PERSONALITY AND SOCIAL PSYCHOLOGY BULLETIN McCann / AMERICAN CHURCH MEMBERSHIPS Threatening Times and Fluctuations in American Church Memberships Stewart J. H. McCann University College of Cape Breton Building on the work of S. M. Sales, who related contemporaneous economic threat to authoritarian behaviors, two studies tested the hypothesis that threat is associated with relatively more attraction to authoritarian churches and less attraction to nonauthoritarian churches. The hypothesis was supported in Study 1, when the annual percentage of changes in memberships (1928-1986) for two authoritarian and two nonauthoritarian denominations were examined in relation to several annual social, economic, and political threat indices, and in Study 2, when changes in the proportion of the population having membership in 25 representative denominations were examined over periods of relatively low threat (1955-1964), high threat (1965-1974), and low threat (1965-1979). Both studies suggest that social and political threat as well as economic threat may activate authoritarian behaviors. Is religion a citadel of hope built on the edge of despair or the opiate of the people as Reinhold Neibuhr and Karl Marx, respectively, assumed? Perhaps not. Although it is rather commonplace to assume that there is a general tendency for people to turn to religion in times of stress, Sales (1972) concluded that only churches with certain qualities become more appealing during threatening times, whereas other churches lose their attractiveness. Specifically, during economically threatening years, Sales found that the rate of conversions to authoritarian churches increased, and the rate of conversions to nonauthoritarian churches decreased. His hypothesis was derived from the theory that threatening circumstances lead to higher levels of authoritarianism (e.g., Fromm, 1941; Rokeach, 1960; Sanford, 1966; Wilkinson, 1972) and consequently to more persons being attracted to relatively authoritarian organizations. According to Sales (1972), the Seventh-day Adventist Church, the Southern Baptist Convention, the Church of Jesus Christ of Latter-day Saints, and the Roman 325 Catholic Church could be labeled authoritarian, whereas the Presbyterian Church in the United States of America, the Protestant Episcopal Church, the Congregational Christian Church, and the Northern Baptist Convention could be classified as nonauthoritarian. 1 Sales was able to categorize denominations in this fashion because churches do appear to differ in ways similar to the individual differences articulated in the components of the authoritarian personality and because the organizational characteristics do tend to form coherent patterns analogous to the individual level syndrome: Some churches demand absolute obedience either to the leadership of the church or to the Divine (authoritarian submission); others allow the parishioner more leeway in his decisions. Some churches condemn disbelievers and heretics (authoritarian aggression); others assume a more tolerant attitude. Some churches emphasize the mystical aspects of religion and apply a literal interpretation of scripture (superstition), while others stress a more intellectualized approach. Some churches exhibit strong concern about sin (including sexuality); others are much less exercised over this issue. Some churches believe that there is only one true church, and that members of other denominations have been somehow misled (ethnocentrism); others feel that there are many equally acceptable forms of belief. Finally, some churches stress the unimaginable might (and potential punitiveness) of God, while others view the Divine more as a friend and helper. (Sales, 1972, p. 422) Author s Note: This research was supported in part by CERP Grant 970-00-1935 at the University College of Cape Breton. Correspondence concerning this article should be addressed to Dr. Stewart J. H. McCann, Department of Behavioral and Life Sciences, University College of Cape Breton, P.O. Box 5300, Sydney, Nova Scotia, Canada B1P 6L2. PSPB, Vol. 25 No. 3, March 1999 325-336 1999 by the Society for Personality and Social Psychology, Inc.
326 PERSONALITY AND SOCIAL PSYCHOLOGY BULLETIN Sales (1972) calculated conversion ratios that indicate the percentage of nonmembers of each church who converted to that church during any given year (p. 423). It is important to recognize that such ratios may reflect gains in interest in particular denominations by outsiders but do not necessarily reflect losses in interest within particular denominations. A church with relatively little extended holding power and, hence, a high turnover might have an elevated conversion ratio but a relatively stable total number of adherents even though the membership is in flux. Or a church with ample holding power might have a low conversion ratio but show a membership increase. If membership and conversion figures do not rise and fall in unison, does membership also fluctuate according to the basic formulation of Sales (1972)? Focusing the hypothesis on church memberships might facilitate an explanation of the waxes and wanes of denominational fortunes, which is a long-standing issue among those interested in the scientific study of religious affiliation (e.g., Hoge & Roozen, 1979; Kelley, 1972; Stark & Glock, 1968). Sales (1972) dwelled exclusively on economic threat, defined operationally as per capita disposable income, and selected the years from 1920 to 1939 partly because no major noneconomic threats existed during the period which might cloud the anticipated relationship between economic conditions and conversions to the denominations (p. 422). However, noneconomic threats also have been theoretically and empirically linked to increases in authoritarian behavior. For instance, Sales and Friend (1973) experimentally demonstrated that failure-induced threat led to increases in authoritarian responses, whereas success led to decreases; Peterson, Doty, and Winter (1993) focused on social issues such as AIDS and drug use as triggers for authoritarian behavior because they pose threats to the American way of life; Altemeyer (1988) used scenarios to manipulate political circumstances 20 years in the future and found differences in the Right-Wing Authoritarianism (RWA) scale scores of students who answered as they thought they would under the hypothetical conditions; and McCann (1997) found that the margin of popular vote victory in U.S. presidential elections is related to an interaction between the strength in the personality of the winner and the combined degree of social, economic, and political threat during election years. Therefore, measures of social and political threat as well as economic threat were included in the current work to provide a more global test of the general hypothesis that increased threat is related to heightened authoritarian behavior and, in this context, to a stronger attraction to authoritarian churches and a weaker attraction to nonauthoritarian churches. This strategy also facilitated comparative tests to explore the relative contributions of various social, economic, and political threat variables to fluctuations in membership figures. The central objective of the two studies reported here was to test the hypothesis that threat is associated with larger increases or smaller decreases in the membership of authoritarian denominations and with larger decreases or smaller increases in the membership of nonauthoritarian denominations. In each study, the primary threat variable was the social, economic, and political threat (SEPT) index (McCann & Stewin, 1989, 1990, 1991), which provides single annual quantitative estimates of prevailing social, economic, and political threat taken as a whole. Secondary threat variables included separate annual social, economic, and political threat ratings obtained from a survey of historians (McCann, 1996), five common economic variables, and a composite of the five. Study 1 examined annual membership changes in two authoritarian and two nonauthoritarian Protestant denominations over the years from 1928 to 1986. Study 2 investigated membership changes in a representative sample of 25 authoritarian and nonauthoritarian denominations during three 10-year periods from 1955 to 1985 that differed in level of threat. STUDY 1 The notion that contemporaneous situational threat is a cause of heightened authoritarianism has existed in some form in authoritarian personality literature from the beginning (e.g., Fromm, 1941; Rokeach, 1960; Sanford, 1966; Wilkinson, 1972). There also is empirical evidence, mostly of a correlational nature, that is consistent with this position (e.g., Doty, Peterson, & Winter, 1991; Jorgenson, 1975; McCann, 1997; McCann & Stewin, 1984, 1990; Padgett & Jorgenson, 1982; Peterson et al., 1993; Sales, 1972, 1973; Sales & Friend, 1973). The theoretical and empirical contributions have fostered fairly broad acceptance of a link between threat and authoritarianism (e.g., Altemeyer, 1988; Doty et al., 1991; McCann & Stewin, 1990; Meloen, 1983; Osterreich, 1992; Peterson et al., 1993; Sales, 1973; Simonton, 1990; Stone & Smith, 1993). Adorno, Frenkel-Brunswik, Levinson, and Sanford (1950) did not dwell on the direct effects of contemporaneous threat on levels of authoritarianism, but other psychodynamic theorists have drawn explicit connections. For example, Wilkinson (1972, pp. 103-111) put forth a composite theory of authoritarian cause with three elements having reciprocal impacts: (a) social flux and threats to culture, (b) personality disposition and behavior, and (c) cultural orthodoxy and traditional ideology. In the face of threat, the model also allows that
McCann / AMERICAN CHURCH MEMBERSHIPS 327 many people will become more intolerant, rigid, hostile to alien influences and reverent to traditional authority, without necessarily becoming more authoritarian in terms of inner personality structure (p. 106). Consequently, both authoritarians and nonauthoritarians are prone to display more authoritarian attitudes and behaviors in the face of threat. In Altemeyer s (1988) redefinition of authoritarian personality theory, authoritarianism is developed and maintained through principles of social learning such as direct reinforcement, modeling, and vicarious reinforcement (Bandura, 1977). To Altemeyer (1988), the level of authoritarianism can change as long as new experiences are possible, as long as new models can emerge, as long as reinforcement contingencies can be upset and sudden changes in the individual and a society can occur because of economic calamity, for example, or domestic upheaval, or war (p. 60). Clearly, threat can influence the attitudes and behavior of both authoritarian and nonauthoritarian persons. Do members of authoritarian denominations score higher than do members of nonauthoritarian denominations on measures of authoritarianism? Yes. For example, Altemeyer (1988) found that a bankable difference appears in study after study (p. 201) between the relatively high-scoring Jehovah s Witnesses, Pentecostals, and Baptists and the low-scoring Anglicans and United Church members, with Lutherans falling in the middle. Earlier research (e.g., Gregory, 1957; Photiadis & Johnson, 1963) found positive correlations between F-scale measures of authoritarianism and religious orthodoxy, which is related to church placement on the authoritarian-nonauthoritarian dimension (Fagan & Breed, 1970; Meadow & Kahoe, 1984, p. 249). Generally, there is a rather strong correlation between authoritarianism and fundamentalism (Altemeyer & Hunsberger, 1992). How were churches chosen for Study 1? Of Sales s (1972) authoritarian denominations, conversions to the Seventh-day Adventist Church showed the highest positive correlation with threat, and conversions to the Southern Baptist Convention showed no significant correlation. Of nonauthoritarian churches, conversions to what is now the United Presbyterian Church showed the highest negative correlation, and conversions to the Protestant Episcopal Church showed the lowest negative correlation (p <.10). Because consistent data for the years 1928 to 1986 also are readily available for these churches but not for others in Sales s study, these four with conversion ratios most and least related to economic threat were chosen. The issue of whether threat is related to aggregate church membership changes was not empirically addressed by Sales (1972). However, he did note that most studies have shown no relationship between threat and the appeal of religion in general. In the current study, inclusive church membership fluctuations were analyzed in relation to the various threat variables. Method and Results ANNUAL PERCENTAGE CHANGES IN CHURCH MEMBERSHIP Seventh-day Adventist, Southern Baptist Convention, United Presbyterian, and Protestant Episcopal membership figures were obtained from Historical Statistics of the United States: Colonial Times to 1970 (U.S. Bureau of the Census, 1975) and the Statistical Abstract of the United States (U.S. Bureau of the Census, several years 2 ). Annual total resident population figures also were taken from these sources. For each denomination, the annual membership figure was divided by population, and the annual percentage change in membership (as a fraction of the population) was computed for the years 1928 to 1986. For United Presbyterians, membership change data for the merger years of 1958 and 1983 were necessarily excluded. Total annual American church membership figures were taken from a compilation by Jacquet (1987, p. 256). The available annual figures for inclusive membership were divided by population, and the percentage changes from year to year were calculated where possible. This procedure resulted in aggregate church membership change data for 45 years between 1932 and 1985. THE INDICES OF SOCIAL, ECONOMIC, AND POLITICAL THREAT The SEPT index. The main threat variable was the McCann and Stewin (1989, 1990, 1991) quantitative American SEPT index. The SEPT index is based on the judgements of 196 American historians who were asked to rate each of the years from 1920 to 1986 on a 7-point scale according to the degree to which they thought the events of the year could be perceived as threatening to the established order and way of life in the United States. The interrater reliability of the index is high, test-retest reliability has been shown to be adequate, and validity has been demonstrated through significant correlations with threat indicators such as the unemployment rate, suicide rate, number of military personnel on active duty, and stock price fluctuations (see McCann & Stewin, 1990). Studies involving threat also have been replicated and extended using the SEPT index (e.g., see McCann, 1991; McCann & Stewin, 1990). Separate social, economic, and political threat ratings. The development of the separate social, economic, and political threat ratings (McCann, 1996) followed the same approach used to produce the SEPT index (McCann & Stewin, 1990). A mailing list was prepared by the Ameri-
328 PERSONALITY AND SOCIAL PSYCHOLOGY BULLETIN can Historical Association that included every second person on their master file who was an assistant, associate, or full professor in the United States who had indicated an interest in one or more relevant areas such as U.S. economic history and U.S. cultural history. 3 From the list of 1,957 professors, 125 were selected by picking approximately every 16th person. In November and December of 1993, the professors anonymously responded to a simple, subjective, annual threat-weighting scale very similar to the one used by McCann and Stewin (1990) which began with the following instructions: Because of various social, political, and economic circumstances, there have been years since 1920 which generally can be viewed as very threatening to the American established order and way of life and years which generally can be viewed as very nonthreatening to the established order and way of life in the United States. I would like you to rate each of the following years from 1920 to 1992 according to the degree to which you feel it can be perceived as economically, politically, and socially threatening to the established order and way of life in the United States. Respondents circled one digit ranging from 1 (very nonthreatening) to 7(very threatening) on the economic, the political, and the social measure for each year from 1920 to 1992. Nineteen professors returned forms for a response rate of 15.2%. However, complete data for every year were provided by only 14 respondents for an effective return rate of 11.2%. The mean ratings of the 14 raters who provided complete data became the annual social, economic, and political threat values. Interrater reliability was high despite the low return rate. For social threat, each rater s responses were significantly correlated with the resulting index, and only one of the 14 coefficients was at a significance level greater than.001. Correlations ranged from.39 to.84, with a median of.63, and Cronbach s alpha was.87. For economic threat, each rater s responses correlated with the index at the.001 level; correlations ranged from.43 to.89 with a median of.78, and Cronbach s alpha was.94. For political threat, one rater s responses were not significantly correlated with the index, but the ratings of 11 were at the.001 level. Correlations ranged from.01 to.75, with a median of.54. Cronbach s alpha was.78. Social threat correlated with economic threat, r(57) =.49, p <.001, and political threat, r(57) =.56, p <.001, but economic and political threat were not correlated, r(57) =.16, ns. The SEPT index correlated with social threat, r(57) =.55, p <.001; economic threat, r(57) =.44, p <.001; and political threat, r(57) =.60, p <.001. Objective economic threat indicators. Annual unemployment rates for 1928 to 1986 were taken from Historical Statistics of the United States: Colonial Times to 1970 (U.S. Bureau of the Census, 1975) and the Statistical Abstract of the United States (U.S. Bureau of the Census, several years). Data from the same sources were used to calculate annual percentage changes in stock prices, housing starts, number of concerns in business, and gross national product (GNP). The composite economic index was created by converting the unemployment rate and four other objective economic change variables to z scores, adding the standardized housing starts, stock prices, business concerns, and GNP change variables and subtracting the standardized unemployment rate. The resulting composite correlated with the unemployment rate, r(57) =.42, p <.001, and with percentage changes in housing starts, r(57) =.62, p <.001; stock prices, r(57) =.64, p <.001; business concerns, r(57) =.55, p <.001; and GNP, r(57) =.46, p <.001. ANALYSES The main analyses were conducted on a maximum common data base (N = 57) containing all the years from 1928 to 1986 except the major Presbyterian merger years of 1958 and 1983. The database for the analyses of aggregate American church membership included the years for which membership change data could be calculated from 1932 to 1938 and 1948 to 1985 (N = 45). For each denomination and for inclusive church membership, correlations between annual percentage changes in membership and SEPT index scores, separate social, economic, and political threat ratings, and economic indicators were computed. Further computations indicated that year was negatively correlated with membership changes for Seventh-day Adventists, r(55) =.29, p <.05; Southern Baptists, r(55) =.53, p <.001; Presbyterians, r(55) =.57, p <.001; Episcopalians, r(55) =.57, p <.001; and inclusive church membership, r(55) =.41, p <.01. Year was also correlated with SEPT scores, r(55) =.39, p <.01; economic threat ratings, r(55) =.31, p <.05; and unemployment rate, r(55) =.42, p <.001. Consequently, partial correlations between threat variables and denominational membership changes were computed with year controlled (see Table 1). Partial correlations between the SEPT index and membership changes were in the expected direction and significant for each denomination. Separate social, economic, and political threat ratings also correlated significantly in the anticipated manner with membership changes for Seventh-day Adventists, Presbyterians, and Episcopalians; for Southern Baptists, only political threat ratings correlated significantly in the expected di-
McCann / AMERICAN CHURCH MEMBERSHIPS 329 TABLE 1: Pearson and Partial Correlations (with year controlled) Between Church Membership Changes and Threat Indicators Church Threat Indicator Pearson Partial Seventh-day SEPT ratings.62****.58**** Adventist Social threat ratings.30**.31**** Economic threat ratings.52****.47**** Political threat ratings.34***.33*** Economic composite.60****.58**** Unemployment rate.61****.57**** Housing index.29**.32*** Stock prices.33***.32*** Business enterprises.38***.38*** Gross National Product.05.04 Southern Baptist SEPT ratings.40****.25** Social threat ratings (.18) (.23) Economic threat ratings (.00) (.21) Political threat ratings.24**.22** Economic composite (.20) (.12) Unemployment rate.27**.07 Housing index.13.20 Stock prices.10.05 Business enterprises.27**.28** Gross National Product (.20) (.26) Presbyterian SEPT ratings.01.36*** Social threat ratings.40****.50**** Economic threat ratings.29**.59**** Political threat ratings.20.32*** Economic composite.08.24** Unemployment rate.07.41*** Housing index (.08) (.14) Stock prices (.11).21 Business enterprises (.09) (.07) Gross National Product.20.26** Episcopalian SEPT ratings.05.28** Social threat ratings.29**.36*** Economic threat ratings.07.32*** Political threat ratings.13.23** Economic composite (.19) (.09) Unemployment rate (.09).19 Housing index (.18) (.27**) Stock prices (.02).05 Business enterprises (.23) (.24) Gross National Product.01.03 Inclusive SEPT ratings (.03) (.19) membership a Social threat ratings (.27**) (.33**) Economic threat ratings (.25**) (.43***) Political threat ratings.03 (.09) Economic composite (.06) (.22) Unemployment rate.02 (.30**) Housing index (.07) (.03) Stock prices (.16) (.24) Business enterprises.09.03 Gross National Product (.06) (.07) NOTE: Parentheses indicate that the correlation is not in the expected direction. SEPT = social, economic, and political threat. a. N = 45 (rather than 57). **p <.05. ***p <.01. ****p <.001. rection with membership changes. Except for Seventhday Adventists, partial correlations between objective economic variables and membership changes were relatively small and inconsistent. In regard to inclusive church membership changes, the significant partial correlations for social threat, economic threat, and unemployment rate as well as the signs for six of the seven nonsignificant correlation suggest a decline in total church memberships in times of threat (see Table 1). To determine whether the relationship between annual percentage changes in denominational membership and threat can be accounted for by economic threat alone, multiple regression equations were computed for each denomination by entering year first, the composite economic variable and the economic threat ratings stepwise second, the social and political threat ratings stepwise third, and the SEPT index last. Social or political threat accounted for additional variance for Seventh-day Adventists, Presbyterians, and Episcopalians in the anticipated fashion; SEPT similarly accounted for additional variance for Presbyterians, Episcopalians, and Southern Baptists (see Table 2). Discussion Study 1 results support the general hypothesis. Annual threat levels were positively related to annual percentage changes in membership in more authoritarian denominations, negatively related to annual percentage changes in membership in less authoritarian denominations, and not related to annual percentage changes in inclusive church membership. Contrary to conventional wisdom and to Argyle s (1958) view that there is very little relation between religion and economic prosperity (p. 139), aggregate church memberships actually may fall rather than rise or remain constant under threatening conditions. Significant negative correlations between inclusive membership changes were found for economic and social threat ratings as well as for unemployment rates, and the sign was negative for six of the seven other threat indices. In addition to extending Sales s (1972) pattern of findings from temporal fluctuations in church conversion ratios to changes in denominational membership figures, the current study demonstrates that such changes may not result solely from dynamics involving economic threat. Regression analysis revealed that the social threat ratings, the political threat ratings, and the SEPT index could account for additional variance in denominational membership changes when economic threat ratings and the composite economic index were statistically controlled. One apparent weakness of Study 1 is that only 11.2% of the professors returned complete social, economic, and political threat ratings. In comparison, the effective return rate for the SEPT index was 25.7% (McCann & Stewin, 1990). Part of the reason for the discrepancy may
330 PERSONALITY AND SOCIAL PSYCHOLOGY BULLETIN TABLE 2: lie in the fact that the current task was more daunting, requiring 219 separate ratings on three dimensions rather than 67 ratings on a single global one. Nevertheless, the new indexes were usable. There was high interrater reliability, correlations between the separate indexes and the SEPT index displayed direction and magnitude that might be expected, and the substantive analyses using the indexes yielded anticipated and interpretable results. It should be noted that this was a rating exercise: The number of raters was lower than hoped for, but the number of cases (i.e., years) remained substantial at 73. STUDY 2 Study 2 was carried out to determine whether a pattern of results similar to that found in Study 1 could be established for a larger sample of authoritarian and nonauthoritarian churches. The replication involved more than two dozen denominations, the 10 threat variables used in Study 1, and membership change data over each of three consecutive 10-year periods between 1955 and 1984. The three 10-year periods generally progressed from low threat to high threat to low threat, but it appeared that economic threat was not a key feature of the transitions, making it again possible to explore the significance of social and political threat relative to economic threat in this context. Method and Results Variance Accounted for by Social Threat, Political Threat, and SEPT Ratings With Year and Economic Threat Controlled Church Variable R 2 R 2 Change Seventh-day Year.08**.08** Adventist Economic composite.39****.31**** Political threat ratings.45****.06** SEPT ratings.53****.07*** Southern Baptist Year.28****.28**** SEPT ratings.32***.04* Presbyterian Year.32****.32**** Economic threat ratings.56****.24**** Political threat ratings.60****.04** Episcopalian Year.32****.32**** Economic threat ratings.39****.07** Economic composite a.46****.07** Social threat ratings.51****.05** SEPT ratings.56****.05** a. The contribution is not in the expected direction. *p <.07. **p <.05. ***p <.01. ****p <.001. DENOMINATIONS For 25 denominations, membership data for 1955, 1965, 1975, and 1985 were selected from a compilation by Jacquet (1987, pp. 254-255). Jacquet provided membership data for 5-year intervals from 1950 to 1985 and for 1940, 1947, and 1984 on 29 U.S. denominations. To Jacquet, the 29 represent a cross-section of theological orientations, ecclesiastical structures, and geographical foci and also satisfy the criterion of reliability and completeness in statistical reporting (p. 254). The Assemblies of God, the Christian and Missionary Alliance, and the North American Baptist Conference were excluded because of pronounced changes over time in the basis of reporting. The Roman Catholic Church was dropped because the relaxed member demands imposed by Vatican Council II may have changed it from a more authoritarian to a less authoritarian church during the 1955 to 1985 period (Kelley, 1978, pp. 166-167). Denominations retained for Study 2 are listed in Table 4. Before proceeding with analyses, certain relevant omissions and anomalies in Jacquet s (1987) table were remedied. For the Church of God (Cleveland, Tennessee), an omitted membership figure for 1985 was taken from Famighetti (1985). For the United Methodist Church, an omitted membership figure for 1985 was obtained from Jacquet and Jones (1991, p. 273). For the Episcopal Church, Jacquet (1987) included the available figures for 1956 and 1966 rather than for 1955 and 1965. To develop an estimate for 1955, the 1956 figure was subtracted from the 1960 figure, divided by four, multiplied by five, and subtracted from the 1960 figure. For 1965, because the peak in membership occurred in 1966 (Kelley, 1972, p. 4), the 1960 figure was subtracted from the 1966 figure, divided by six, multiplied by five, and added to the 1960 figure. Authoritarian and nonauthoritarian denominations. Each denomination was categorized as authoritarian or nonauthoritarian based on information from Kelley (1972, 1978), Hoge (1979), Hudson (1961), Ahlstrom (1972), Bibby (1978), Jacquet (1987), and Peele (1984). Evidence for classification of many denominations was found in several sources, but only one reference is cited for most churches in the groupings that follow. Kelley (1972) placed a number of denominations along a strictness gradient, having characteristics with a familiar authoritarian ring at one end and a less authoritarian flavor at the other. To Kelley, churches toward the strict end are characterized by discipline (e.g., willingness to obey the commands of...leadership without question ), conformity (e.g., intolerance of deviance or dissent ), absolutism (e.g., closed system of meaning and value which explains everything ), commitment (e.g., willingness to sacrifice status, possessions, safety, life itself, for the cause or the company of the faithful ), fanaticism (e.g., keep yourselves unspotted from the world... cloister ), and missionary zeal (e.g., eagerness
McCann / AMERICAN CHURCH MEMBERSHIPS 331 to tell the good news of one s experience of salvation to others ). Churches at the lenient end display individualism (e.g., unwillingness to give unquestioning obedience to anyone ), diversity (e.g., appreciation of individual differences ), dialogue (e.g., appreciative of outsiders rather than judgemental ), relativism (e.g., belief that no one has a monopoly on truth; that all insights are partial ), reserve (e.g., reluctance to expose one s personal beliefs or to impose them on others ), and lukewarmness (e.g., reluctance to sacrifice all for any single set of values ) (all quotes found in Kelly, 1972, p. 84). Kelley s strictness gradient was empirically tested by Hoge (1979), who had 17 denominations rated by 21 experts on eight different dimensions derived from Kelley s writings. Close correspondence was found between the ratings and Kelley s strictness placements. On the basis of Kelley s (1972) strictness gradient, eight churches were classified for Study 2 as authoritarian (i.e., Jehovah s Witnesses, the two Church of God denominations, Church of Jesus Christ of Latter-day Saints, Reorganized Church of Jesus Christ of Latter-day Saints, Seventh-day Adventists, Southern Baptist Convention, Lutheran Church Missouri Synod); seven others were classified as nonauthoritarian (i.e., United Church of Christ, United Methodist Church, Presbyterian Church U.S.A., Episcopal Church, Reformed Church in America, Lutheran Church in America, American Lutheran Church). The American Lutheran Church was more or less at the middle on Kelley s gradient, but Hoge s (1979) results clearly put it in the nonauthoritarian group. Further strictness gradient inclusions (Kelley, 1978) justified categorizing the Church of the Nazarene and the Salvation Army as authoritarian. Hoge s (1979) analysis placed the Christian Church (Disciples of Christ) in the nonauthoritarian group. The Free Methodist Church of North America, Wisconsin Evangelical Lutheran Synod, Baptist General Conference, and Evangelical Covenant Church were put in the authoritarian category on the basis of Hudson (1961), who referred to the Free Methodists and the Wisconsin Evangelical Lutherans as conservative Protestants (p. 161) and noted that even greater affinities with Fundamentalism have been displayed by (p. 162) those in the Baptist General Conference and the Evangelical Covenant Church. Ahlstrom (1972) stated that Cumberland Presbyterians piety and church life was more closely related to the South s evangelical mainstream than that of other Presbyterian bodies (p. 726), so it also was classified as authoritarian. Bibby (1978) placed the Mennonite Church with other conservative churches such as the Church of the Nazarene, the Church of God, and the Salvation Army, so it too was placed in the authoritarian group. Finally, the Church of the Brethren was classified as nonauthoritarian because of its membership in the National Council of the Churches of Christ in the U.S.A. (Jacquet, 1987), a criterion often used for separating liberal and conservative churches (e.g., Peele, 1984). RELATIVE THREAT DURING THE THREE 10-YEAR PERIODS The database allowed the examination of changes in membership during three 10-year periods that differed in level of societal threat. To provide empirical verification for my perception that the 1965 to 1974 period was more threatening in a general sense but perhaps somewhat less threatening economically than the 1955 to 1964 and 1975 to 1984 periods, comparisons of the three periods were made on the 10 objective and subjective threat variables used in Study 1 (see Table 3). Planned t tests of the differences between the means of the 10 variables for the three time periods showed that SEPT scores, social threat ratings, and political threat ratings were significantly higher during the 1965 to 1974 period than during either of the other periods. But as expected, the seven economic threat variables did not fluctuate in this way. Economic threat ratings were significantly higher in the 1965 to 1974 period than during the 1955 to 1964 period but also were nonsignificantly higher in the 1975 to 1984 period than in the 1965 to 1974 period. Unemployment was lowest during the 1965 to 1974 period. The four other objective economic indices and the composite showed no significant differences across periods CALCULATION OF PERCENTAGE CHANGES IN CHURCH MEMBERSHIP Membership change variables were calculated from data for 1955, 1965, 1975, and 1985 (Jacquet, 1987, pp. 254-255). A percentage change in membership variable adjusted for population increases was computed. For each denomination, the membership figures for 1955, 1965, 1975, and 1985 were divided by the total population figures for those years (U.S. Bureau of the Census, 1986, p. 5), and the percentage changes from 1955 to 1965, 1965 to 1975, and 1975 to 1985 in the proportion of the population who were members of the denomination were calculated. PREDICTIONS The general hypothesis led to two predictions. Both clearly tested the hypothesis, but the first was somewhat more stringent. The first was that the authoritarian denominations incurred their greatest membership gains or smallest losses, whereas the nonauthoritarian denominations incurred their smallest gains or greatest losses during the more threatening 1965 to 1974 period. In other words, the prediction was that the peak for authoritarian denominations and the trough for nonau-
332 PERSONALITY AND SOCIAL PSYCHOLOGY BULLETIN TABLE 3: Comparisons of Means on Social, Economic, and Political Threat Indicators in Periods of High and Low Threat Low Threat High Threat Low Threat Threat Indicator 1955-1964 Difference t 1 1965-1974 Difference t 2 1975-1984 SEPT ratings 3.42 7.66**** 4.90 8.17**** 3.67 Social threat ratings 3.55 5.58**** 5.19 8.10**** 3.60 Economic threat ratings 2.67 6.30**** 3.72 1.54 4.06 Political threat ratings 3.66 5.77**** 4.74 6.78**** 3.65 Economic composite.31.22.18.05.14 Unemployment rate 5.37 1.86 4.61 6.08**** 7.66 Housing index.57.03.83.43 5.57 Stock prices 10.36 1.46 2.65.66 6.12 Business enterprises.42 1.18.28 1.29 1.37 Gross National Product 3.68.03 3.64.65 2.81 1. This column contains the t values for the differences on the threat indicators between 1955-1964 and 1965-l974. 2. This column contains the t values for the differences on the threat indicators between 1965-1974 and 1975-1984. p <.10. ****p <.001. thoritarian denominations occurred during the 1965 to 1974 period. The second prediction was that the change in membership for the more threatening 1965 to 1974 period was above the mean of the change in membership for the less threatening 1955 to 1964 and 1975 to 1984 periods for authoritarian denominations but below the mean of the changes in membership for the less threatening 1955 to 1964 and 1975 to 1984 periods for nonauthoritarian denominations. Therefore, even if the membership trend was decidedly upward or downward from the 1955 to 1964 period to the 1975 to 1985 period, the second prediction was for an elevation in the trend line for the more authoritarian churches and a depression in the trend line for the less authoritarian churches during the 1965 to 1974 period. It should be apparent that if the first prediction is supported, then so too must be the second. But the second prediction may receive support without the first being verified. TESTING THE PREDICTIONS Both predictions were tested with the chi-square goodness-of-fit statistic. For the first prediction, the observed frequency was simply the total number of denominations authoritarian and nonauthoritarian that showed the respective predicted pattern across the three periods, that is, the number of authoritarian denominations that had their best years during the more threatening 1965 to 1974 period plus the number of nonauthoritarian denominations that had their worst years during that same period. Of the 25 denominations, 17 supported the first prediction, χ 2 (1, N = 25) = 3.24, p <.05, one-tailed. For the second prediction, the observed frequency was simply the number of denominations that showed the respective predicted pattern across the three periods. That is, the observed frequency was the number of authoritarian denominations with the membership change during the 1965 to 1974 period above the mean of the membership changes for the 1955 to 1964 and 1975 to 1984 periods plus the number of nonauthoritarian denominations with the membership change during the 1965 to 1974 period below the mean of the membership changes for the 1955 to 1964 and 1975 to 1984 periods. Of the 25 denominations, 20 supported the second prediction, χ 2 (1, N = 25) = 9.00, p <.001, one-tailed. The percentage changes in the proportion of the population who were members of the more authoritarian and the less authoritarian denominations over each of the three periods are displayed in Table 4. Support for each prediction also is indicated. Discussion The general hypothesis that authoritarian churches are more attractive during threatening times and nonauthoritarian churches are more attractive during nonthreatening times gained support in Study 2. The first prediction that the authoritarian churches incurred their greatest membership gains or smallest losses whereas the nonauthoritarian churches incurred their smallest gains or greatest losses during the more threatening 1965 to 1974 period was verified; 68% of the denominations satisfied this pattern. The second prediction that the change in membership for the more threatening 1965 to 1974 period was above the mean of the changes in membership for the less threatening 1955 to 1964 and 1975 to 1984 periods for authoritarian churches but below the mean of the changes in member-
McCann / AMERICAN CHURCH MEMBERSHIPS 333 TABLE 4: Percentage Changes in the Proportion of the Population in Authoritarian and Nonauthoritarian Churches During Periods of High and Low Threat Low Threat High Threat Low Threat Church 1955-1964 1965-1974 1975-1984 Prediction 1 Prediction 2 Authoritarian Church of God (Andersson, Indiana) 1.0 4.4 1.0 Yes Yes Church of God (Cleveland, Tennesee) 23.0 50.3 37.3 Yes Yes Church of the Nazarene 8.4 15.6 7.0 Yes Yes Cumberland Presbyterian 20.7 7.2 5.7 Yes Yes Free Methodist Church 1.4 1.5 2.6 Yes Yes Jehovah s Witnesses 50.8 52.8 17.8 Yes Yes Mennonite Church 2.7 5.8 12.5 Yes Yes Salvation Army 1.5 20.2.5 Yes Yes Seventh-day Adventists 12.4 22.3 18.9 Yes Yes Wisconsin Evangelical 6.9.8 5.0 Yes Yes Southern Baptist 8.6 6.4 2.8 No Yes Baptist General Confererence 37.1 19.7 3.9 No No Evangelical Covenant 1.6 1.8 7.2 No No Lutheran (Missouri) 14.8 7.7 14.0 No No Mormon 24.2 17.5 49.4 No No Reorganized Mormon 4.3 15.7 10.1 No No Nonauthoritarian American Lutheran 13.5 14.5 12.7 Yes Yes Christian (Disciples) 13.7 38.9 22.5 Yes Yes Episcopal Church 5.7 24.4 13.3 Yes Yes Lutheran (in America) 2.8 14.5 12.2 Yes Yes Reformed (in America) 3.1 17.2 12.8 Yes Yes United Church of Christ 16.5 21.0 16.3 Yes Yes United Methodist 5.8 19.8 15.7 Yes Yes Church of the Brethren 15.0 17.2 19.7 No Yes Presbyterian (U.S.A.) 8.1 20.2 22.0 No Yes ship for the less threatening 1955 to 1964 and 1975 to 1984 periods for nonauthoritarian churches also was substantiated; 80% of the denominations satisfied this pattern. Study 2 also showed that noneconomic threats are related to fluctuations in church memberships. Analyses of the changes in the 10 threat variables demonstrated that the three periods progressed from low to high to low threat only as measured by the SEPT, social threat, and political threat indexes. The seven economic threat variables failed to distinguish the three periods in this manner, but the predictions still were supported. Therefore, it seems clear that economic threats are not essential to the link between threat and changes in church membership. Study 2 results enhance the generalizability of the principle that authoritarian churches are more attractive during threatening times, whereas nonauthoritarian churches are more attractive during less threatening times. Some support for the linkage has now been found for 20 representative American denominations. The predicted patterns also have been shown to occur over relatively short time periods in an ABA research design (i.e., high threat to low threat), making explanations based on long-term trends unlikely. GENERAL DISCUSSION Study 1 and Study 2 join a growing strand of archival researches that have predicted regularities in mass behavior on the basis of authoritarian personality theory and found empirical verification in transhistorical data (e.g., Doty et al., 1991; Jorgenson, 1975; McCann, 1991, 1997; McCann & Stewin, 1984, 1987, 1990; Padgett & Jorgenson, 1982; Rokeach, 1960; Sales, 1972, 1973). Generally, the present inquiry has provided correlational evidence that, at least indirectly, buttresses the validity of authoritarian personality theorizing, which holds that contemporaneous threat is a cause of heightened authoritarianism (e.g., Fromm, 1941; Rokeach, 1960; Sanford, 1966; Wilkinson, 1972). That is, the results of Study 1 and Study 2 are consistent with the theory that people become more authoritarian during stressful times and consequently gravitate toward or experience a greater affinity to more authoritarian churches. However, the notion that people become more authoritarian during stressful times and that this causes the shifting de-
334 PERSONALITY AND SOCIAL PSYCHOLOGY BULLETIN nominational trends must remain an inference because authoritarianism was not directly assessed in this research. On the basis of the current work, the possibility cannot be ruled out that an individual difference variable other than authoritarianism interacted with threat to produce the results. But what are plausible alternatives perhaps conservatism, dogmatism, or integrative complexity? Wilson (1973) put forth a conservatism syndrome that incorporates religious dogmatism, and a central proposition is that the common basis for the various syndrome components is a generalized susceptibility to experiencing threat or anxiety in the face of uncertainty (p. 259). Rokeach s (1960) dogmatism theory includes the assumption that the more threatening a situation is to a person, the more closed his belief system will tend to become and that the effects should last as long as the person experiences threat (p. 377). Porter and Suedfeld (1981) suggest that societal threat decreases integrative complexity so that persons are more likely to evaluate stimuli rigidly, reject dissonant information in order to minimize conflict, engage in categorical black-white thinking, minimize conflict by fitting stimuli into existing categories or excluding them from consideration, and show a general inability to understand the perspective of other people (p. 322). It is conceivable that conservatism, dogmatism, or integrative complexity might provide an explanatory framework for the present results. However, Eckhardt (1991) reviewed the literature of authoritarianism, conservatism, dogmatism, and other similar constructs and concluded that these concepts appear to be different aspects of the same thing (p. 98), and Stone (1993) criticized integrative complexity as an extreme simplification of authoritarian personality theory (p. 168). So, the new interpretive veneer would still stem from the authoritarianism concept. Furthermore, authoritarian personality theory prompted the present research, and it appears to me to furnish a rich explanatory foundation. If conservatism, dogmatism, or integrative complexity can provide superior or even comparably satisfying alternative explanations in this context, then those articulations are best left to those who feel less comfortable with the present authoritarianism interpretation. Is it possible that some factor other than threat led to the membership changes observed in Study 1 and Study 2? Perhaps, but none is apparent to me, and two broad categories of potential explanatory factors probably can be eliminated from serious consideration. First, because membership fluctuations systematically changed in accord with predictions for so many denominations, it is even more obvious here than it was for Sales (1972) that events specific to any single church (e.g., turnover in the national leadership; promulgation of new dogma) cannot account for the present data (p. 425). Second, Sales s (1972) argument that since churches with broad national representation...exhibited patterns similar to those of more locally based churches...itisunlikely that region-specific factors could explain these findings (p. 425) is even more compelling here. Many more churches have been studied, both with broader national membership bases and more regional followings. It has sometimes been proposed that people who are upwardly or downwardly mobile also shift their affiliations to churches with commensurate socioeconomic status (e.g., Newport, 1979; see Roberts, 1984, pp. 159-160). Different denominations tend to draw their adherents from different rungs in the socioeconomic ladder (e.g., see Roberts, 1984, pp. 283-289), and members of authoritarian churches generally are lower in socioeconomic status than are members of nonauthoritarian churches. Therefore, it might be argued, as Sales (1972, p. 425) recognized, that membership changes toward authoritarian churches during times of high threat are simply due to greater downward social mobility in poorer times and that membership changes toward nonauthoritarian churches during times of low threat are due to greater upward mobility in prosperous times. But such an explanation would not appear to suffice for the present data. In Study 1, SEPT index scores generally were related to membership fluctuations even when economic variables were statistically controlled. Even more convincing evidence comes from Study 2, in which the predictions were supported across time periods that did not systematically differ in terms of economic threat. Consequently, the social mobility argument would not seem to apply because economic threat and consequent mobility changes are not necessary to produce the observed pattern of church membership fluctuations. Sales (1972) acknowledged that, at the root of it, the pattern of attraction to authoritarian and nonauthoritarian churches in times of relative stress and relative tranquillity does not necessarily support the hypothesized link between threat and authoritarian tendencies. A variable common to authoritarian denominations and a variable common to nonauthoritarian denominations, neither of which is related to authoritarianism, could function in threatening and nonthreatening times to generate the denominational attraction patterns found here. At the theoretical level, this counterargument is as applicable to the results of Study 1 and Study 2 as it was to Sales s findings, but it remains merely speculative with no hint of the nature of the hypothetical factors involved. Sales defended against such criticism by noting
McCann / AMERICAN CHURCH MEMBERSHIPS 335 similarities between his results and those of two laboratory experiments (Sales & Friend, 1973), which demonstrated that failure-induced threat led to increases in authoritarian behavior, whereas success led to decreases. Several other diverse studies since 1972 also have provided results consistent with the hypothesized threatauthoritarianism link (e.g., Doty et al., 1991; Jorgenson, 1975; McCann, 1991, 1997; McCann & Stewin, 1984, 1987, 1990; Padgett & Jorgenson, 1982; Peterson et al., 1993; Sales, 1973), and because these roughly parallel results have been found in various contexts, it seems unlikely that the current findings can be attributed to some as yet undefined, elusive, extraneous difference between authoritarian and nonauthoritarian churches. The present findings may have implications for Kelley s (1972) influential theory of church growth and decline articulated in Why Conservative Churches are Growing. Kelley sought to explain why some churches were gaining members whereas others were losing members during the middle to late 1960s and early 1970s. It was Kelley s thesis that the membership of strong churches is destined to increase and that the membership of weak churches is destined to decrease. The pattern of church membership changes during the middle to late 1960s and early 1970s appears to offer support to Kelley s contention. However, as we have seen earlier, Kelley s strong and weak churches show correspondence to Sales s (1972) categorization of authoritarian and nonauthoritarian churches. Therefore, in light of the current findings, an alternative explanation for at least some of the variance in membership changes seems plausible: Because the years under consideration by Kelley were characterized by above average threat, higher levels of authoritarianism may have contributed to the differential changes in membership statistics in strong and weak churches. In fairness to Kelley, however, despite the fact that threat may have slowed, accelerated, and even reversed membership trends differentially for strict and lenient denominations, the Study 2 data show that the membership share of the majority of strict churches did continue to increase, whereas the membership share of almost all lenient churches continued to shrink through 1985. Understanding the dynamics of church membership fluctuations is broadly important because denominational oscillations are not insulated from the wider society. Threats to the established order and way of life may sway the fortunes of authoritarian and nonauthoritarian denominations, but because membership in authoritarian and nonauthoritarian churches is related to stands taken on many divisive contemporary issues (e.g., McKinney & Olson, 1991; Peele, 1984), so too may membership fluctuations have ramifications for threats to the established order and way of life. The enduring and pervasive nature of authoritarian and nonauthoritarian churches ensures their capacity to be strong forces in the chronic struggle to democratically produce viable and equitable policy and legislation. Winds of threat fan the flames of authoritarian sentiment, and the strength of the storm can determine the course of the nation. NOTES 1. Authoritarian and nonauthoritarian are used as convenient descriptive labels for, respectively, more authoritarian and less authoritarian denominations. Of course, it is recognized that, in a sense, all churches may be said to be authoritarian at their core. It should also be emphasized that this dichotomous categorization is meant to be entirely descriptive rather than evaluative and does not imply relative superiority of the churches in either category. 2. Whenever possible, cumulative tables were consulted. However, in many instances, data had to be taken from yearly presentations in annual volumes. For convenience, the designation several years is used throughout this paper. 3. The complete list of relevant areas can be found in McCann (1997). REFERENCES Adorno, T. W., Frenkel-Brunswik, E., Levinson, D. J., & Sanford, R. N. (1950). The authoritarian personality. New York: Harper. Ahlstrom, S. E. (1972). A religious history of the American people. New Haven, CT: Yale University Press. Altemeyer, B. (1988). Enemies of freedom: Understanding right-wing authoritarianism. San Francisco: Jossey-Bass. Altemeyer, B., & Hunsberger, B. (1992). Authoritarianism, religious fundamentalism, quest, and prejudice. International Journal for the Psychology of Religion, 2, 113-133. Argyle, M. (1958). Religious behaviour. London: Routledge and Kegan Paul. Bandura, A. (1977). Social learning theory. Englewood Cliffs, NJ: Prentice Hall. Bibby, R. W. (1978). Why conservative churches really are growing: Kelley revisited. Journal for the Scientific Study of Religion, 17, 129-137. Doty, R. M., Peterson, B. E., & Winter, D. G. (1991). Threat and authoritarianism in the United States, 1978-1987. Journal of Personality and Social Psychology, 61, 629-640. Eckhardt, W. (1991). Authoritarianism. Political Psychology, 12, 97-124. Fagan, J., & Breed, G. (1970). A good short measure of religious dogmatism. Psychological Reports, 26, 533-534. Famighetti, R. (Ed.). (1985). The world almanac and book of facts 1986. Mahwah, NJ: Funk and Wagnalls. Fromm, E. (1941). Escape from freedom. New York: Holt, Rinehart and Winston. Gregory, W. E. (1957). The orthodoxy of the authoritarian personality. Journal of Social Psychology, 45, 217-232. Hoge, D. R. (1979). A test of theories of denominational growth and decline. In D. R. Hoge & D. A. Roozen (Eds.), Understanding church growth and decline: 1950-1978 (pp. 179-197). New York: Pilgrim. Hoge, D. R., & Roozen, D. A. (1979). Understanding church growth and decline: 1950-1978. New York: Pilgrim. Hudson, W. S. (1961). American protestantism. Chicago: University of Chicago Press. Jacquet, C. H. (Ed.). (1987). Yearbook of American & Canadian churches 1987. Nashville, TN: Abingdon. Jacquet, C. H., & Jones, A. M. (Eds.). (1991). Yearbook of American and Canadian churches 1991. Nashville, TN: Abingdon. Jorgenson, D. O. (1975). Economic threat and authoritarianism in television programs. Psychological Reports, 37, 1153-1154.