u Cju^rt+čf Pw-r* <&**& >;^-Introduction ^\^, ^V*~> £j t* H. NEWTON MALONY The way people think about religion, the importance they place on it, the reasons they have for being religious—all provide the background for behavior. This section deals with the effect religious attitudes have on such nonrcligious issues as prejudice. This contrasts with the next section, "Psychodynamics and Religion," in which the effects of personality differences on religious behavior are considered. Allport was very concerned with the evidence that religious persons are more prejudiced than nonreligious persons. Attempting to understand this issue, he suggested that there are two types of religious persons—those who use religion (the extrinsic) and those who live religion (the intrinsic). He and Ross in their article report research comparing these types of persons. They suggest that persons who are churchgoers, for any and all reasons, are more prejudiced than any other types of persons. Allport and Ross' study has exerted considerable influence during the last decade over the psychology of religion. Hunt and King survey and summarize this research and compare the intrinsic-extrinsic concept with many other similar ideas. They also discuss the Religious Orientation Scale, which has been widely used to measure these types of attitudes toward religion. Dittcs compares the intrinsic-extrinsic npologv with theorizing by sociologists, such as Troeltsch and Weber. He notes that theorists have long been concerned with pure (sect) and historical (church) religion. He concludes that religion is much more complex than any twofold typology can encompass. This portion of the book will acquaint the student with the continuing dialogue about various kinds of religious persons and ongoing efforts to differentiate them, one from another. 116 Personal Religious Orientation and Prejudice GORDON W. ALLPORT AND > MICHAEL ROSS Previous psychological and survey research has established three important facts regarding the relationship between prejudiced attitudes and the personal practice of religion. 1. On the average, church attenders are more prejudiced than nonattenders. 2. This overall finding, if taken only by itself, obscures a curvilinear relationship. While it is true that most attenders are more prejudiced than nonattenders. a significant minority of them are iess prejudiced. 3. It is the casual, irregular fringe members who are high in prejudice; their religious motivation is of the extrinsic order. It is the constant, devout, internalized members who are low in prejudice; their religious motivation is of the intrinsic order. The present paper will establish a fourth important finding— although it may properly be regarded as an amplification of the third. The finding h that a certain cognitive style permeates the thinking of many people in such a way that the) are indiscriminately proreligious and, at the same time, highly prejudiced. '• ~"~~'' -"^ Buf first let us make clear the types of evidence upon which the first three propositions are based and examine their theoretical significance. CHURCHGOERS ARE MORE PREJUDICED Beginning the long parade of findings demonstrating that churchgoers are more intolerant of ethnic minorities than nonattenders is a study by Allport and Kramer (1946). These authors discovered that Reprinted with permission of the authors and publisher from: hurnai uf Personality and Social Psychology, 1967, 5 (4), 4Í2-443, Copyrighl (1967) by the American Psychiilogical Association. 117 ^^^Bffport and J. M. Ross * students who claimed no religious affiliation were less likely to he anti- * Negro than those who declared themseh es to be protestant or Catholic. % Furthermore, students reporting a strong religious influence at home were higher in ethnic prejudice than students reporting only slight or no religious influence. Rosenblith (1949) discovered the same trend among students in South Dakota. The Authoritarian Personality (Adorno, Frenkel-Brunswik, Lcvinson, & Sanford, 1950, p. 212) stated that scores on cthnocentricism (as well as on authoritarianism) are significantly higher among church attenders than among non-attenders. *. Gough's (1951) findings were similar. Kirkpatrick (19491 found reli- | gjous people in general to be slightly less humanitarian than nonrchgious I people. For example, they had more punitive attitudes toward crimi- s nals, delinquents, prostitutes, homosexuals, and those in need of i psychiatric treatment. Working with a student population Rokeach '■% (1960) discovered nonbelievers to be consistently less dogmatic, less | authoritarian, and less ethnocentric than believers. Public-opinion i polls (as summarized by Stember, 1961) revealed confirmatory evidence | across the board. I Going beyond ethnic prejudice, Stouffer (1955) demonstrated that among a representative sample of American church members . those who had attended church within the past month were more in- ■ tolerant of nonconformists (such as socialists, atheists, or communists) ,r than those who had not attended. It seems that on the average religious . people show more intolerance in general—not only toward ethnic but also toward ideological groups. Is this persistent relationship in any way spurious? Can it be due, for example, to the factor of educational level? Many studies show ■' that people with education tend tobe appreciably less prejudiced than people with low education. Perhaps it is the former group that less -often goes to church. The reasoning is false. Sociological evidence has a shown conclusively that frequent church attendance is associated w ith high socioeconomic status and with college education (Demerath, 1965). Furthermore. Stouffer's study found that the intolerant tendency among churchgoers existed only when educational level was held constant. Struening (1963), using as subjects only faculty members of a large state university (all highly educated), discovered that non-attenders were on the average less prejudiced than attendcrs. These '* studies assure us that the association between churehgoing and prejudice is not merely a spurious product of low education. Turning to the theoretical implications of these findings, shall we say that religion in and of itself makes for prejudice and intolerance? There are some arguments in favor of such a conclusion, 118 Personal Religious Orientation and ľreiudice especially when we recall that certain powerful ikeo'ogical positions —those emphasizing revelation, election (chosen jvople), and theocracy (Allport, 1959, 1966)—have throughout histor> turned one religion against another. And among sociological factors in religion we find many that make for bigotry. One thinks of the narrow composition of many religious groups in terms of ethnic and class membership, of their pressure toward conformity, and of the -n petition between them (see Demerath, 1965; l.enski, 1961). It does ."-.in that religion as such makes for prejudice. And yet it is here that we encounter the grand paradox, One may not overlook the teachings of equality and brotherhood, of compassion and humanheartedness, that mark all the great world religions. Nor may one overlook the precept and example of great figures whose labors in behalf of tolerance were and are religiously motivated—such as Christ himself, Tertullian, Pope Gelasius I, St. Ambrose, Cardinal Cusa, Sebastian Castellio, Schwenekfeld, Roger Williams, Mahatma Gandhi, Martin Luther King, and many others, including the recently martyred clergy in our own South, These lives, along with the work of many religious bodies, councils, and service organizations, would seem to indicate that religion as such unmakes prejudice. A paradox indeed. THE CURVILINEAR RELATIONSHIP If religion as such made only for prejudice, we would expect that churchgoers who expose themselves most constantly to its influence would, as a result, be more prejudiced than those who seldom attend. Such is not the case. Many studies show that frequent attenders are less prejudiced than infrequent attenders and often less prejudiced even than non-attenders. Let us cite one illustrative study by Struening (1963). The cut\ ilinear trend is immediately apparent in Table 1. In this particular study nonattenders had lower prejudice scores than any group, save only those de\ otees who managed to attend 11 or more times a month. Without employing such fine time intervals other studies have shown the same curvilinear trend. Thus, in The Authoritarian Personality (p. 212) we learned that in 12 out of 15 groups "regular'" attenders (like nonattenders) were less prejudiced than "seldom" or "often" attenders. Employing a 26-item Desegregation Scale in three separate studies, Holtzman (1956) found the same trend as shown in Table 2. If more evidence for the curvilinear relationship is needed, it will be found in community studies made in New Jersey (Friedrichs, 1959), 119 tV. Allport and J. M. Ross I TABLE 1 Chi rch Attendance and Prejudici among ť ah i n M embirs Ol A MIDWESTERN ĽNIVEKSITY Frequency of attendance (.limes pet mo.) 0 1 2 3 4 5-7 8-10 11 or more Note. -From Slruening ^ 19>7.. \ Prejudice score 261 14.7 143 25.0 103 26.0 84 23.8 157 22.0 94 19.9 26 16.3 21 11.7 TABLE 2 Church Attendance and Prejudice among Students in the Border States 1956 study Mean score on D scale "k intolerant 19S8 study 1960 study Nonattenders 37 413 Once a mo. 66 48.5 Twice a mo. 67 50.6 Once a wk. or oflener 49 44.5 38.1 51.4 48.4 443 Note. —Adapted Stata Holt/man i 1956;. Kelli. F (-[»on. and lloltimjin ' I9ÍS . Ymitig, Bi-nwn,. ändtluli™uiL]9601- North Carolina (Tumin, 1958), New England (Pettigrew, 1959), and Ohio and California (Pinkney, 1961). One could almost say there is a unanimity of findings on this matter. The trend holds regardless of religion, denomination, or target of prejudice (although the case seems less clear for anti-Semitism than for prejudice against other ethnic groups). What are the theoretical implications? To find that prejudice is related to frequency of church attendance is scarcely explanatory, since it may reflect only formal behavior, not involvement or commitment to religious values. And yet it seems obvious that the regular attenders who go to church once a i\ eek or oftener (and se\ eral studies indicate that oftener than once a week is especially significant) are 120 Personal Religious Orie Urion and Prejudice u e\ people who receive something of special ideological and experiential meaning. Irregular, casual fringe members, on the other hand, regard their religious contacts as less binding, less absorbing, less integral ith their personal lives. At this point, therefore, we mu«: pass from external behavioral idence into the realm of experienc-> ..c.,\ motivation. Unless we do so we cannot hope lo understand the mr- - i linear relationship that has been so clearly established. EXTRINSIC VERSUS INTRINSIC MOTIVATION Perhapsjhc briefest way to characterize the two poles of sub-jective religion is to say that the extrinsically motivated person uses his religion, whereas the intrisically motivated lives his religion. As we shall see later, most people, if they profess religion at all, fall upon a continuum between these two poles. Seldom, if ever, does one encounter a "pure" case. And yet to clarify the dimension it is helpful to characterize it in terms of the two ideal tvpes. Extrinsic Orientation Persons with this orientation arc disposed to use religion for their own ends. The term is borrowed from axiology, to designate an interest that is held because it serves other, more ultimate interests. Extrinsic values are always instrumental and utilitarian. Persons with this orientation may find religion useful in a variety of ways—to provide security and solace, sociability and distraction, status and self-justification. The embraced creed is lightly held or eist selectively shaped to fit more primary needs. In theological terms the extrinsic type turns to God, but without turning away from self. Intrinsic Orientation Persons with this orientation find their master motive in religion. Other needs, strong as they may be, are regarded as of less ultimate "gnihvance, and they are, so far as possible, brought into harmony with »»e religious beliefs and prescriptions. Having embraced a creed the '"dividual endeavors to internalize it and follow it fullv. It is in this «■'nse that he lives his religion. A clergyman was making the same distinction when he said, Some people come to church to thank Cod, to acknowledge His glory, and to ask His guidance-----Others come for what thev can get. Their interest in the church is to run it or exploit it rather than to Serve it 121 ü. W. Allport and J. M. Ross ŕ Approximate parallels to these psychological types have been proposed by the sociologists Fichter (1954) and Lenski (1961). The former, in studying Catholic parishioners, classified them into four groups: the dormant, the marginal, the modal, and the nuclear. Omitting the dormant, richtet estimated in terms of numbers that 20% are marginal, 70% modal, and less than 10% nuclear. It is, of course, the latter group that would most closely correspond to our conception of the "intrinsic." Lenski distinguished between church members ~ whose involvement is "communal" (for the purpose of sociability and i status} and those who are "associational" (seeking the deeper \alues of their faith). These authors see the significance of their classifications for , the study of prejudice. Fichter has found less prejudice among devout j (nuclear) Catholics than among others (see Allport, 1954, p. 421), . Lenski (1961, p. 173) reported that among Detroit Catholics 59% of J those with a predominantly "communal" involvement favored segre- 5 gated schools, whereas among those with predominantly an "associa- % tional" involvement only 27% favored segregation. The same trend held ; for Detroit Protestants. ^ The first published study relating the extrinsic-intrinsic j-dimension directly to ethnic prejudice was that of Wilson (I960). :' Limiting himself to a 15-item scale measuring an extrinsic (utilitarian-1 institutional) orientation, Wilson found in 10 religious groups a median í correlation of .65 between his scale and anti-Semitism. In general these • correlations were higher than he obtained between anti-Semitism and . the Religious-Conventionalism Scale (Levinson, 1954). From this finding Wilson concluded that orthodoxy or fundamentalism is a less * important factor than extrinsieness of orientation. 2 Certain weaknesses may be pointed out in this pioneer study.!; Wilson did not attempt to measure intrinsicness of orientation, but" assumed without warrant that it was equivalent to a low score on the'; extrinsic measures. Fuuher, since the items were worded in a uni-J directional way there may be an error of response set. Again, Wilson J dealt only with Jews as a target of prejudice, and so the generality^ of his finding is not known. •; Finally, the factor of educational level plays a part. Wilson used? the California Anti-Semitism scale, and we know that high scores^ on this scale go with low education (Christie, 1954; Peltigrcw, 1959í-Titus & Hollander, 1957; Williams, 1964). Further, in our own study the extrinsic subscale is negatives correlated with degree of education (r— -.32). To an appreciable extent, therefore, Wilson's high cor-, relations may be "ascribed" to educational le\el, * 122 Personal Religious Orientation and Prejudice At this point, however, an important the-...i tical observation must be made. Low education may indeed predispose a person toward an cxelusionist, self-centered, extrinsic, religious orientation and may dispose him to a stereotyped, fearful image of jews. This fact does not in the least affect the functional relationship between the religious and the prejudiced outlooks. It is a c.wr.nn error for investigators to "control for" demographic factors ■ nlnait considering the danger involved in doing so. In so doing tht> are often obscuring and not illuminating the functional (i.e., psychological) relationships that obtain (see Allport, 1950). Folio« ing Wilson the task of direct measurement was taken up by Fcagin (1964) who used a more developed scale—one designed to measure not only extrinsic orientation but also the intrinsic. His scales arc essentially the same as those discussed in a later section of this paper. In his study of Southern Baptists Feagin reached four conclusions: (a) Contrary to expectation, extrinsic and intrinsic items did not fall on a unidimcnsional scale but represented two independent dimensions; (b) only the extrinsic orientation was related to intolerance toward Negroes; (c) orthodoxy as such was not related to the extrinsic or intrinsic orientation; (d) greater orthodoxy (fundamentalism of belief) did, however, relate positively to prejudice. Taking all these studies together we are justified in assuming that the inner experience of religion (what it means to the individual) is an important causal factor in developing a tolerant or a prejudiced outlook on life. Yet, additional evidence is always in place, and new insights can be gained by a closer inspection of the rather coarse relationships that have been established up to now. THE PRESENT STUDY We wished to employ an improved and broader measure of prejudice than had prev iously been used. And since direct measures of Prejudice (naming the target groups) have become too sensitive for "ide use, we wished to try some abbreviated indirect measures. Further, we wished to make use of an improved Extrinsic-Intrinsic scale, one that would give reliable measures of both extrinsic and 'ritrinsic tendencies in a person's religious life. For these reasons the following instruments were adopted. Social Problems Questionnaire This scale, devised by Harding and Schuman (unpublished1; 123 _.. W. Allport and ). M. Ross | see also Schuman & Harding, 1963, 1964). is a subtly worded instru- ■■. merit containing 12 anti-Negro, 11 anti-Jewish, and 10 anti-other items (pertaining to Orientals, Mexicans, and Puerto Ricans). The wording is varied so as to avoid an agreement response set. Indirect Prejudice Measures Six items were taken from Gilbert and Levinson's (1956) -' Custodial Mental Illness Ideology Scale (CMI). Example; "We should* be sympathetic with mental patients, but we cannot expect to under-1 stand their odd behavior, a) I definitely disagree, b) 1 tend to disagree. ^ c) í tend to agree, d) I definitely agree." i Four items are related to a "jungle" philosophy of life, suggesting,. a generalized suspiciousness and distrust. Example; "The world is f a hazardous place in which men are basically evil and dangerous, a) 11 definitely disagree, b) I tend to disagree, c) I tend to agree, d) I definitely s agree." I In all cases the most prejudiced response receives a score of 5 and '■>' the least prejudiced response, 1. No response was scored 3- | TABLE 3 I IN TtitCORR ELATION S BETWEEN FlVE MtASlRI S OF pRLJUDICť 1 _____ ________ '& Anti-Negro .65 .70 .20 Anti-Jewish .67 .24 Anti-Other .33 Jungle Nole. - \ = Í09. From Table 3 we see that while the indirect measures have aj positive correlation with each other and with direct measures, the rela-s tionship is scarcely high enough to warrant the substitution of the? indirect for the direct. The high correlations between prejudice fofj; the three ethnic target groups once again illustrate the well-established"* fact that ethnic prejudice tends to be a broadly generalized dispositions in personality. & Religious Orientation Measure í The full scale, entitled "Religious Orientation," is availably from ADI.* It separates the intrinsically worded items from thel .25 .31 36 .43 124 Personal Religious Orientalin., and Prejudice extrinsic, gives score values for each item, and reports on item reliabilities. In all cases a score of 1 indicates the most intrinsic response, a score of 5. the most extrinsic. While it is possible to use all 20 items as one continuous scale, it will soon become apparent that it is often wise to treat the two subscales separately, A sample item from the extrinsic subseale follows: "What religio. ;ifers me most is comfort when sorrows and misfortune strike, a) K -_.«*-itely disagree, 1. b)I tend to disagree, 2. c) I tend to agree, 4. d) I definitely agree, 5." A sample item from the intrinsic subseale: "My religious beliefs are what really lie behind my whole approach to life, a) this is definitely not so, 5. bj probably not so, 4. c) probably so, 2. d) definitely so, 1. ' SAMPLE While our sample of six groups of churchgoers shows some di\crsity of denomination and region, it is in no sense representative. Graduate-student members of a seminar collected the 309 cases from the following church groups: Group A, 94 Roman Catholic (Massachusetts); Group B, 55 Lutheran (New York State); Group C. 44 i\azarene (South Carolina); Group D, 53 Presbyterian (Pennsylvania); Group E, 35 Methodist (Tennessee); Group F, 28 Baptist (Massachusetts). "We labeled the groups alphabetically since such small sub-samples could not possibly lead to \alid generalizations concerning denominations as a whole. All subjects knew that they were invited to participate as members of a religious group, and this fact may well have introduced a "proreligious" bias. GROSS RESULTS If we pool all our cases for the purpose of correlating religious orientation with prejudice, we discover that while the findings are in the expected direction they are much less impressive than those of previous studies, especially Wilson's. Correlations with Extrinsic Subseale Since Wilson employed an extrinsic scale similar to ours, we hrst present in Table 4 our findings using this subseale and the various Measures of prejudice. Whereas Wilson found a correlation of .65 between his extrinsic and anti-Semitic measures, our correlation falls to •21. In part the reason no doubt lies in certain features of Wilson's Method winch we have criticized. 125 .riport and J. M. Ross Correlations with Combined Extrinsic-Intrinsic Scale From the outsel it was our intention to broaden Wilson's • undircctional (extrinsic) measure to see whether our hypothesis might hold for the total scale (combined scores for the 11 extrinsic and 9 -intrinsic items). As Table 5 shows, matter do not improve but seem to worsen. The logic of combining the two subscales is of course to augment the continuum in length and presumably enhance the reliability of the total measure.lt soon became apparent, however, that subjects who endorse extrinsically worded items do not necessarily ä reject those worded intrinsically, or vice versa. It turns out that there * is only a very low correlation in the expected direction between the two ^ subscales (r = .21). Obviously at this point some reformulation is ) badlv_neededL ......____— —------~—■—_-----------__ ~"1 REFORMULATION OF THE APPROACH I Examination of the data reveals that some subjects are indeed "consistently intrinsic,'" having a strong tendency to endorse intrin-' sically worded items and to reject the extrinsically worded. Corre-' spondingly others are "consistently extrinsic." Yet, unfortunately for our neat typology, many subjects are provokingly inconsistent. They persist in endorsing any or all items that to them seem favorable to religion in any sense. Their responses, therefore, are "indiscriminately proreligious." TABL£4 ' and Prejudice ŕ . ■' (.4- q--. Z£<- CORRELATIONS Bi lWEfcN" EXTRINSIC SlBSCALE Anti-Negro Anti-Jewish Anti-Other Jungle C Ml .26 .21 32 .29 .44 Note.—N = 309 TABLE 5 CORRILAIIONS BI ľW> EN ToTAl. ExiRrNSIC-lNTRINSIC St. AU AND pRLJlnlCF. ■* !' Anti-Negro Anti-Jewish Anti-Other lungle CMI .26 .18 .18 .21 .17 Nute -N - 4U9 126 Personal Religious Orientation and Prejudice The problem is essentially the same as that < rvui.ntered by the many investigators who have attempted to reverse the i. c r ding of items comprising the F scale, in order to escape an unwanted response-set bias. Uniformly the effort has proved to be frustrating, since so many subjects subscribe to both the positive and negative wording of the same question (sec Bass, 1955; Chapman & Bock. 1958; Chapman & Campbell, 1959; Christie, 1954; Jackson & M^i-.k, 1957). An example from our own subsc; is vould be: "My religious beliefs arc what really He behind my whole ^proaeh to life" (intrinsic). ' Though I believe in my religion. I feel there are many more important things in my life" (extrinsic). The approach used by Peabody (1961) offers us a model for analyzing our data in a meaningful way. Peabody administered both positive and negative F-scale items to subjects at two different testing sessions. By comparing each individual's responses to the same question stated positively at one time and in reverse at another he was able lo separate out those who were consistently pro or anti toward the content of authoritarian items. But he found many who expressed double agreement (or disagreement) with both versions of the same question. Table 6 applies Peabody's paradigm to our data. In assigning our 309 cases to these categories we employed the following criteria, ~Tntrinsk type includes individuals who agree with intrinsically worded items on the intrinsic subseale, and who disagree with extrin-sically stated items on the extrinsic subseale. By the scoring method employed these individuals fall below the median scores on both sub-scales. Extrinsic type includes individuals who agree with extrinsically stated items on the extrinsic subseale, and who disagree with items on the intrinsic subseale. By our scoring method these individuals all fall above the median scores on both subscales. Indiscriminately proreligious includes those who on the intrinsic snbscalc~ščore ať least Tí'point s less than on the extrinsic subseale. 'This figure reflects the fact that a subject gives approximately 50% more intrinsic responses on the intrinsic subseale than we should expect from his extrinsic responses to the extrinsic subseale.) Indiscriminately antireligious ur nonreligious includes those who "oukl show a strong tendency to disagree with items on both subscales. Mnee nonchurchgoers are excluded from our samples, such cases are n'>t iounci. (Some pilot work with markedly liberal groups indicates 'oat this type does exist, however, even among members of "religious" "»"gani/atjons.) 127 . Allport and J. M. Ross ■•-i* .TABLF>6,. _0, , ■ ■ For r Patterns of Rľi igiols Oriem ation M-i..,' ,«ŕ Agrees wish intrinsic choice Disaurees with intrinsic choice Agrees with extrinsic choice r' ' Disagrees with extrinsic choke lndiscrinünijjely proreligious Consistently intrinsic in tvpe C i insistently \, extrinsic in typej ^ Indiscriminately «'«^5? antirel igkitts or «*i^ nonrc-]igk>usa tl.i,'£'1'< *?sol found 111 prcvni samplr f ^ Table 1 gives the percentage of the three types. RESULTS OF THE REFORMULATION % The five measures of prejudice were analyzed by a 6 (Groups) X*J 3 (Religious Types) analysis of variance. Table 8 presents the overall effects for religious types for each of the five measures of prejudice. The multivariate analysis of variance indicates that there is both a^ significant difference between the three types of religious orientation,. and between the six subsamples in the level of Dreiudice.]_Examination^ of the means shows two trends: (a) The extrinsic type is more?. "prejudiced than the intrinsic type for both direct and indirect^ measures; (b) the indiscriminate type of religious orientation is rnoře^ prejudiced lhan either of the two consistent types. Statistically all^ these trends are highly significant. ^ * .. ■. '' ■ TABLE 7 ;.. . . PtHCiM vGE of Each Ri-liGii.it.'s Tvpi- in Each Slesampie Religious Consistently' Con s i st e ntl v Indiscriminately group N intrinsic extrinsic proreligious A (94) 36 34 30 B (55) 35 36 29 C (44) 36 39 25 D (53) 32 30 38 E !35) 31 29 40 F í 28) 39 39 22 128 Personal Religious Orientation and Prejudice ~i* ■ .■ , ■ -TABLES • ;. .- . PRFJUMCr \ND RfI IGIOL"S ORIENTATiOs Mean prejudice score rarset of F p re) u d ice Intrinsic Extrinsic Inconsistent tvpe ratio tvpe tvpe \ - 95 N = 108 \ = 106 Anti-Negro 28.7 33.0 36.0 8.6** Anti-lcwish 22.6 24.6 28.9 11.r* \nti-Other 20.4 23,3 26.1 10.9" Jungle 7.9 8.7 9.6 8.4" CMI 10.2 11,8 13.4 20.4"- Mi Liltivariatc analysis of variance Source of variation F ratio df Religious type (A) Sample groups (B) A x B 5.96"* 3.19"* 1.11* 10,574 25,668 50,1312 > • 21. "p -• .001 '"(I -■ .0005. We note especially that the scores of the indiscriminate type are markedly higher on all measures than the scores of the intrinsic type. Corresponding F ratios for paired comparisons range from 8.4 for the jungle scale to 20.4 for the CMI scale. The differences between the indiscriminate and extrinsic types are smaller. For the anti-Jewish and CMI scales these differences are, however, beyond the .005 level; for the anti-Negro the difference falls below significance. The relationship between the indiscriminately proreligious orientation and prejudice receives support (see Table 9) when we compare subjects who are moderately indiscriminate with those who are exircmely indiscriminate. (In the first group the scores on the intrinsic subscale average 16 points lower than on the extrinsic subscale, whereas the extreme cases average 23 points less on the intrinsic than on the extrinsic subscale.) The discovery that the degree of indiscriminateness tends to relate directly to the degree of prejudice is an important finding. It tan only mean that some functional relationship obtains between religious muddle headedness (for that is what indiscriminate scores imply) and antagonism toward ethnic groups. We shall return to this interpretation in the concluding section of this paper. 129 ,4. Allport and J. M. Ross RESULTS FOR SUBSAMPLES It would not be correct to assume that the variance is distributed equally over all the subsamples, for it turns out that the denominational groups differ appreciably in prejudice scores and in religious type, as Tables 10 and 11 indicate. TABLE 9 DEGKEESOi IMJISCRIMINATENESS WDAVERU.t PnhjLDici Scores Moderately F jitrem e] y Target of indiscriminate indiscriminate prejudice N - 56 N = 39 F ratio A n ti- Negro 35.4 37.9 .97 Anti-Jewish 28.0 30.1 .90 Anti-Other 24.9 28.2 3.25* Jungle 9.5 10.2 1.11 C MI 10.2 14.6 3.99' •p .. .05. , , TABLE 10 ^ Ami Negro Prejudice. Mean Scores on '::y • ------ Social Problems Sc ALE ".' ' '' ;. -o t Religious Intrinsic Ex Irin sic Indiscriminate group type type type Či roup M A 27.4 134) 34.8 (32) 32.2 (28) 31.4(94) B 27.2 (19) 32.3 (20) 31.9(16) 30.4(55) C 22.4 (16) 36.2(17) 35.0 Hi) 30.9 (44) D 35.5 (17) 28.7 (16) 42.5 (20) 36.1(53) E 40.5 (11) 35.5 (10) 43.0 (141 40.1(35) F 22.6 (11) 27.9(11) 28.7 ífi) 26.0(28) Tvpe AJ 28.7 '1081 33.0 (106) A nah ms of \ananci 36.0 (95! 32.5 1.309) •'-'■>•-,'. -. Source of var ration le lAl äf ,VÍ.S F ratio Religious tvp 2 1077.8 S.6" Religious group'.B) 3 952,2 7.6** A x B 10 251.1 2.0* Error <\\ i 291 125.6 It is true that when we combine subsamples all the trends; arc in 130 Personal Religious Orientation and Prejudice the expected direction, but troublesome excepii-ns occur for single groups as indicated by the nearly significant interaction effects. The must troublesome contradictions appear in relation to the anti-Negro measures based on the Harding-Schuman scale. Table 10 discloses certain sore points, even though the average trend over all the sub-samples is in the predicted direction. For Groups A, B, and C we m c that the indiscriminate type is slightly less prejudiced than the extr;.nsn. type, and for Groups D and F the extrinsic type seems actually less prejudiced than the intrinsic. (Groups D and E are consistently more troublesome than other sub-samples, perhaps because of some salient racial issue in the local community. It will be noted that both these groups are considerably more anti-Negro than the other subsamples.) By way of contrast we present in Table 11 the results for the short (five-item) CMI scale. With the exception of the indiscriminate typt in Group F, the progression of scores is precisely as expected. Ea.!. ^ubsample shows that the intrinsic type is less prejudiced toward the mentally ill than the extrinsic type- and the extrinsic type is less prejudiced than the indiscriminately proreligious." Indirect (CMI) Measl re of Prejudice Religious Intrinsic Extrinsic Indiscriminate eroup tvpe type tvpe Group ill A 11.2 (34) 12.4 (32) 13.6 (28) 12.3 (94) B 10.1 (19) 10.8 (20) 13.4 116) 11.3 (55) C 9.5 (16) 12.2 (17) 12.6 ill) 11.3(44) D 10.6 (17) 11.4 (16) 14.8 (20) 12.4 Í53) E 8.6 01) 12.9 (10) 13.6 '14) 11.8(351 F 9.2(11) 10.7 (11) 9.2 <6) 9.8 (281 Tvpe AI 10.2 (108) 11.8 (106) 13.4 (95) 11.9(309) Analysis of variance Source of \aridtiun df .VIS F ratio Religious tvpe (A) 2 255.0 20.4** Religious group (B) 3 36.5 2.9* A x B 10 15.3 1.2 Error l«i 291 12.5 T 05 "r .ooi. Returning in a different way to the original question of whether 131 ô. W. Allport and J. M. Ross consistent extrinsic and intrinsic orientations make for prejudice and for tolerance, rcspecthelv, we shall now examine this matter in each subsample separately. Inspection of the mean scores and variance for the total scale indicates that we are dealing with a relative!} narrow range of variation. To minimize the effect of a narrow range of scores and skewed distributions, we used Kendal's (1955) tau as a measure of degree of relationship between prejudice and consistent religious orientation. The results are given in Table 12. While the correlations -are not high (14 are significant in the expected direction), only one (in the troublesome Group E) is significant in the reverse direction. EDUCATIONAL DIFFERENCES Computing the actual years of schooling for all groups we find that the indiscriminate type has significantly less formal education than the intrinsic cases (p > .005, F = 18.29), and somewhat less than the extrinsic type (p > .10, F = 2.89). Comparing extrinsic with intrinsic types we find that the former has finished fewer years of schooling (p > .10, F = 3.45). (Oddly enough the groups with highest average education are D & E, which also displayed the highest anti-Negro and anti-Semitic prejudice—perhaps because of particular local conditions.) In our survey of earlier studies we saw that educational level is often a factor in the various relationships discovered between religion and prejudice. We have also argued that demographic factors of this sort should not be allowed to obscure the functional (psychological) analysis that the data call for. Granted that low education ; makes for indiscriminate thinking, the mental confusion that results from low education may have its own peculiar effects on religious and ethnic attitudes. TABLF 12 CORRl L-U'lONSBl TWl-tN CoMMNT D ExTRINSK IM RINi.lt Rt UGIOLS ScORtS (forConmsti nt Sľbjkltm andPkeildki 'Kind.a! s Tai i Religion s group Anti-Nťgro Anu-Jewish Anti-Othtr Jungle C MI A 31"* .26*** .24*** .14* .19*** B .19* .13 .15 .05 .03 C 32"** .17* 35*-" .14* .28'" D -.12 .05 - m .03 .11 E -.24* -.11 -.13 .26" 46*** F 39-*' ,1i .25* -.01 .24' "F 10 -r -.os "•f - 01 _ j! ■ " L ,f.'- ' . "X> 132 • Personal Religious Orkn'iition and Prejudice SUMMARY AND INTERPRETATIONS At the outset we stated three propositions that seem to be firmly ťstablished: (a) Churchgoers on the broad average harbor more ethnic prejudice than nonchurchgoers; (b) in spite of this broad tendency a curvilinear relationship in fact exisí; k■) the intrinsically motivated churchgoers are significantly less t s "ŕ i diced than the extrinsically motiv ated. Our present research supplies additional strong support for the second and third of these propositions. To these propositions we add a fourth: churchgoers who are indiscriminately proreligious are more prejudiced than the consistently extrinsic, and very much mure prejudiced than the consistently intrinsic types. The psychological tie between the intrinsic orientation and tolerance, and between the extrinsic orientation and prejudice, has been discussed in a series of papers by Allport (1959, 1963, 1966). In brief the argument holds that a person with an extrinsic religious orientation is using his religious views to provide security, comfort, status, or social support for himself—religion is not a value in its own right, it serves other needs, and it is a purely utilitarian formation. Now prejudice too is a "useful" formation: it too provides security, comfort, status, and social support. A life that is dependent on the supports of extrinsic religion is likely to be dependent on the supports of prejudice, hence our positive correlations between the extrinsic orientation and intolerance. Contrariwise, the intrinsic religious orientation is not an instrumental device, It is not a mere mode of conformity, nor a crutch, nor a tranquilizer, nor a bid for status. All needs aTe subordinated to an overarching religious commitment. In internalizing the total creed of his religion the individual necessarily internalizes its values of humility, compassion, and love of neighbor. In such a life (where religion is an intrinsic and dominant value) there is no place for rejection, contempt, or condescension toward one's fellow man. Such is our explanation for the relationship between extrinsic religion and prejudice, and between intrinsic religion and tolerance. Our present task is to discover, if we can, some similar functional tie hetween prejudice ias measured both directly and indirectly) and 'he indiscriminately proreligious orientation. The common factor seems to be a certain cognithe style. Technically it might be called undifferentiated thinking," or excessive "category width," as defined bv Pettigrew (1958). Rokeach (1960) notes the inability of the "dogmatic mind to perceive differences; thus, whereas some people 133 í. W. Allport and ]. M. Ross '■á distinguish in their thinking and feeling between Communists and a Naňs, the undifferentiated dogmatist has a global reaction (^cognitive . and emotional) toward "Communazis." We have no right, of course, to expect all our subjects to make discriminations exactly corresponding to our own logic. iNor should we <-> expect them to read and respond to every item on the Extrinsic- -i Intrinsic scale according to its full meaning as intended by the in- ■& vestigators. Perhaps we should he gratified that two-thirds of our cases & can be safely classified as "consistent" (i.e., having about the same p strength of disposition toward an extrinsic or intrinsic orientation % across most of the items). These consistent eases, as we have seen, s support the hypothesis with which we started. It is the remaining § (indiscriminate) one-third of the cases which obscure the trend (orl diminish its statistical significance). 1 In responding to the religious items these individuals seem to % take a superficial or "hit and run" approach. Their mental set seems W to be "all religion is good." "My religious beliefs are what really lie* behind my whole life"—\es! "Although I believe in my religion, I feel s there are many more important things in my life"—Yes! "Religion* is especially important to me because it answers many questions about 1 the meaning of life"—Yes! "The church is most important as a place | to formulate good social relationships"—Yes! M There seems to be one wide category—"religion is OK." Tromi the way in which the scale is constructed this undifferentiated» endorsement can be the product of an agreement response set. Our a inconsistently proreligious may be "yeasayers" í Couch & Keniston,« 1960). But if so, we are still dealing with an undifferentiated! cognitive disposition. We recall likewise that the inconsistent cases!« have a lower level of formal education than the consistent cases. Thisa factor also is relevant to the formation and holding of overwidfB categories. g| But why should such a disposition, whatever its source, be sdl strongly related to prejudice, in such a way that the more, undifa ferentiated, the more prejudiced—as Table 9 shows? -a The answer is that prejudice itself is a matter of stereotype» over generalization, a failure to distinguish members of a minorit! group as individuals (Allport, 1954, Chaps. 2, 10). It goes withoin saying that if categories are overwide the accompanying feeling ton"! will be undifferentiated. Thus, religion as a «hole is good; a minorit] group as a whole is bad. . It seems probable that people with undifferentiated styles ti thinking (and feeling! are not entirely secure in a world that fot th 134 Personal Religious Orientation and Prejudice most part demands fine and accurate distinct >us. The resulting diffuse anxiety may well dispose them to grapple onto religion and to distrust strange ethnic groups. The positive correlation between the jungle items and other prejudice scales (Table i) is evidence for this interpretation. Our line of reasoning, reade:- will recognize, is compatible with various previous contributions io t be theory of prejudice. One thinks here of Rokeach's concept oi Jogmatism; of Schuman and Harding's (1964) discovery of a "confused" type in their study of the relation between rational consistency and prejudice; of the same authors' work on sympathetic identification (1963); of studies on the dynamics of scapegoating, the role in insecurity, of authoritarian submission, of intolerance for ambiguity, and of related concepts. All in all we conclude that prejudice, like tolerance, is often embedded deeply in personality structure and is reflected in a consistent cognitive style, Both states of mind are enmeshed with the individual's religious orientation. One definable style marks the indh idual who is bigoted in ethnic matters and extrinsic in his religious orientation. Equally apparent is the style of those who are bigoted and at the same time indiscriminately proreligious. A relatively small number of people show an equally consistent cognitive style in their simultaneous commitment to religion as a dominant, intrinsic value and ethnic tolerance. One final word: our research argues strongly that social scientists who employ the variable "religion" or "religiosity" in the future will do well to keep in mind the crucial distinction between religious attitudes that arc intrinsic, extrinsic, and indiscriminately pro. To know that a person is in some sense "religious" is not as important as to know the role religion plays in the economy of his life. (The categories of nonreligious and indiscriminately antireligious will also hir some purposes be of central significance, although the present research, confined as it is to churchgoers, does not employ them.) NOTES AND REFERENCES 'J Hauling and H. Schuman. "Social Problems Questionnaire." Cornell Universit) . I In. full Religious Orientation scale has been deposited u iih the American Documentation Institute. Order Document No. 926Ü from \D1 \u*ihary Publications Project fbiitdduplicaiion Service, I ibrarv uf Conuress, Washington, D. C, 20^-10 Remit ii ai!\ance Si.25 for microfilm or SI.25 for photocopies and make checks pavahle to Chief, Photoduplication Sen ice. Library of Congress. 'he multivariate F reported here is Willi's lambdu ' \ndtrson. 1958 . Statistics! computations are summamed by Bock tl%ii and programmed for the IBM 7090 b> ■**!' >nd Cramer 11962 j. The univariate tests to be reported are adjusted for unequa. 13!