The Impact of Shared Religious Affiliation on the Rate of Currently Divorced in the United States in 1990 and 2000

Larry C. Mullins, Kimberly P. Brackett, Nelya J. McKenzie, and Donald W. Bogie
Auburn University Montgomery

Introduction

[1] While many factors influence the rate of divorce in the United States (Cherlin; Eshleman; South 2001; South, Trent, and Shen; Sweeney and Phillips), recent research has indicated that the list of factors may be extended to include the impact of shared religious affiliation (Mullins, Brackett, Bogie, and Pruett 2004, 2006). Extending the previous work done on this issue, the current paper examines: (a) the impact of shared religious affiliation on the divorce rate in both 1990 and 2000; and (b) the extent of change between 1990 and 2000 in the divorce rate and the level of shared religious affiliation. The essential issue in this research: Does the degree of shared religious affiliation influence the divorce rate at the county level?

Conceptual and Theoretical Considerations

[2] A greater degree of shared religious affiliation generally enhances the likelihood of friendships, networks, and general mutual support systems for those in the same religious group. Likewise, expectations extending from the community would be consistent with the religious ideas of the homogenous group and could give direction and focus to married couples. Further, the prevailing religious context, over and above the influence of other variables (for which data are available), may help to stabilize marital relationships, reflected by a lower rate of divorce. Specifically, we expect that the rate of currently divorced persons at the county level will be lower when there is a greater degree of shared religious affiliation of formal religious groups, i.e., when relatively more persons are adherents to a fewer number of congregations, and conversely, even after controlling for established covariates of divorce. This is consistent with recent work by Gruber, who suggests in his research that divorce rates are lower where there is greater church attendance, which is part of his idea of “religious density.”

[3] Durkheim’s theories of religion (1965) and suicide (1951) could provide at least a partial explanation of the impact of religion on marriage. Booth, Johnson, Branaman, and Sica contend that these two theories provide a basis for discussions of the way in which religion decreases suicide and how they may apply to the thinking about the impact of marriage on suicide. Durkheim’s theory of suicide addressed the impact of institutional expectations on individual behavior, i.e., the influence of marriage on taking one’s own life, while his theory of religion emphasizes its effects on society rather than the individual. Succinctly, while religion does not provide a normative basis for influencing marriage, it encourages an “intense collective life” (Durkheim 1951: 150) and could serve as a positive and integrative force on human society (D’Antonio; Hargrove) that helps shield persons from the conflicting expectations that are associated with the institution of marriage in modern society. While Durkheim views religion and marriage as independent, they are nevertheless integrative in their effects on society.

Methods

Design and Sample

[4] The analysis in this study is based on a 20 percent random sample of counties drawn from each of the 50 states, using 1990 as the benchmark – 621 of the 3,111 U.S. counties in 1990 were selected for the original sample. For the sake of consistency in comparisons, data from the same 621 counties were subsequently compiled for the 2000 data set. The data used in calculating the shared religious affiliation index for the 621 counties for both 1990 and 2000 were drawn from the extensive databases of United States church membership/religious affiliation that have been constructed by the Glenmary Research Center (Bradley et al.; Jones et al.). All other variables utilized in this analysis are from the 1990 and 2000 Censuses of Population and Housing (U.S. Bureau of the Census 1991, 1992a, 1993, 2001, 2002).

[5] The 621 counties is sufficiently large to be both representative of the United States as a whole and to supply sufficient numbers for the analysis of the variables that are included in the 1990 and 2000 data sets. A complete set of data was compiled for each selected county for each variable for the years 1990 and 2000.

Dependent Variable: Rate of Persons Currently Divorced, 1990 and 2000

[6] The divorce rate for both 1990 and 2000, derived from the 1990 and 2000 decennial censuses using the question on current marital status that was asked on the census short form, measures the total number of persons who are currently divorced at these two points in time per 1,000 population aged 15 and older. The divorce measure utilized in this analysis does not include multiple instances of divorce, nor does it reflect rates that are confined to a specific period of time (such as over one year).

The Key Independent Variable: Shared Religious Affiliation, 1990 and 2000

[7] Following the approach utilized by Ellison, Burr, and McCall, the independent variable of primary interest is derived from a statistical index that identifies the extent of shared religious affiliation within each of the counties included in the sample. Shared religious affiliation is operationally defined as an index of the concentration of individual religious denominations within a county as determined from the Glenmary data. It is measured through an adaptation of the “Herfindahl Index,” which was first used in antitrust policy in the determination of monopolistic share.

[8] The “Herfindahl Index” was originally designed to measure the extent of corporate concentration within a given market area (Herfindahl). The index showed the share of each firm competing in a market by squaring the individual shares and summing the results.

[9] For purposes of the present research, affiliation with various religious denominations is substituted for market share; thus, the Index of Shared Religious Affiliation becomes a measure of the overall degree of shared religious orientation. The general formula for the Shared Religious Affiliation Index is: SRAj =∑N2ij. N represents the number of adherents in each denomination within a county divided by the total number of church adherents in that county; i represents the index of summation over all religious denominations in county j. SRA represents the probability that any two persons, selected at random, within a county are affiliated with the same organized religious group (Iannacone).

[10] For example, assume that a given county has four discrete religious group affiliations, e.g., Baptist, Methodist, Catholic, and Jewish, each with a 20 percent market share of adherents. The Index of Shared Religious Affiliation for the county is SRA=.202 +.202 +.202 +.202 or .16. Thus, if two persons were selected at random from that county, the odds are 16 percent that they would be adherents of the same recognized denominational group. Another example is a county that has five discrete religious groups with market shares of 50, 20, 15, 5, and 5 percent, e.g., Catholic, Mormon, Presbyterian, Jehovah’s Witness, and Muslim, respectively. The index then would be SRA = .502 +.202 +.152 +.052 +.052, or .32. In this case, the odds are 32 percent that two people selected at random will have the same religious affiliation.

[11] Theoretically, the index scores can range from 0.00 (no adherents with any affiliations) to 1.00 (all adherents have a single affiliation). For this analysis, we have multiplied the Index score by 1,000 to facilitate ease of use. Thus, the Index of Shared Religious Affiliation used here can range from 0 to 1,000.

Covariates of Divorce

[12] In order to examine more clearly the effects of shared religious affiliation on divorce, the impact of other potential influences must be held constant. Previous research has shown that divorce rates are associated with higher levels of geographic mobility and, by extension, lower levels of community involvement and integration (Breault and Kposawa; Glenn and Shelton). Percent population change and percent urban, therefore, are used as indicators of population mobility and community cohesiveness. It also has been generally established that a higher concentration of young adults in the population is associated with an elevated divorce rate (Martin and Bumpass); thus, we include the percentage of the population aged 15-34. Likewise, race and ethnicity are shown to have an impact on divorce rates (U.S. Bureau of the Census 1992b; Sweeney and Phillips). We use percentage white and percentage Hispanic as measures of race and ethnicity. The African-American percentage is excluded due to its high correlation (r = -.82 in 1990 and -.81 in 2000) with the white percentage.

[13] Previous research also shows that the relative concentration of males and females within a population influences the divorce rate (Guttentag and Secord; Trent and South). Here, we use the percentage of females in the population as an indicator of gender concentration. Also, given evidence that the level of economic instability is positively associated with the divorce rate (Martin and Bumpass; Sayer and Bianchi; South 1985), we include two census-based measures to represent different dimensions of this variable (both of which are sensitive to economic fluctuations): percentage of the civilian labor force employed in manufacturing, and percentage unemployed.

[14] Regional differences in the propensity toward divorce also have been observed since the nineteenth century (Eshleman). Generally, the divorce rate increases from east to west. Rank-ordering divorce rates by region, the rate is lowest in the Northeast, moderate in the Midwest, next to highest in the South, and highest in the West. It is widely assumed that the divorce rate is directly related to the regional level of cultural similarity, i.e., the greater the level of similarity, the lower the rate of divorce. Traditionally, there has been a higher level of cultural similarity and normative agreement in the older parts of the United States than in the developing areas of the West (Glenn and Shelton). Likewise, higher geographic mobility is associated with the western region of the U.S., suggesting another basis for the elevated divorce rates observed in that part of the country.

[15] The four major regions that are historically utilized by the U.S. Census Bureau in reporting population data are used in our analysis: Northeast, South, West, and Midwest. Counties are identified as either being in a designated region, i.e., Northeast, Midwest, South, or West, or not in that region. Those counties identified as being in a particular region are coded as Yes = 1, while counties not in that region are coded as No = 0. For each of the covariates noted above, data for both 1990 and 2000 are utilized.

Descriptive Results

[16] Examination of the intercorrelations of the variables for 1990 and for 2000 indicates a relatively independent set of variables for each time frame (see Table 1). No correlation between any two variables in 1990 is greater than -.424 (the correlation between percent white and residence in the South). A similar conclusion can be drawn concerning the intercorrelations between the variables in the 2000 data set; no correlation is greater than -.542 (the correlation between percent white and percent unemployed). The multiple linear regression analyses in both 1990 and 2000 utilize a reasonably independent set of variables relative to their effects on the dependent variables, i.e., the rate of currently divorced persons in U.S. counties in 1990 and 2000. Multicollinearity is not an issue in the analysis.

Table1. Intercorrelations of Variables in 1990 and 2000 (N=621)a,b

Variables	1	2	3	4	5	6	7	8	9	10	11	12	13	14
Divorce Rate (1)	1.000	-.135	.211	.048	-.002	-.104	.021	.160	.152	.147	.136	-.226	-.053	.146
Shared Religious Affiliation (2)	-.144	1.000	.026	-.039	-.225	.156	-.034	.138	-.052	-.134	.056	-.307	.068	.218
%Population Change (3)	.408	.001	1.000	.250	-.042	.119	-.288	-.049	-.020	.186	.281	-.243	-.120	.094
%Age 15-34 (4)	.246	-.092	.370	1.000	-.296	.144	-.245	.250	.080	.480	-.020	-.182	.006	.185
%White (5)	-.084	-.065	-.055	-.233	1.000	-.199	.009	-.542	.075	-.141	-.063	.421	.088	-.403
%Hispanic (6)	.028	.143	.114	.029	-.173	1.000	-.207	.150	-.307	.299	.285	-.259	-.034	.062
%Female (7)	-.171	-.089	-.248	-.179	-.033	-.118	1.000	.035	.269	.070	-.224	.062	.073	.062
%Unemployed (8)	.127	.063	-.062	.058	-.307	.162	.056	1.000	-.049	.011	.123	-.247	.003	.147
%Manufacturing Occupations (9)	.103	-.010	.010	.165	-.027	-.288	.287	.013	1.000	-.049	-.388	.110	.026	.158
%Urban (10)	.336	-.109	.255	.420	-.106	.280	.133	.007	-.054	1.000	.164	-.110	.136	-.082
West (11)	.318	.026	.233	-.060	-.022	.265	-.337	.015	-.327	.119	1.000	–	–	–
Midwest (12)	-.253	-.250	-.254	-.198	.409	-.234	.040	-.139	-.028	-.094	–	1.000	–	–
Northeast (13)	-.048	.108	.013	.092	.096	-.020	.037	-.010	.052	.113	–	–	1.000	–
South (14)	.040	.164	.069	.185	-.424	.045	.182	127	.232	-.053	–	–	–	1.000

a Correlations in the upper right quadrant are from 2000 data; correlations in the lower left quadrant are from 1990 data.
b Correlations greater than or equal to .080 are significant at the p < (at least) .05 level of significance (two-tailed).

Multiple Linear Regression Analysis of the 1990 Data

[17] The first issue concerns the multiple linear regression analysis results using the 1990 data (see Table 2). The 12 independent variables together accounted for 39.9 percent of the explained variance in the 1990 divorce rate at the county level (p < .01, 12/608 df).

[18] When the 1990 divorce rate is regressed on these social, economic, and geographic factors, the index of shared religious affiliation in 1990 is a statistically significant (p < .01) predictor (Beta = -.159). That is, independent of the other variables in the equation, the greater the degree of shared religious affiliation in 1990, the lower the rate of currently divorced persons at the county level in 1990. Other factors from the 1990 data set that are significantly related to higher county-level divorce rates in 1990 include greater percent urban population, residence in the West, Midwest, and Northeast, greater percent population change from 1980-1990, greater percent unemployed, greater percent employed in manufacturing occupations, lesser percent Hispanic, lesser percent female, and lesser percent aged 15-34. Not statistically related to the 1990 divorce rate is percent white.

Table 2. Multiple Linear Regression Analysis of the 1990 Divorce Rate on Included Variables (N=621)

Coefficients
Model		Unstandardized Coefficients		Standardized Coefficients	t	Sig.
		B	Std. Error	Beta
1	(Constant)	162.928	21.333		7.637	.000
	Shared Religious Affiliation	-.020	.004	-.159	-4.736	.000
	%Population Change	.291	.045	.237	6.479	.000
	%Age 15-34	-.317	.161	-.082	-1.972	.049
	%White	.064	.042	.057	1.509	.132
	%Hispanic	-.292	.070	-.152	-4.161	.000
	%Female	-1.862	.372	-.195	-5.008	.000
	%Unemployed	.656	.127	.173	5.160	.000
	%Manufacturing Occupations	.351	.062	.204	5.622	.000
	%Urban	.220	.026	.346	8.613	.000
	West	10.016	2.075	.191	4.826	.000
	Midwest	-7.828	1.622	-.200	-4.826	.000
	Northeast	-6.794	2.466	-.094	-2.755	.006

R2 = .399 (12, 608 df), F = 33.625, p < .01

[19] Rank ordering the standardized slopes from greatest to least impact on the divorce rate, among the variables that are statistically significant reveals that greater percentage urban has the greatest impact. The Index of Shared Religious Affiliation is ranked eighth, ahead of percent white, percent Hispanic, percent aged 15-34, and residence in the Northeast.

Multiple Linear Regression Analysis of the 2000 Data

[20] Table 3 shows for 2000 the multiple linear regression analysis results, using the same set of independent variables as those in 1990. All but one of the independent variables, i.e., residence in the West, is significantly related to the divorced rate. Altogether, the 12 variables account for 27.5 percent of the variance in the divorced rate in 2000 (p < .01, 12/608 df).

[21] The Index of Shared Religious Affiliation once again shows a significant, inverse association with divorce. Hence, the lesser the concentration of county-level denominational affiliation (Beta = -.162), the higher the rate of currently divorced. Also, significantly related to a higher divorce rate in 2000 are greater percent population change from 1990-2000, lesser percent aged 15-34, greater percent white, lesser percent Hispanic, lesser percent female, greater percent unemployed, greater percent employed in manufacturing occupations, greater percent urban, and not residing in the Midwest and in the Northeast.

Table 3. Multiple Linear Regression Analysis of the 2000 Divorce Rate on Included Variables (N=621)

Coefficients
Model		Unstandardized Coefficients		Standardized Coefficients	t	Sig.
		B	Std. Error	Beta
1	(Constant)	99.732	18.739		5.322	.000
	Shared Religious Affiliation	-.017	.004	-.162	-4.294	.000
	%Population Change	.134	.041	.130	3.248	.001
	%Age 15-34	-.818	.183	-.212	-4.460	.000
	%White	.228	.047	.229	4.887	.000
	%Hispanic	-.347	.064	-.220	-5.411	.000
	%Female	-.683	.324	-.089	-2.105	.036
	%Unemployed	1.981	.275	.315	7.212	.000
	%Manufacturing Occupations	.346	.075	.184	4.593	.000
	%Urban	.166	.027	.291	6.163	.000
	West	.924	2.067	.019	.447	.655
	Midwest	-13.090	1.602	-.367	-8.169	.000
	Northeast	-10.675	2.534	-.162	-4.213	.000

R2 = .275 (12, 608 df), F = 19.196, p< .01

[22] Again, rank ordering the standardized slopes (Betas) from greatest to least impact on the divorce rate shows that not residing in the Midwest has the greatest effect on the rate of currently divorced in 2000. Of the 12 variables, the Shared Religious Affiliation Index is tied for eighth in the rank-order of the relative impact on the divorce rate at the county level.

Paired t –Test Comparisons Between the 1990 and 2000 Variables

[23] Table 4 presents the results of the differences between the means for each of the ten substantive variables used in the 1990 and 2000 analyses. The results show a statistically significant change for nine of the variables. The divorce rate did not show a statistically significant change in the means between 1990 and 2000. The divorce rate increased by 0.7 persons per 1,000 population from 1990 to 2000.

Table 4. Paired t – Test Results Between 1990 and 2000 Data (N=621)

	1990		2000
Variables	M	SD	M	SD	t
Divorce Rate	75.5	18.6	76.2	16.9	-1.36
Shared Religious Affiliation	297.3	46.2	325.8	63.3	-8.06*
%Population Change	3.6	15.2	10.0	6.4	-15.10*
%15-34 Years	28.6	4.8	25.6	4.4	30.24*
%White	86.0	16.6	85.5	16.0	10.66*
%Hispanic	4.1	9.7	5.8	10.7	-17.50*
%Female	50.9	2.0	50.4	2.2	8.37*
%Unemployed	6.9	4.9	5.8	2.7	6.59*
%Manufacturing Occupations	18.6	10.8	15.8	9.0	15.72*
%Urban	34.8	29.3	38.4	29.6	-7.60*

*p<.001 (2-tailed)

[24] Between 1990 and 2000, there was a significant increase observed for the Shared Religious Affiliation Index (297.3 compared to 325.8). Significant increases between 1990 and 2000 were observed also for percent population change, percent Hispanic, and percent urban population. On the other hand, there were significant decreases over the 10-year time frame in percent population aged 15-34, percent white, percent female, percent unemployed, and percent employed in manufacturing occupations.

[25] Overall, these data indicate little change in the divorce rate over the ten-year period, but a higher degree of shared religious affiliation. Particularly noteworthy is the apparent stability of the divorce rate. While an increasingly higher divorce rate has marked much of the twentieth century (most visibly observed immediately following World War II and during the 1960s and 1970s), it began to level off during the 1980s (Eshleman). Equally noteworthy is the indicated increase in the county level Index of Shared Religious Affiliation.

Discussion and Conclusions

[26] Clearly, our findings indicate that the extent of shared religious affiliation is a separate and independent factor in the explanation of the rate of currently divorced at the county level. In the multiple linear regression analyses for both time periods, the Index of Shared Religious Affiliation is inversely related to the divorce rate, i.e., the greater the county-level concentration of shared religious affiliation, the lower the rate of the currently divorced in these counties. Furthermore, based on the comparison of the 1990 and the 2000 mean scores, the divorce rate is not substantially, nor significantly, different in 1990 than in 2000. The Shared Religious Affiliation Index, however, is significantly greater in 2000 than in 1990.

[27] It is clear that the Index of Shared Religious Affiliation, our independent variable of focus, continues to impact the likelihood of divorce. While this factor is not one of the strongest correlates of divorce, its effect in the two time periods is significant and appears undiminished. In fact, the extent of shared religious affiliation may be gaining strength in the United States. This may be one of the reasons that it continues to have an effect on the divorce rate.

[28] However, this observation must be tempered by possible deficiencies in the data. Many of the smaller, less formally organized religious groups in the U.S. are not included in the Glenmary dataset; hence, our data may indicate more shared religious affiliation in American society than is really the case. In fact, the same growing individualism that may have an impact on divorce also would appear to be characteristic of religion as well. Indeed, in an increasingly multi-faceted society, the religious system may be characterized by a greater variety and less overall influence by any one group.

[29] Perhaps this makes it even more noteworthy that there is a sustained association between shared religious affiliation and the divorce rate. That shared religious affiliation continues to have an impact on the divorce rate in 2000, particularly in a society where rates of divorce are by any measure elevated, adds an extra measure of validity to the findings. Despite the problems that institutionalized religion has recently endured, e.g., the highly publicized morality failings of well-known conservative ministers and wide-scale accusations of child abuse within the Roman Catholic Church, religion is one of the few large institutional structures, if not the only one, whose “message” still exerts impact in a world of competing messages and normative confusion.

[30] But this also may be one of the strengths of religion in our society. In an era of increasing anomie, people apparently are searching for a strong, singular voice that shields the human experience from conflicting demands and confusing expectations. Thus, religion, as Durkheim posited, “protects man against the desire for self-destruction . . . because it is a society” (1951: 170). While divorce is obviously not always “destructive” and, conversely, religion may serve to keep people together in destructive situations, the overall impact of religion on marriage nevertheless appears to be one of enhancing family systems, and by extension, societal stability. As Durkheim stated, “The stronger the integration of the religious community, the greater its preservative value” (1951: 170). The current research lends support to this position.

Bibliography

Booth, Alan, David Johnson, A. Branaman, and A. Sica

1995 “Belief and Behavior: Does Religion Matter in Today’s Marriage.” Journal of Marriage and the Family 57: 661-73.

Bradley, Martin, Norman Green, Jr., Dale Jones, Mac Lynn, and Lou McNeil

1992 Churches and Church Membership in the United States 1990: An Enumeration by Region, State and County Based on Data Reported for 133 Church Groupings. Atlanta: Glenmary Research Center.

Breault, Kevin D., and Augustine J. Kposawa

1987 “Explaining Divorce in the United States: A Study of 3,111 Counties, 1980.” Journal of Marriage and the Family 49: 549-58.

Cherlin, Andrew

1996 Public and Private Families: An Introduction. New York: McGraw-Hill.

D’Antonio, William

1983 “Family Life, Religion, and Societal Values and Structures.” Pp. 81-108 in Families and Religion: Conflict and Change in Modern Society. Edited by W. D’Antonio and J. Aldous. Beverly Hills: Sage.

Durkheim, Emile

1951 Suicide: A Study in Sociology. New York: Free Press.

1965 The Elementary Forms of Religious Life. New York: Free Press.

Ellison, Christopher G., Jeffery Burr, and Patricia McCall

1997 “Religious Homogeneity and Metropolitan Suicide Rates.” Social Forces 76: 273-99.

Eshleman, Ross J.

2003 The Family. Tenth edition. Boston: Allyn and Bacon.

Glenn, Norval, and Beth Ann Shelton

1985 “Regional Differences in Divorce in the United States.” Journal of Marriage and the Family 47: 641-52.

Gruber, Jonathon

2005 “Religious Market Structure, Religious Participation, and Outcomes: Is Religion Good for You?” Advances in Economic Analysis & Policy 5, article 5. Available online at http://www.bepress.com/bejcap/advances/vol5/issue1/art5.

Guttentag, Marcia, and Paul Secord

1983 Too Many Women? The Sex Ratio Question. Beverly Hills: Sage.

Hargrove, Barbara

1983 “The Church, the Family, and the Modernization Process.” Pp. 21-48 in Families and Religion: Conflict and Change in Modern Society. Edited by W. D’Antonio and J. Aldous. Beverly Hills: Sage.

Herfindahl, Orris

1950 Concentration in the Steel Industry. Unpublished Dissertation, Columbia University, New York.

Iannocone, Lawrence R.

1991 “The Consequences of Religious Market Structure: Adam Smith and the Economics of Religion.” Rationality and Society 3: 156-77.

Jones, Dale E., Sherri Doty, Clifford Gammich, James E. Horsch, Richard Houseal, Mac Lynn, John P. Marcum, Kenneth M. Sachagrin, and Richard H. Taylor

2002 Religious Congregations and Membership in the United States 2000: An Enumeration by Region, State and County Based on Data Reported for 149 Religious Bodies. Nashville: Glenmary Research Center.

Martin, T. Castro, and Larry L. Bumpass

1989 “Recent Trends in Marital Disruption.” Demography 26: 37-51.

Mullins, Larry C., Kimberley P. Brackett, Donald W. Bogie, and Daniel Pruett

2004 “The Impact of Religious Homogeneity on Divorce in the United States.” Sociological Inquiry 74: 338-54.

2006 “The Impact of Concentrations of Religious Denominational Affiliations on the Rate of Currently Divorced in the United States.” Journal of Family Issues 27: 976-1000.

Sayer, Liana, and Suzanne Bianchi

2000 “Women’s Economic Independence and the Probability of Divorce: A Review and Reexamination.” Journal of Family Issues 21: 906-43.

South, Scott

1985 “Economic Conditions and the Divorce Rate.” Journal of Marriage and the Family 4: 31-41.

2001 “Time Dependent Effects of Wives’ Employment on Marital Disruption.” American Sociological Review 66: 226-45.

South, Scott, Katherine Trent, and Yang Shen

2001 “Changing Partners: Toward a Macrostructural-Opportunity Theory of Marital Dissolution.” Journal of Marriage and Family 63: 743-54.

Sweeney, Megan, and Julie Phillips

2004 “Understanding Racial Differences in Marital Disruption: Recent Trends and Explanations.” Journal of Marriage and Family 66: 239-50.

Trent, Katherine, and Scott South

1989 “Structural Determinants of the Divorce Rate: A Cross-Sectional Analysis.” Journal of Marriage and the Family 52: 983-92.

U.S. Bureau of the Census

1991 1990 Census of Population and Housing. Population and Housing Counts, CPH-2 Series. Washington, DC: Government Printing Office.

1992a “Marriage, Divorce, and Remarriage in the 1990s.” Current Population Reports, P23-180. Washington, DC: U.S. Government Printing Office.

1992b “Marital Status and Living Arrangements.” Current Population Reports, Series P-20, No. 468. Washington, DC: Government Printing Office.

1993 Census of the Population: 1990. Social and Economic Characteristics. CP-2 Series. Washington, DC: Government Printing Office.

2001 “Census 2000 Summary File 1.” Available online at http://www.census.gov/Press-Release/www/2001/sumfile1.html.

2002 “Census 2000 Summary File 3.” Available online at http://www.census.gov/Press-Release/www/2002/sumfile3.html.