Recursive models for categorical data

Standard methods for recursive models with continuous endogenous variables are extended to models with categorical endogenous variables. The concept of a reduced-form equation is generalized in a natural way to cover nonlinear regression functions and, in particular, models with categorical endogenous variables. Maximum-likelihood estimation and asymptotic chi-square tests are described. Two numerical examples are presented: a linear recursive two-equation model for all-categorical data, and a combined linear and logit three-equation recursive model with both categorical and continuous endogenous variables. Limitations of the present work and directions for further extension are noted.     

Some models of response uncertainty for panel analysis

Data of L. M. Wiggins from three-wave panels, each with a single dichotomous response, illustrate the use of models with response probabilities that vary over occasions or over individuals, or neither, or both, with a “no interaction” combination of the two being specified for the last case, which can also be derived from the Rasch measurement model. Models more complicated than these, allowing for changes in individual parameters (interaction of occasions and persons) or serial dependence of responses are considered when the Rasch-type model does not adequately describe the data.     

Confession of prior knowledge about experimental procedures as a function of evaluation apprehension and commitment

Subjects' evaluation apprehension (anxiety about being judged by the experimenter) and commitment to (involvement in) the experiment were manipulated in an experimental study of subjects' honesty in postexperimental interviews. Previous studies have shown that subjects are typically dishonest in postexperimental interviews by refusing to admit prior knowledge of the experiment imparted to them by a confederate. It was found that both high apprehension and low commitment lead to greater subject honesty. Other features of the experimental design were successful in inducing subjects to be generally more honest than their counterparts in previous confederate tip-off studies.     

Dynamic models and attainment in the workplace

The two most marked trends in recent stratification studies are: (1) the shift away from a statical approach to attainment and toward a dynamical representation of achievement, and (2) the shift away from the assumption that achievement is largely a matter of individual characteristics, and toward the view that achievement is the outcome of an employer-employee exchange of productive resources for earnings and status. This paper forges a link between these parallel trends by elaborating on previous formulations of dynamic models of achievement and applying the results to the analysis of earnings attainment in an internal labor market. The modeling section of the paper joins within a single framework the growing interest in ascertaining how a given structure of opportunity shapes achievement and in determining the different points in the career line at which individual background and resource variables impact attainment. Special attention is devoted to the problems facing researchers who wish to bring a dynamic conceptualization of achievement to cross-sectional or otherwise deficient data. Although the empirical application of the various models is largely meant to be illustrative, it is of interest in its own right because it goes substantially beyond previous efforts in this area.     

A dynamic macro social indicator model of changes in marriage, family, and population in the United States: 1947–1974

This paper presents an integrated 21-equation model of how marriage, family, and population conditions, as indexed by macro social indicators, affect each other and are affected by other social, demographic, and economic forces. An opportunity structures theoretical paradigm is applied to the specification of dynamic structural equations for determining changes (both trends and cyclical fluctuations) in marriage formation and dissolution, family and household composition, fertility, mortality, population growth, and population distribution. The equations are estimated on annual national data for the United States during the post-World War II years 1947 to 1972, and then they are used to make conditional forecasts of the values of some of the endogenous variables for 1973 and 1974. It is found that the equations fit the observed data well, lack demonstrable autocorrelation of disturbances, and forecast the 1973 and 1974 values usually with less than 2% error. Strategies are sketched for further refining some of the equations, and it is suggested that this model should be integrated into a larger societal model in order to estimate some of the effects of changes in demographic phenomena on other social conditions.     

The use of nativity data to estimate ethnic characteristics and patterns

Until recently, nativity data for the first and second generations (the foreign stock) have been the primary source of census information on white ethnic groups in the United States. With the development of an ancestry item, it is now possible to deal with all generations regardless of birthplace. The 1979 Current Population Survey provides a rare opportunity to cross-tabulate the results obtained from each system in order to determine the degree of overlap in the populations and the extent to which inferences about the social characteristics of ethnic groups are altered. The degree of overlap varies enormously among groups, but the results show that inferences based on nativity data are fairly close to those that would be obtained with the use of the ethnic ancestry item. This is an encouraging result for researchers obliged to use nativity data in working with earlier censuses.     

Alternatives and options in computer content analysis

First developed in the early 1960s, computer content analysis methods stimulated renewed interest in the quantitative analysis of verbal materials. Continued research and development of such methods produced a range of current alternatives and options. Investigators can choose between advanced systems varying widely in amount of theoretical prestructuring required and complexity of units of analysis, or they can devise less complex systems built upon basic work processing programs.     

Separating age,period, and cohort effects in white U.S. fertility, 1920–1970

Four models are developed to describe the odds transformation of period- and age-specific fertility rates as products of age, period, and cohort effects. These are applied to data for white U.S. women age 15–44 from 1920 to 1970, with equal weights given to each rate. All models which include age fit subsets of the data extremely well. Per effect, the incorporation of periods improves the fit much more than the incorporation of cohorts. It is shown that first differences are invariant in two-effect models, and second differences are invariant in the three-effect models.     

Determinants of fertility levels and change among developed countries: 1958–1978

This article provides a multivariate cross-national test of the hypothesis that national population/family planning policies have effected levels, and changes in fertility in developed nations over the past two decades. Variation is assessed in the total fertility rates (TFR) in 1978, and in the change in these rates between 1958 and 1978, among thirty developed countries. Measures include socioeconomic development, divorce, percent in consensual unions, female labor force participation, abortion policy, and level of contraceptive use by married couples and, government population/family planning policy. Seventy percent of the variation in 1978 TFR is related to the percent contracepting, female labor force, and the population/family planning policy measures. These are the only measures with significant direct effects. A longitudinal analysis of 1958 to 1978 change in TFRs is also conducted. This model increases R² to 75%, and the three independent variables remain significant. Implications of these findings for policy makers interested in increasing or decreasing fertility rates are noted.     

Some cross-population comparisons of family bias in the effects of schooling on occupational status

It is a truism of research on social stratification that the effects of socioeconomic or family background on educational attainment and adult success lead to biases in the simple regressions of occupational status (or other putative outcomes of schooling) on educational attainment. The present analysis compares findings of family bias in the effects of schooling on occupational status across samples of siblings drawn from the Wisconsin Longitudinal Study and from Olneck's sample of brothers from Kalamazoo, Michigan. The comparative analysis shows that family bias in the effect of schooling on occupational status may be much less than is commonly believed and that very large samples may be needed to measure it reliably. Moreover, the analysis suggests that estimates of family bias are very sensitive to the specification of response variability in schooling. The analysis also illustrates some useful methods for cross-population comparison of structural equation models.     

Chance in Guttman scaling

In their recent paper (1977, Social Science Research, 6, 188–196), Kenny and Rubin offer an interpretation of “chance” in Guttman scaling. Their approach is based on the matrix of probabilities p_ij, giving the chance that the ith respondent passes the jth item. It is shown that their interpretation of chance in fact corresponds to a stochastic model of a perfect unidimensional scale. An alternative condition for chance based on the matrix of p_ij's is proposed. This is then proved to be equivalent to the usual interpretation.     

Household migration and the labor market position of married women

The present paper conceptualizes labor force participation among U.S. wives as both a determinant and consequence of household migration. A theoretical justification is provided for decomposing these relationships by occupational position of the wife. Applying log-linear models to data from the 1970 One-in-a-Thousand Public Use Sample, we corroborate previous research in that, on the average, the likelihood of interstate migration between 1965 and 1970 was greatest among families in which the wife was not employed prior to the move. However, among employed wives, those classified as professionals or managers were also found to accentuate the probability of household migration, particularly if the husband was not employed in 1965. The implication is that all types of employment by the wife need not limit family mobility as previous studies have suggested. Although migration was found to have a disrupting effect on the continuity of employment among most wives, this effect was most in evidence among wives employed in blue collar occupations. The data suggest the need for a careful reappraisal of traditional economic notions of migration, particularly as they apply to married women.     

Population growth and the birth squeeze

Birth squeezes result from an imbalance in the number of males and females in the childbearing ages, and can have a significant impact on the level and distribution of births. To measure the effects of a birth squeeze, the observed male and female age-specific birth rates were distinguished from the underlying fertility magnitudes, which reflect the mutual propensity of males and females of specified ages to have a child. The fertility magnitudes, readily expressible in terms of the birth rates, are nonetheless independent of the age-sex composition of the population, while the birth rates are affected by compositional changes. The two-sex Total Fertility Rate, TFR2, the average of the male and female TFRs, is advanced as a summary measure of fertility independent of the birth squeeze, and a simple index, U, is defined to measure the severity of the squeeze. Analyses involving stable population models, alternative population projections, and the birth squeezes found in contemporary populations documented the importance of the birth squeeze for studies of fertility. In particular, the size of the distortions introduced by the birth squeeze was seen as related to the level of fertility itself. In high-fertility populations, the birth squeeze depresses observed female age-specific birth rates, and thus conventional measures can significantly underestimate the fertility reduction needed to end population growth.     

Comparing measures of homicide trends: Methodological and substantive differences in the Vital Statistics and Uniform Crime Report time series (1933–1975)

Several Uniform Crime Reports (FBI) and Vital Statistics (NCHS) homicide time series for the United States, 1933–1975, are compared with respect to (a) definitions and collection procedures used; (b) differences in estimating procedures; and (c) substantive differences that result from use of either series. Among the findings from this investigation are that: (1) the sampling procedures employed by the NCHS appear to be superior to those of the FBI; (2) the FBI time series most commonly employed by researchers is discontinuous prior to 1958 due to revision procedures which were instituted at that time. Analyses over time periods which began prior to this year should employ a more accurate FBI time series which is published by the Office of Management and Budget (1973). (3) The specific time period under study may lead to different substantive conclusions with different indicators; (4) the results of statistical analyses of structural equation models contained herein indicate that while the difference is not generally statistically significant, the deterrent effect of execution is consistently estimated to be greater with the Vital Statistics series. This finding has important policy implications if estimates of the number of lives saved by executing prisoners are projected and utilized by policy makers. (5) Neither series indicates serious problems of autocorrelation of disturbances with the exogenous variables employed here; and (6) use of either series for the period after 1949 appears to produce the same substantive results.