value assumptions, seldom elicits indifference. Moreover, re­ searchers’ personal responses to these value assumptions undoubtedly affect how ques­tions of validity are framed. Indeed, one might suggest that, failing all else, the ROS might serve as a sort of projective test of the values and predilections of psychologists of religion! Having said this, what data (if any) might speak regarding the validity of the Extrinsic and Intrinsic scales?

If the Intrinsic scale indeed measures “re­ligion as a master motive,” it should be rather strongly positively correlated with measures tapping commitment to, or as­cribed importance of, religion. This is, in fact, the case (Donahue, 1985b). Moreover, numerous studies have found a positive correlation between the Intrinsic scale and measures of one’s general sense of purpose in life, also consistent with the “master motive” conceptualization (see Batson et al.,1993, chap. 8).

If the Extrinsic scale indeed taps need to “look good” to oneself and others. Defenders of the intrinsic orientation have responded by suggesting that measures of social desirability are biased against religious respondents or, alternatively, that in­trinsically oriented individuals report being more socially desirable because they are more socially desirable (e.g., Watson, Mor­ris, Foster, & Hood, 1986; Richards, 1994), although empirical support for either con­tention remains rather questionable (Burris,Ll 994; Leak & Fish, 1989). Similar denunci­rations (e.g., Gorsuch, 1993) have been made with respect to research linking the intrinsic orientation with subtle and not-so-subtle forms of prejudice (see Batson & Burris, 1994, for a review). In contrast, this re­ viewer is unaware of any comparable accu­sations of bias with respect to research demonstrating links between the extrinsic orientation and unsavory variables such as ethnocentrism. To be certain, the conceptual and empiri­ cal claims and counterclaims raised regard­ing the extrinsic and intrinsic orientations remain sensitive and controversial. That is precisely the point: Allport’s framework, “hands-off” attitude toward religion, then it “‘=· should not be positively correlated with measures of religious commitment. One again, this is the case (Donahue, 1985b Moreover, if the Extrinsic scale assesses one’s frank admission of using religion for comfort, then it should be linked to variables suggestive of stress and maladjustment, i.e., variables that might encourage an otherwise irreligious person to “try ” religion. Research is generally consistent with this suggestion: Batson et al. (l 993) concluded, based on a review of findings from over 40 studies, that there is “considerable evidence that this [extrinsic, means] dimension is negatively associated with several conceptions of mental health,” including “appropriate social behavior” and “freedom from worry and guilt” (p. 286, brackets inserted). Moreover, Burris, Batson, and Wag- oner (1992) found that persons randomly as- signed to complete an esteem-threatening writing task subsequently scored higher on the personal comfort subscale of the Extrinsic scale (Kirkpatrick, 1989) than did those who completed a neutral writing task, offering some experimental evidence for validity.

Direct evidence of this sort suggesting that the Extrinsic scale effectively taps the use of religion to bolster one’s social status has yet to be produced, however.

In short, the research reviewed here­ limited in scope due to the sheer breadth of the literature, and conservatively selective due the the value-ladenness of the con­ structs-seems generally supportive of the validity of the extrinsic and intrinsic orien­tations and the scales used to measure them. The research does not, however, support Allport’s conceptualizations in every detail, a fact that will undoubtedly elicit sparring between apologists and critics of religious orientation for some time yet (e.g., Kirk­ patrick & Hood, 1990; Masters, 1991). In this reviewer’s opinion, essential to ensur­ing a “fair fight” is a retooling of Allport’s model based upon what has been learned in the past three decades. Such a revised framework should remain true to the mo­ tive-centered spirit-if not the letter-of Allport’s extrinsic-intrinsic framework, but should also be capable of incorporating reli­gious orientations more recently identi­fied-e.g., Batson’s (1976) quest. Such a re­ vised framework should also move beyond theological prescriptions to psychological principles as a basis for predicting and ex­ plaining relationships between religious ori­entation and variables of interest, e.g., prej­udice, prosocial behavior, and mental health. It is a hopeful sign that attempts at such a framework are beginning to appear (e.g., Pargament, 1992; Burris, 1997). Given the fervency with which he strove to understand the vagaries of religious motiva­tion during his own life, we can only as­sume that Allport would have wanted it this way.

Religious Orientation Scale (ROS)

Please indicate the extent to which you agree or disagree with each item below by using the following rating scale:*

Extrinsic ( sub )scale**

1. Although I believe in my religion, I feel there are many more important things in my life.
2. It doesn’t matter so much what I believe so long as I lead a moral life.
3. The primary purpose of prayer is to gain relief and protection.
4. The church is most important as a place to formulate good social relationships.
5. What religion offers me most is comfort when sorrows and misfortune strike.
6. I pray chiefly because I have been taught to pray.
7. Although I am a religious person I refuse to let religious considerations influence my everyday affairs.
8. A primary reason for my interest in religion is that my church is a congenial social activity.
9. Occasionally I find it necessary to compromise my religious beliefs in order to pro­ tect my social and economic well-being.
10. One reason for my being a church member is that such membership helps to establish a person in the community.
11. l The purpose of prayer is to secure a happy and peaceful life.
12. *** Religion helps to keep my life balanced and steady in exactly the same way as my citizenship, friendships, and other memberships do.

Intrinsic ( sub )scale**

1. I. It is important for me to spend periods of time in private religious thought and meditation.
2. If not prevented by unavoidable circumstances, I attend church.
3. I try hard to carry my religion over into all my other dealings in life.
4. The prayers I say when I am alone carry as much meaning and personal emotion as those said by me during services.
5. Quite often I have been keenly aware of the presence of God or the Divine Being.
6. I read literature about my faith (or church).
7. If I were to join a church group I would prefer to join a Bible study group rather than a social fellowship.
8. My religious beliefs are really what lie behind my whole approach to life.
9. Religion is especially important because it answers many questions about the meaning of life.

•Many researchers have used a 9-point response format.

.. The ordering of all 20 items should be scrambled.

••• Indicates an additional Extrinsic item used by Feagin (1964) but not by Allport and Allport and Ross (1967).

Batson’s Supplementary “End Dimension” Scales

Please indicate the extent to which you agree or disagree with each item below by using the following rating scale:*

Internal scale

1. My religious development is a natural response to our innate need for devotion to God.
2. God’s will should shape my life.
3. It is necessary for me to have a religious belief.
4. When it comes to religious questions, I feel driven to know the truth.
5. (-) Religion is something I have never felt personally compelled to consider.
6. (-) Whether I tum out to be religious or not doesn’t make much difference to me.
7. I have found it essential to have faith.
8. I find it impossible to conceive of myself not being religious.
9. (-) For me, religion has not been a “must.”

External scale

1. The church has been very important for my religious development.
2. My minister (or youth director, camp counselor, etc.) has had a profound influence on my personal religious development.
3. A major factor in my religious development has been the importance of religion for my parents.
4. My religion serves to satisfy needs for fellowship and security.
5. Certain people have served as “models” for my religious development.
6. (-) Outside forces (other persons, church, etc.) have been relatively unimportant in my religious development.

Doctrinal Orthodoxy scale

1. I believe in the existence of a just and merciful personal God.
2. I believe God created the universe.
3. I believe God has a plan for the universe.
4. I believe Jesus Christ is the divine Son of God.
5. I believe Jesus Christ was resurrected (raised from the dead).
6. I believe Jesus Christ is the Messiah promised in the Old Testament.
7. I believe one must accept Jesus Christ as Lord and Savior to be saved from sin.
8. I believe in the “second coming” (that Jesus Christ will one day return to judge and rule the world).
9. I believe in “original sin” (we are all born sinners).
10. I believe in life after death.
11. I believe there is a transcendent realm (an “other” world, not just this world in which we live).
12. I believe the Bible is the unique authority for God’s will.

Note.(-) indicates a reverse-scored item. For additional unscored buffer items, as well as details regard­ing scoring procedures, see Batson et al. (1993).

Location:

Subsequent Research:

References