BOWLING ALONE
Page 71
7. Of the ninety-one possible bivariate correlations among these fourteen indicators, eighty-eight are statistically significant in the proper direction at the .05 level or better, and none are in the wrong direction. The mean inter-correlation across the ninety-one is r = .56. This concordance is impressive, given that the underlying data come from three independent survey archives and three different government agencies. The summary index is simply the average of the standardized scores on the fourteen component measures. To maximize the number of cases, we computed this average even for those few cases in which data were missing on as many as five of the underlying fourteen indicators; this procedure enabled us to include all states except Alaska and Hawaii in our analysis. Effectively, this index is identical to the factor score from a principal components analysis of the fourteen component variables.
8. The few exceptions from the surprisingly smooth gradients in figure 80 are intuitively explicable—Nevada is unusually low, whereas Mormon Utah is relatively high.
9. One other plausible measure of social capital—church attendance—is empirically quite unrelated with the other indicators used here. The fraction of all respondents in the 1974–94 General Social Survey who report attending religious services at least “nearly every week” is essentially uncorrelated with our Social Capital Index (r = -.06). Some states with high levels of religious observance (Alabama, for example) are very low on our measure of community-based social capital, but other relatively religious states (Minnesota, for example) are very high in social capital. Conversely, South Dakota is high on social capital but low on church attendance, while Hawaii is relatively low on both.
10. Tocqueville, Democracy in America, 81.
11. Our measure of social capital at the state level in the 1980s and 1990s is correlated R = .52 with the measure of “state political culture” invented by Daniel J. Elazar, American Federalism: A View from the States (New York: Crowell, 1966), based on descriptions of state politics in the 1950s and subsequently quantified by Ira Sharkansky, “The Utility of Elazar’s Political Culture,” Polity 2 (1969): 66–83. In a fascinating and important study, Tom W. Rice and Jan L. Feldman, “Civic Culture and Democracy from Europe to America,” Journal of Politics 59 (1997): 1143–1172, report that “the civic attitudes of contemporary Americans bear a strong resemblance to the civic attitudes of the contemporary citizens of the European nations with whom they share common ancestors,” even though the last direct contact with the “mother country” may have been several generations ago.
CHAPTER 17: EDUCATION AND CHILDREN’S WELFARE
1. Urie Bronfenbrenner, Phyllis Moen, and James Garbarino, “Child, Family, and Community,” in Ross D. Parke, ed. Review of Child Development Research, vol. 7. (Chicago: University of Chicago Press, 1984).
2. Kids Count Index from Annie E. Casey Foundation (Baltimore, Md., 1999), Web site: www.aecf.org /kidscount/index.htm.
3. The Pearson’s r correlation coefficient is +0.80. A score of 1 would represent perfect linear association; social scientists generally consider scores above .40 to constitute strong correlation.
4. This conclusion is based on ten ordinary least squares multivariate regression analyses. The units of observation were the fifty states, excluding D.C. The following ten dependent variables were used: births per one thousand females aged fifteen to seventeen in 1995; percent of children in poverty in 1995; percent of babies born at subnormal weight in 1995; percent of teens (sixteen to nineteen) not attending school and not working in 1995; infant mortality rate in 1995; child death rate (aged one to fourteen) in 1995; percent of teens (sixteen to nineteen) who are high school dropouts; death rate of teens (fifteen to nineteen) by accident, homicide, and suicide in 1995; arrest rate of juveniles (ten to seventeen) for violent crimes in 1995; as well as the comprehensive Kids Count index for 1997. In each regression model, the following control variables were included simultaneously: the state poverty rate (1987–92); the fraction of the 1990 population that was white; the fraction of all families with children that are headed by single parents; and the fraction of adults who have graduated from high school. In the full models, the poverty rate was a signifi- cant (p <.05 or better) predictor of seven negative outcomes; meanwhile, a low score on the Social Capital Index was a significant predictor of five negative outcomes. Racial composition and the fraction of families headed by single parents were significant in four and three of the models, respectively, but the magnitude of the effect was small, and these predictors were also associated in the wrong direction in two and three of the models, respectively. Adult high school graduation rates linked in the wrong direction for seven of the ten variables. The fraction of adults who are college grads was also explored, but it also performed poorly as a predictor. In predicting the overall measure of child welfare, only poverty and social capital had a major independent effect, both at the .001 level of statistical significance.
5. Jill E. Korbin and Claudia J. Coulton, “Understanding the Neighborhood Context for Children and Families: Combining Epidemiological and Ethnographic Approaches,” in Jeanne Brooks-Gunn, Greg J. Duncan, and J. Lawrence Aber, eds., Neighborhood Poverty, Volume II (New York: Russell Sage Foundation, 1997), 65–79. See also Susan P. Lumber and Maury A. Nation, “Violence within the Neighborhood and Community,” in Violence against Children in the Family and the Community, eds. Penelope K. Trickett and Cynthia J. Schellenbach (Washington, D.C.: American Psychological Association, 1998), 191–194; Robert J. Sampson, Jeffrey D. Morenoff, and Felton Earls, “Beyond Social Capital: Spatial Dynamics of Collective Efficacy for Children,” American Sociological Review 64 (1999): 633–660.
6. James Garbarino and Deborah Sherman, “High-Risk Neighborhoods and High-Risk Families: The Human Ecology of Child Maltreatment,” Child Development 51 (1980): 188–198.
7. D. K. Runyan, W. M. Hunter, et al., “Children Who Prosper in Unfavorable Environments: The Relationship to Social Capital,” Pediatrics 101 (January 1998): 12–18; Howard C. Stevenson, “Raising Safe Villages: Cultural-Ecological Factors that Influence the Emotional Adjustment of Adolescents,”Journal of Black Psychology 24 (1998): 44–59; A. J. De Young, “The Disappearance of ‘Social Capital’ in Rural America: Are All Rural Children ‘At Risk’?” Rural Special Education Quarterly 10 (1989): 38–45.
8. Ronald A. Wolk, ed., Quality Counts: A Report Card on the Condition of Public Education in the 50 States (Washington, D.C.: Editorial Projects in Education, 1997), 3.
9. Excluding the District of Columbia from analysis, we find that the Social Capital Index is correlated with each of seven National Assessment of Educational Progress tests administered in the 1990s: fourth-grade math, 1992: r = .81; fourth-grade math, 1996: r = .67; eighth-grade math, 1990: r = .90; eighth-grade math 1992: r = .91; eighth-grade math, 1996: r = .88; fourth-grade reading, 1994: r = .68; eighth-grade science, 1996: r = .85. In addition, the Social Capital Index is correlated with state average scores on the Scholastic Assessment Test (1993), adjusted for test-participation rates across states (r = .67). The Social Capital Index is also negatively correlated with the state high school dropout rate aggregated over the period 1990–95 (r = -.79).
10. Author’s analysis of state-level data on educational performance and data from the DDB Needham Life Style and Roper Social and Political Trends archives, aggregated to the state level, along with state-level data on racial composition, poverty, and educational levels of the adult population. All analyses of state educational performance in this chapter control for single-parent rate, 1984–90; pupil-teacher ratio, 1988–90; state poverty rate, 1987–90; percent of population nonwhite, 1990; mean personal per capita income, 1980–90; income inequality (Gini coefficient), 1990; fraction of adult population with at least high school degree, 1990; total educational spending per pupil, 1989–90 to 1991–92 (in real dollars) and mean teacher salaries, 1989, both adjusted for differences in state cost of living; fraction of elementary and secondary students in public schools; Catholic percentage o
f state population; and a composite survey-based measure of religious observance.
11. Strictly speaking, the statistical analysis suggests that to bring North Carolina’s educational performance to the level of Connecticut merely by adjusting the student-teacher ratio would require a cut in average class size of twenty to twenty-five pupils per class, but the average class size in North Carolina at the time these data were collected was actually seventeen students. This fact represents statistically the practical impossibility of relying solely on smaller class size to fix educational problems.
12. Author’s analysis of state-level data on educational performance and data from the DDB Needham Life Style and Roper Social and Political Trends archives, aggregated to the state level.
13. In a multivariate regression with an index of student misbehavior as the dependent variable, community social capital had a standardized beta of -.612, compared with .333 for single-parent rate, .261 for fraction of the adult population with at least four years of high school, and .226 for the pupil-teacher ratio. All were significant at p <.05 or better (social capital was significant at p = .0002). Other demographic, economic, and educational variables that were included in the initial model were nonsignificant. The dependent variable was an index composed of high school teachers’ perceptions of the seriousness of four problems: student weapon possession, absenteeism, and apathy, as well as student-on-student violence.
14. P. W. Cookson, School Choice: The Struggle for the Soul of American Education (New Haven, Conn.: Yale University Press, 1994); Sharon G. Rollow and Anthony S. Bryk, “The Chicago Experiment: The Potential and Reality of Reform,” Equity and Choice 9, no. 3 (spring 1993): 22–32.
15. James S. Coleman and Thomas Hoffer, Public and Private High Schools: The Impact of Communities (New York: Basic Books, 1987), 94, 133–135, 231, 229. For contrary evidence, see Stephen L. Morgan and Aage B. Sørensen, “A Test of Coleman’s Social Capital Explanation of School Effects,” American Sociological Review 64 (1999): 661–681.
16. Anne T. Henderson and Nancy Berla, A New Generation of Evidence: The Family Is Critical to Student Achievement (Washington, D.C.: National Committee for Citizens in Education, 1994), 1.
17. Roger G. Barker and Paul V. Gump, Big School, Small School: High School Size and Student Behavior (Stan-ford, Calif.: Stanford University Press, 1964); Kenneth R. Turner, “Why Some Public High Schools Are More Successful in Preventing Dropout: The Critical Role of School Size,” unpublished dissertation, Harvard University Graduate School of Education, 1991.
18. Anthony S. Bryk, Valerie E. Lee, and Peter B. Holland, Catholic Schools and the Common Good (Cam-bridge, Mass.: Harvard University Press, 1993). For example, a public school in the fiftieth percentile for teacher enjoyment of work would move to the eighty-fourth percentile if Catholic school “communal organization” were adopted. Likewise, a fiftieth percentile public school that became more communal would move to the eighty-ninth percentile for staff morale; the thirtieth percentile for rates of class cutting; the twenty-eighth percentile for classroom disorder; and the sixty-sixth percentile for student interest in academics. See page 288.
19. Bryk, Lee, and Holland, Catholic Schools (1993), 314.
20. James P. Comer and Norris M. Haynes, Summary of School Development Program Effects (New Haven, Conn.: Yale Child Study Center, 1992).
21. James P. Comer, School Power: Implications of an Intervention Project (New York: Free Press, 1980), 126–28. See also Wendy Glasgow Winters, African-American Mothers and Urban Schools: The Power of Participation (New York: Lexington Press, 1993).
22. Anthony S. Bryk and Barbara Schneider, “Social Trust: A Moral Resource for School Improvement,” in G. G. Whelage and J. A. White, eds., Rebuilding the Village: Social Capital and Education in America (London: Falmer Press, forthcoming). See also Donald Moore, “What Makes These Schools Stand Out?” (Chicago: Designs for Change, April 1998), 1–19 and 83–103.
23. That smaller schools foster greater student engagement in curricular and extracurricular activities is a common finding among educational researchers, as is the generalization that extracurricular participation in school is a strong predictor of civic engagement in later life. See sources cited in note 17 above and in note 4 of chapter 24.
24. Coleman, “Social Capital in the Creation of Human Capital.”
25. Frank F. Furstenberg Jr. and Mary Elizabeth Hughes, “The Influence of Neighborhoods on Children’s Development: A Theoretical Perspective and a Research Agenda,” in Jeanne Brooks-Gunn, Greg J. Duncan, and J. Lawrence Aber, eds., Neighborhood Poverty: Volume II (New York: Russell Sage Foundation, 1997), 43.
26. Nancy Darling and Lawrence Steinberg, “Community Influences on Adolescent Achievement and Deviance,” in Brooks-Gunn, Duncan, and Aber, eds. Neighborhood Poverty: Volume II, 120–131; Jay Teachman, Kathleen Paasch, and Karen Carver, “Social Capital and the Generation of Human Capital,” Social Forces 75 (1999): 1343–1359.
27. Frank F. Furstenberg Jr. and Mary Elizabeth Hughes, “Social Capital and Successful Development among At-Risk Youth,” Journal of Marriage and the Family 57 (August 1995): 580–592.
28. Ernest T. Pascarella and Patrick T. Terenzini, How College Affects Students: Findings and Insights from Twenty Years of Research (San Francisco: Jossey-Bass, 1991); Uri Treisman, “Studying Students Studying Calculus: A Look at the Lives of Minority Mathematics Students in College,” College Mathematics Journal 23 (1992): 362–372; Alexander W. Astin, “What Matters in College,” Liberal Education (fall 1993): 4–14; Alexander W. Astin, “Involvement in Learning Revisited: Lessons We have Learned,” Journal of College Student Development 37 (1996): 123–134.
CHAPTER 18: SAFE AND PRODUCTIVE NEIGHBORHOODS
1. Robert J. Sampson and Jeffrey D. Morenoff, “Ecological Perspectives on the Neighborhood Context of Urban Poverty: Past and Present,” in Jeanne Brooks-Gunn, Greg J. Duncan, and J. Lawrence Aber, eds., Neighborhood Poverty: Volume II (New York: Russell Sage Foundation, 1997), 1–22; Robert J. Sampson, “The Community” in Crime, James Q. Wilson and Joan Petersilia, eds. (San Francisco: Institute for Contemporary Studies Press, 1995), 193–216.
2. Jacobs, Death and Life of Great American Cities, 56.
3. The Pearson’s r correlation coefficient between the average murder rate in a state (1980–95) and the Social Capital Index is -0.8 where -1.0 would constitute a perfect negative linear association.
4. In a multiple regression, with the fifty states as units of analysis, the best-fitting model includes four statistically significant variables: the Social Capital Index, the mean poverty rate (1987–90), the fraction of the population that is white (1990), and the fraction of the population classified as urban (1990). Other variables that were entered but not found to be statistically significant were mean single parent rate (1984–90); personal per capita income (1990, in $1992); fraction of the population with at least four years of college education (1990); fraction of the population with at least four years of high school (1990); fraction of the population that is Catholic; Gini index of income inequality (1990); and responses to the DDB Needham Life Style survey question about “worry that my family may become a victim of crime.” If the causal arrow ran from high crime to low social capital, as fear of crime inhibited social intercourse, then controlling for fear of crime should eliminate the crime–social capital correlation, but it does not; the partial correlation of crime and social capital remains a highly significant r = -.53. Mitchell B. Chamlin and John K. Cochran, “Social Altruism and Crime,” Criminology 35 (1997): 203–227, report that (controlling for other relevant factors, such as poverty, inequality, race, residential mobility, and family structure) crime is lower in cities where the ratio of United Way contributions to city income is higher, another indication of social capital.
5. Sheldon Hackney, “Southern Violence,” American Historical Review 73 (1969): 906–925, quotation at 925; Richard E. Nisbett and Dov Cohen, Culture of Honor: The Psychology of Violence in the South (
Boulder, Colo.: West-view Press, 1996); Raymond D. Gastil, “Homicide and a Regional Culture of Violence,” American Sociological Review 36 (1971): 412–427; Steven F. Messner, “Regional and Racial Effects on the Urban Homicide Rate: The Subculture of Violence Revisited,” American Journal of Sociology, 88 (1983): 997–1007; and (for a critical view) Colin Loftin and Robert H. Hill, “Regional Subculture and Homicide: An Examination of the Gastil-Hackney Thesis,” American Sociological Review 39 (1974): 714–724.
6. This conclusion is based on extensive multivariate predictions of the murder rate 1980–95 in the fifty states, based on poverty rates, income level, income inequality, educational levels, degree of urbanism, and racial composition, along with our standard measure of social capital and the north-south distinction. In virtually all specifications, when social capital is introduced, the north-south distinction becomes insignificant. The most robust predictors across various models are the percentage of nonwhites in the population, the poverty rate, urbanism, and social capital, all of equivalent significance. Among the thirty-nine states outside the old Confederacy, the bivariate correlation between social capital and the murder rate is a very strong r = -.74.
7. Author’s analysis of DDB Needham Life Style surveys. Here too the effects of social capital masquerade as regional differences. Southerners appear more pugnacious than northerners, but once we control for differences in social capital, those regional differences disappear, whereas if we control for region (by looking only at northern states, for example), the negative correlation between social capital and physical truculence persists.