Robert L. Kaufman

Economic Dualism: A Critical Review

In this paper we examine the model underlying the dual economy approach to labor market segmentation. We provide a specification of the components of the model as well as the linkages between components. This construction of the underlying model allows us to evaluate it systematically as a model, to consider and to evaluate each component and linkage separately. While we conclude that there are several important contributions to retain from the dual approach, including the insight that capital structure conditions labor market outcomes, we argue that it is inadequate for sustaining theoretical and empirical development. As a new starting point for the study of economic and labor market segmentation we suggest a resource perspective which retains insights from the dual approach but avoids its overly restrictive assumptions.

Housing and wealth inequality: Racial-ethnic differences in home equity in the United States

In our study, we took a first step toward broadening our understanding of the sources of both housing and wealth inequality by studying differences in housing equity among blacks, Hispanics, Asians, and non-Hispanic whites in the United States. Using data from the American Housing Survey, we found substantial and significant gaps in housing equity for blacks and Hispanics (but not for Asians) compared with whites, even after we controlled for a wide range of locational, life-cycle, socioeconomic, family, immigrant, and mortgage characteristics. Furthermore, the payoffs to many factors are notably weaker for minority than for white households. This finding is especially consistent across groups for the effects of age, socioeconomic status, and housing-market value. Blacks and Hispanics also uniformly receive less benefit from mortgage and housing characteristics than do whites. These findings lend credence to the burgeoning stratification perspective on wealth and housing inequality that acknowledges the importance of broader social and institutional processes of racial-ethnic stratification that advantage some groups, whites in this case, over others.

Assessing Alternative Perspectives on Race and Sex Employment Segregation

Four major explanations for employment segregation-skill deficits, worker preferences, economic and organizational structure, and stereotyping/queuing-are assessed using a diverse and overlapping set of predictors: general skills and training, product market structure, race- and sex-typed tasks and conditions, desirable employment and growth rates, and links to other labor market actors. A two-stage measurement and analytic strategy controls for relevant worker-level factors. Data from the 1990 census PUMs are analyzed to measure the employment segregation of black women, black men, and white women in relation to white men across 1,917 labor market positions, net of human capital, family structure, geographic residence and labor supply. Archival data provide measures of variables characterizing labor market positions. Stereotyping and queuing explanations are broadly consistent with nearly all results, while a worker preference approach applies to somewhat fewer predictors and is largely but not wholly compatible with their effects. A skill deficits explanation applies to, and is supported by, a narrow set of findings, while the economic and organizational structure explanations are restricted in their relevance and receive limited support

How low can it go? Declining black-white segregation in a multiethnic context.

We extend research on whites’ neighborhood contact with blacks, population composition, and prospects for desegregation by developing a new measure of the floor of racial residential segregation under conditions of low black-white contact. The measure incorporates the way in which multi ethnic contexts further constrain levels of black-white segregation. The results show that black-white desegregation is likely when the black population is small, but is unlikely otherwise. Yet, when multiple ethnic groups are sufficiently large, a moderate level of black-white segregation is necessary for whites to maintain low neighborhood contact with blacks, even when the proportion of African Americans is small.

A Structural Decomposition of Black-White Earnings Differentials

This research demonstrates the impact of labor market structure and segmentation on differentials in black and white earnings. This paper argues that there are two very different factors which create black-white earnings differences: (1) differences between blacks and whites within divisions of the labor market, and (2) differences between labor market divisions in earnings combined with the differential distribution of blacks and whites across labor market divisions. Use of a decomposition based on a regression standardization approach discloses that the second factor accounts for a minimum of 14% of the black-white earnings gap. This implies that eliminating all black-white differences within labor market divisions would still leave a significant earnings gap between blacks and whites due to the structure of the labor market.

USING ADJUSTED CROSSTABULATIONS TO INTERPRET LOG-LINEAR RELATIONSHIPS*

Log-linear analysis was developed as a more powerful technique for the analysis of multivariate tables of categorical variables. It was specifically designed to deal with some of the problems that plagued users of crosstabulations. of the major drawbacks to the traditional crosstabulation approach is the difficulty of controlling for the effects of more than a single other variable. This is often due to small sample sizes within cells of a crosstabulation entailing further control variables. In addition, since there is no direct test of the significance of the effect of a single independent variable, it is difficult to interpret the import of a series of crosstabulations of two variables within the categories of other control variables. Instead, there are as many test statistics as there are combined categories of the control variables. Thus, it is often unclear whether an effect is produced by an independent variable or through its interaction with control variables. Goodman (1970, 1972a, b) presented the first comprehensive treatment of log-linear analysis as a response to the problems of crosstabular analysis. About the same time, logistic analysis was developed as an alternative to crosstabular analysis for the case where the dependent variable is categorical and the independent variables are interval (Bishop et al., 1975; Fienberg, 1980; Nerlove and Press, 1973).1 Log-linear analysis, developed for the situation where all of the variables are categorical, overcame most of the problems of crosstabular analysis. Log-linear analysis provides better and more appropriate statistical tests. It provides a global test for each independent variable rather than a series of tests as was the case for crosstabular analysis. In addition, the results allow for a test of main effects separate from interaction effects. Moreover, log-linear surpasses crosstabulation by simultaneously estimating the effects of multiple variables, much in the same way that multiple regression is an advance over bivariate correlation. But in the process of moving from the simplicity of a crosstabular analysis to the sophistication of a log-linear analysis something was given up. In contrast to crosstabulation, log-linear analysis suffers from a lack of intuitive interpretability of the results. By an appropriate percentaging of the table by rows or columns, crosstabulations provided a straightforward estimate of how the distribution of the dependent variable changed across levels of the independent variables. On the other hand, researchers often find it difficult to interpret log-linear parameters and to communicate this interpretation intelligibly to readers who are

Going up the Ladder: Multiplicity Sampling to Create Linked Macro-to-Micro Organizational Samples

We argue that researchers should use representative samples to address many issues and that long-standing interest in the connections between macrolevel and microlevel processes is also central to organizational analysis. Our literature review suggests that designs that link organizations to suborganizational units or members have deficiencies involving atypicalness of cases studied or inadequate and unreliable data on organizations derived from more representative samples of individuals. Instead we

Risk-based crime statistics: A forecasting comparison for burglary and auto theft

Abstract A major criticism of official statistics on crime is that they use inappropriate bases for computing rates. Here we investigate whether computing crime rates that contain in their denominators the number of exposures to risk of a specific event (e.g., residential burglary and auto theft) provides more accurate forecasts than employing the traditional FBI denominators as a base (e.g., the number of auto thefts and burglaries per 100, 000 persons living in the United States). Single equation, macrodynamic structural models are fitted to both the “traditional” and “alternative” forms of computing auto theft and burglary rates over the twenty-seven-year period from 1947–1974, in order to determine how well they perform on statistical and substantive grounds over the estimation period. Ex-post forecasts of the 1975–1979 observed crime rates, used to gauge the accuracy of these models, reveal few differences between the two kinds of rates in terms of how well they forecast. Both types of rates forecast well with the exogenous variables employed here and lead to similar substantive conclusions. The forecasts of the “traditional” rates are consistently, but only slightly, more accurate than those of the “alternative” rates (in most cases the differences are less than 1 percent). It is argued that the criticism of official data may be overstated and that little benefit accrues from the modification of the rate base for some purposes.

Variations on a Theme: More Uses of Odds Ratios to Interpret Log-Linear Parameters

This article presents an expository discussion of the use of odds, odds ratios, and functions of odds ratios as an aid to the interpretation of log-linear parameters. While there have been some previous presentations on the use of odds ratios, this method has never been fully and systematically developed for complex situations, nor have the potential problems of interpretation using odds ratios been discussed. The presentation in this article more fully elaborates the possible modes of drawing contrasts with odds ratios using any of the three common parameterizations of a log-linear model (ANOVA-like, regression-like, and logit). The sampling distribution and the calculation of standard errors for odds ratios are also discussed as are some cautions on the interpretation of odds ratios. A data analysis of unemployment using polytomous variables illustrates the application and interpretation of the odds-ratio approach.

Issues in Multivariate Cluster Analysis

Using a Monte Carlo simulation, the research in this article addresses two key questions about the accuracy of cluster analysis in reproducing a known true cluster model. First, how is the accuracy affected by different ways of measuring interunit similarity; in this case, different ways of using principal components analysis. Second, how is the accuracy affected by the quality of the characteristics data and by different procedures for handling missing information? The results indicate that using principal components analysis is superior to not using it and that the choice of how to utilize the principal components results may be critical. The results also indicate that the impact of data quality differences may be minimal, but that there are important differentials among the procedures for handling missing data.