THE STRANGE REAPPEARANCE OF CIVIC AMERICA:RELIGION AND VOLUNTEERINGAndrew Greeley |
|
|
| .INTRODUCTION | ||
| Harvard Professor Robert D. Putnam (with notable help from New York Times columnist Anthony Lewis) has argued recently that "Civic America" has "strangely disappeared." I propose to contest that assertion by considering an aspect of American civic life which Professor Putnam has "strangely" ignored - the volunteer phenomenon. I will argue in the body of that paper that volunteering in America is higher than anywhere else in the world, that it has increased sharply in the last ten years, and that higher levels of religious devotion are in substantial part responsible for America's lead over fifteen other countries in the proportion of its population who engage in unpaid, volunteer service. | _ |
|
 |
In an appendix I will challenge the
methodology which Professor Putnan uses in his treatment of data from NORC's General
Social Survey to prove his thesis. . My theoretical intent in this analysis is to turn the economic models of religion on their head (which does not mean, be it noted, to refute them). I will ask whether, in addition to being an example of consumption capital, religion can also be considered productive capital in the sense that it leads to financial input to the economy. In this perspective religion becomes not merely a dependent variable but also an independent variable. I take volunteer service as the dependent variable and ask whether religion contributes to the economic input which volunteering can be considered to represent. More specifically I ask whether the much higher proportion of Americans who volunteer is the result of higher levels of religious practice in this country. VOLUNTEERS AROUND THE WORLD The European Value Study asked the following question in 1991: "Please look carefully at the following list of voluntary organizations and say . . . a) Which, if any, do you belong to? b) Which, if any are you currently doing unpaid voluntary work for? A. Social welfare services for elderly, handicapped or deprived people. B. Religious or Church organizations. C. Education, arts, music or cultural activities D. Trade Unions E. Political parties or groups F. Local community action on issues like poverty, employment, house, racial equality G. Third world development or human rights H. Conservation, the environment, ecology I. Professional associations J. Youth work (e.g. scouts, guides, youth clubs etc.)1 K. Sports or recreation L. Women's groups M. Peace movement N. Animal rights O. Voluntary organizations concerned with health P. Other groups. A summation of the number of people who volunteered for at least one activity provides a statistic for the proportion of a population currently volunteering. Figure 1 shows the proportions for sixteen countries. The highest rate of volunteers is in the United States where 47% of the population reported that it had volunteered. This is consistent with the proportion reported Hodgkinson and Weitzman(1992) in response to a question about who had volunteered during the last year.2 Only Canada had a rate comparable to that of the United States. Rates in three Scandinavian countries and the Netherlands were between 30% and 40% . Most of the other European countries reported rates between 20% and 30%, though the rates were lower than 20% in Portugal and Great Britain. Men are marginally more likely to volunteer than women and those between thirty and fifty more likely to volunteer than those who are younger or older. (Figure 2)The latter finding is replicated in Hodgkinson and Weitzman for the United States. However, they report women more likely than men (53% to 49%). Factor analysis was applied to the sixteen response categories in the EVS questionnaire. Three factors emerged from the varimax rotation, one emphasizing the more "modern" kinds of volunteer activity, a second, a more "traditional" variety, and the third a political/social dimension. There is no gender difference between men and women in "modern" volunteer work, women are more likely than men to engage in "traditional" work, and men more likely than women in "political/social" volunteering. Thus the seeming contradiction between the two studies is a result of different question wordings, the EVS questionnaire being more likely to ask about activities that men would favor. Volunteering in the EVS sample also correlates strongly with education. Those who continued in school after they were twenty years old were twice as likely to report some kind of volunteering than those who had left school before they were twelve. Both frequency of Church attendance and membership in Church organizations correlate strongly with voluntary service (Figures 3 and 4). Those who attend services once a week or more are approximately twice as likely to volunteer than those who attend rarely if ever. Moreover, (Figure 4) while organizational membership (a sum of all secular organizational memberships) does, as one might expect, powerfully predict voluntary service, membership in a religious organization has its own independent effect. The correlations between the two variables and volunteering is .32 for religious membership and .50 for the sum of secular membership. The beta weights (correlation net of one another) is .44 for secular membership and .21 for religious membership. Denomination in those six countries where there are sufficiently large samples of both Catholics and Protestants to make comparisons possible has an impact only in two countries - Canada where Protestants are significantly more likely to volunteer than Catholics and Northern Ireland where the reverse is true. There are no significant differences in Britain, the United States, West Germany, and the Netherlands. By way of summary, education, age, church attendance and organizational membership all influence the propensity to volunteer. Denomination does not influence it in countries where comparison is possible. CHANGES IN VOLUNTARY SERVICE Eleven of the sixteen outlets for voluntary activity were proposed in the 1981 EVS study. While the changing of the wording of the question (by adding five new items to the list) makes strict comparison across time questionable, especially on individual items3, it is not impossible that the new questions will not affect the total response pattern. Thus, I compared the 1981 and 1990 responses to the original item list. Table 1 shows that there are only three countries where statistically significant change in the proportion volunteering has occurred - Austria where there has been a seven percentage point decline and the North American countries where the proportion volunteering (as measured by the original 1981 items) has increased, by seven points in Canada and twelve points in the United States. Figure 5 shows that this increase in the United States is especially pronounced among the cohorts born since 1940, each one of which has not only increased its volunteer rate over that of the last ten years but has a higher rate than its predecessor cohort did when it was in the same age position. Thus the cohort born in nineteen forties not only has almost doubled its volunteer rate in the ten year period, it is also fourteen percentage points ahead of where the nineteen thirties cohort was when it was the same age. The increase in volunteering in America has been created by the "Baby Boomers," the "Me Generation," and "Generation X," all of whom, if one is to believe the popular media are inherently selfish and "uncommitted." Moreover the increase continues because even those born during the nineteen seventies and hence at the most only twenty at the time of the 1990 survey are already more likely to volunteer than those born in the nineteen sixties were ten years ago when they were in their twenties and those born in the nineteen fifties when they were in their thirties, the prime age for voluntary service. However, those born in the nineteen twenties also show an increase in volunteering over the last ten years. The increase then is affecting major segments of the population regardless of age. A similar phenomenon is taking place in Canada. There too the more recent cohorts account for much of the change. There too there is an increase through the eighties among those born in the nineteen twenties. During the eighties, then, volunteer rates remained stable in Europe and increased dramatically in North America. WHY AMERICANS VOLUNTEER MORE Contrary then to what one might believe from reading both Robert Bellah and Pope John Paul II, Americans do not appear to be selfish individualists when it comes to volunteering. As the first column in Table 2 demonstrates Americans are significantly more likely to report volunteer service than citizens of thirteen of the other fourteen countries included in this analysis, Canada alone being the exception. The question then becomes whether the higher levels of religious devotion in the United States can account for the higher levels of voluntary service. The protocol for considering religious influence in contemporary social analysis is that social structural variables be considered before religious ones so that whatever impact religion may have will be residual, that is, an impact that cannot be accounted for by social structure. As the second column of the table, indicates social structural variables4 reduce the advantage somewhat save in the Scandinavian countries (which for our purposes include Iceland). The final number in each column represents the added impact of the country dummy variables when they are added to the model the column represents. Thus by themselves the dummy variables explain 3.58% of the variance. Their explanatory power is reduce to 2.31% of the variance when the social structural variables - age, sex, education, occupation, and income are added in a new model. In the third column the number of secular memberships is added to the existing model. Thus the numbers in the column are those differences which remain between the various countries and the United States after age, sex, education, occupation, income, and organizational membership are taken into account. In this model the country differences are reduced save in the Netherlands and the Scandinavian countries; and the difference between Italy and the United States becomes statistically insignificant. Moreover the amount of variance attributable to the country variables declines to 1.90%. When two religious behavior variables - frequency of Church attendance and membership in church-related organizations5 - are introduced in the fourth column of the table, the advantages of higher levels of American religious behavior on voluntary service becomes evident. A statistically significant disadvantage in comparison with the United States persists in only four countries - Britain, Northern Ireland, Denmark, Iceland, and the Netherlands. 6 Moreover the addition of the two religion items diminishes explanatory power of the country variables to .53. The final model in the fourth column reduces the variance attributable to country to a little more than 1/7 of what it was in the first row. Thus it would appear that, net of the other influences, the higher levels of religious practice in the United States and Canada make a major contribution to the higher levels of voluntary service and indeed these religious differences are a major factor in the explanation of country differences. VOLUNTEER PROFILES FOR THE SIXTEEN COUNTRIES As Table 3 shows, these two religious behavior variables affect volunteering except in Ireland where frequency of Church attendance has no impact on volunteering, most likely because of the uniformity of attendance rates in Ireland. Even in the countries where religious activity is not high, such as the Scandinavian countries, religious behavior still has a significant impact on voluntary service. Motivations for volunteering differ considerably between North American and Canada on the one hand and Western Europe on the other(Table 4). The former countries emphasize the idealistic and moral reasons for volunteering while the latter emphasize the more pragmatic reasons. In both cases it is altogether possible that the reasons given are the culturally fashionable rationalizations for generosity. Thus it is the allegedly pragmatic North Americans who are the idealists and the allegedly principled Europeans who are the pragmatists. Wuthnow (1991) in his important work on volunteering attempts to reconcile the high levels of such activity in the United States with the "individualism" of Americans. In fact, the data in the present analysis suggest that the real individualists are the Dutch and the Scandinavians and the real idealists or "communalists" are the North Americans. Perhaps individualism has so many different meanings that it means nothing at all. Clearly then there are different cultural approaches to voluntary service in the sixteen countries under consideration, with Scandinavia and North America having the highest volunteer scores and the opposed motivational patterns, the former the more "idealistic" and "traditional" the latter more "pragmatic" and "social." SECULAR VOLUNTEERING Religious devotion and Religious group membership could be expected to influence volunteering that is church-related. Do they also influence volunteering that is not church-related. Figure 6 shows the rates for the sixteen countries of "secular" volunteering, that is voluntary service that is not related to church organizations. When religious volunteering is excluded, American rates(34%) fall somewhat beneath those of Canada (38%)and Sweden (38%) and are virtually equal with those of Norway (34%), Iceland (35%) and the Netherlands (33%). The American advantage in volunteering is then clearly concentrated in church-related voluntary efforts. Twenty eight percent of Americans volunteer for church-related services as do fifteen percent of the Canadians. In all other countries the rates are lower than ten percent. However, church attendance has a positive impact on secular voluntary service as does membership in a church-related organization. The correlations between these two variables and all volunteering is .16 and .32 and between them and secular volunteering is .09 and .19. HOW MUCH IS RELIGION WORTH? Hodgkinson and Weitzman observethat the mean number of weekly volunteer hours for an American who does volunteer is 4.2 hours. They calculate that more than fifteen billion hours of volunteer service are offered each year. Estimating that a volunteer hour is worth $11.58 (the average hourly wage for non-agricultural workers in 1991), suggest that the volunteer input to the economy in 1992 was a hundred and seventy six billion dollars. Since the difference between the United States and West Germany can be explained entirely by the different religious practices in the two countries, one can say that if Americans were like the West Germans in their religious behavior, almost forty percent of that contribution or some seventy billion dollars would be lost. CONCLUSION The North American countries, often dismissed as selfish and materialistic, are the most likely to have high rates of voluntary service. In substantial part these rates are accounted for by higher levels of religious practice and explained by the volunteers in "idealistic" and "moral" terms. Moreover volunteer rates increased in the North American countries in the nineteen eighties while they remained stable in Europe so that at the end of the decade they were even higher in North American than at the beginning of the decade. In both North American countries the increase occurred in most birth cohorts, especially the younger ones. These generous, religiously-driven "habits of the
heart" make a major contribution to the economies of both countries. Economists ought
not to overlook the importance of religion and religious motivations as they elaborate
their models. Appendix Three Methodological Errors in Robert Putnam's Use of the General Social Survey If one has to choose between unpaid volunteer service and membership in "civic" groups as an indicator of "social capital," many would surely choose the former because membership does not of itself imply that one does anything except belong (and perhaps pay dues) while volunteering does suggest that one is doing something for others. However I proposed to argue in this note that the decline in membership in groups is illusory, at least as this decline is supposedly established by data in NORC's General Social Survey. In his two recent works on "The Strange Disappearance of Civil America, Putnam (1995, 1996) makes three serious methodological mistakes in his use of data from the General Social Survey: ==> While he correctly notes a decline in the number of total group memberships between 1974 and 1993 (r=-.04), he fails to disaggregate the various kinds of membership to determine which kinds of memberships may have increase and which may have increased. As Table 5 shows membership in unions, farm groups church groups, traditional fraternal groups, veterans groups and fraternities and sororities have declined, while membership in service clubs, nationality groups, professional groups, and hobby groups have increased. The declines in farm, union, fraternal, and veteran groups and the increase in professional groups do not seem surprising, though the decline in church-related groups does. However, the decline of social capital such as it may be affects only some kinds of memberships while other kinds of membership have actually increased during the two decades being measured. One must note the absence in the General Social Survey questionnaire two other kinds of groups which have become important in the last two decades - community organizations and environmental groups. Data from the two waves of the values Study (1981 and 1990)show a marginal (and not statistically significant) increase in American group membership between 1981 and 1990 (Table 6). These confirm that membership in both community organizations and environment groups significantly increased during the 1980s. On two questions with virtually the same wording as the General Social Survey questions - membership in unions and in church-related organizations there is virtually no difference in the coefficients, suggesting that both surveys are dealing with similar phenomena. A much more modest claim by Putnam might be justified: membership in some kinds of organizations had decreased an in other kinds of organization has increased. This fact, combined with the data on volunteers in the main part of this paper suggest that the civil culture in America has not "suddenly" disappeared, but may not have disappeared at all. ==> Moreover in both his papers Putnam blames television for the alleged disappearance of civic America. He shows an increase in television view as recorded in the General Social Survey, a decline in membership, and a correlation between television viewing and membership and concludes that therefore television is causing the decline in membership. However, as one learns in introduction to analytic methodology, one begs the question with this kind of argument unless one proves that in fact television view is responsible for the negative correlation which exists between time and group membership. There are many ways this can be done, most notable an OLS regression equation in which membership is regressed on year and hours spent watching television. In the General Social Survey 1974 to 1993 the raw correlation between years and group membership is .03877 when hours of television is added to the equation the beta diminishes to .0364. The correlation is reduced by only .0023 by television or by six percent. Thus ninety four percent of the relationship between time and group membership remains unexplained. The null hypothesis that television had very little to do with the decline in group membership as measured by General Social Survey data cannot be rejected. ==> Finally, the complex relationship between age, cohort and group membership requires much closer examination than Putnam (1996)gives to it. He does distinguish between life cycle, period, and generational effects, but he does not seem to understand these differences or now to extricate them one from another. The only way to eliminate the possibility that what seems to be a pure period effect (a decline for example in civic virtue in American culture ) is in fact nothing more than a life cycle effect is to follow birth cohorts through the life cycle. Thus (page 44) Putnam asserts that "members of the generation born in the 1920s belong to almost twice as many civic associations as those born in the late 1960's." But those born in the late 1960s would have been interviewed for the General Social Survey while they were still under thirty years old, while many of those born in the 1920's would have been interviewed while they were still in their forties. This would make no difference in the analysis if people were as likely to join organizations when they were twenty five as when they were forty five. In fact, however, we know that organizational membership peaks between the ages of thirty and fifty. Perhaps the younger generation is not less civic but only younger. Inexplicably, Putnam does not take this possibility into account. One must compare cohorts when they are the same age to examine this possibility. Are those born in the 1950s (the much maligned "Boomers") for example, less likely in their middle forties to belong to organizations as were those born in the 1920s when they were that same age? If such be the case, than Putnam is right, there is a pure period effect, an actual decline in social capital and a disappearance, though perhaps not so sudden, of "civic America." If, on the other hand, those born in the 1950s are as likely as earlier generations to increase their group membership as they approach the peak membership years, then the apparent decline in social capital is not a pure period effect (as Putnam's argument implies, though he defines the word differently) but a life cycle effect which implies no change in American culture and no long term decline in social capital. As Figure 7 shows there appears to be a life cycle curve between age and membership. Thus "Boomers" at the age of forty five are not statistically different from those born in the 1930s or the 1920s when these two latter groups were in their middle forties and have actually passed those born in the 1940s when they were that age. Moreover, the cohorts born in the 1960s and the 1970s are following the same pattern as they move through the life cycle. The "decline" of social capital which Putnam thinks he has found might be the result of nothing more than the shifting through the life cycle of smaller cohorts (pre-1950) when they were at their peak organizational ages and larger cohorts when they were still in their lower organizational ages. This possibility can be examined for numbers of group membership with simple OLS regression analysis. If one enters separately age and cohort into a regression equation and the negative correlation between years and membership does not diminish substantially and then puts both age and cohort in at the same time and the negative correlation disappears one has proved that the "decline" is merely the result of the phenomenon described in the previous paragraph - a shifting through the life cycle of smaller and larger cohorts. This is exactly what happens. The simple correlation between number of groups and year is -.047. Neither age nor cohort by themselves diminish this negative relationship in the slightest. But when age and cohort are added at the same time the relationship goes down to -.01 and statistical insignificance. As far as data in the General Social Survey can be used to test the matter, the apparent disappearance of civic America, is in fact the result of the relative position in their life cycles of the larger and smaller age cohorts. The same phenomena applies both to the proportion belonging to organizations (the dependent variable in Figure 7) when one uses logistic regression analysis. The amount of variance (the "pseudo" R2) which can be accounted for by a pure period effect is less than one tenth of one percent when age and cohort are taken into account. If one also applies this logistic regression model to the apparent decline in membership in church-related organizations, the pure period effect is less than three tenths of one percent. Why Putnam did not use these techniques in his analysis and why the referees for his publications did not insist on them is a very interesting question. Almost anyone who has done demographic analysis knows about the age/cohort phenomenon in contemporary American life, as the younger and larger cohorts catch up with the older and smaller ones. It is not altogether responsible, however, to indulge in broad generalizations about the "decline" of something American without examining the elementary methodological questions raised in this note. Such generalizations, however, are extremely popular in certain elite circles, such as the op ed page of the New York Times. These three methodological flaws in Putnam's analysis, in addition to the sharp increase in voluntary activity, means that it is very hard to reject the null hypothesis that "Civic America" has not disappeared at all, much less "strangely" disappeared. Not even at Harvard. 1 In an earlier study (1981) items K to O were not asked. Since the wording of the question was changed in this way comparisons with the earlier data must be made cautiously. 2 Studies done in England and France replicate the percentages reported in the Values Study. 3 The Directors of the Value Study do not tell their readers in their report that the question was changed. Moreover, they create a "religious " index in which membership in religious organizations is used as an item, without noting the change in the question wording and hence taking into account the possibility that modification of the proportion reporting membership in religious organizations could easily be the result of the modification of the wording of the question. This possibility is especially likely since in many countries sports organizations are sponsored by the churches. 4 Men are marginally more likely to report volunteering in the EVS questions then women. Both men and women in their thirties are more likely to volunteer than younger or older people. Education attainment is the most powerful of the social structural variables. 5 All other religious variables have an insignificant impact on volunteering when entered into regression equations with these two religious behavior items. 6 7 The television question was not asked every year so the
negative correlation is slightly lower in this equation. Charts 1. Volunteers by
Country Tables Changes in Volunteer
Rates |
|
