James B. Kirby

Improper solutions in structural equation models: Causes, consequences, and strategies

In this article, the authors examine the most common type of improper solutions: zero or negative error variances. They address the causes of, consequences of, and strategies to handle these issues. Several hypotheses are evaluated using Monte Carlo simulation models, including two structural equation models with several misspecifications of each model. Results suggested several unique findings. First, increasing numbers of omitted paths in the measurement model were associated with decreasing numbers of improper solutions. Second, bias in the parameter estimates was higher in samples with improper solutions than in samples including only proper solutions. Third, investigations of the consequences of using constrained estimates in the presence of improper solutions indicated that inequality constraints helped some samples achieve convergence. Finally, the use of confidence intervals as well as four other proposed tests yielded similar results when testing whether the error variance was greater than or equal to zero.

Explaining Racial and Ethnic Disparities in Health Care

Objectives:The substantial racial and ethnic disparities in access to and use of health services are well documented. A number of studies highlight factors such as health insurance coverage and socioeconomic differences that explain some of the differences between groups, but much remains unexplained. We build on this previous research by incorporating additional factors such as attitudes about health care and neighborhood characteristics, as well as separately analyzing different Hispanic subgroups. Methods:We use the Oaxaca-Blinder regression-based method to decompose differences among racial and ethnic groups in 3 measures related to access, quantifying the portion explained by each of a number of underlying characteristics and the differences that remain unexplained. We use data from the 2000 and 2001 Medical Expenditure Panel Survey (MEPS), a nationally representative survey of the noninstitutionalized U.S. population. We link these data to detailed neighborhood characteristics from the Census Bureau and local provider supply data from the Health Services Resource Administration (HRSA). Results:Consistent with earlier studies, we find insurance status and socioeconomic differences explain a significant part of the disparities. Additionally, neighborhood racial and ethnic composition account for a large portion of disparities in access, and language differences help explain observed disparities in the use-based access measure. However, much of the differences between racial and ethnic groups remain unexplained. We also found substantial variation in the level of disparities among different groups of Hispanics. Conclusions:Researchers and policymakers may need to broaden the scope of factors they consider as barriers to access if the goal of eliminating disparities in health care is to be achieved.

Finite Sampling Properties of the Point Estimates and Confidence Intervals of the RMSEA

A key advantage of the root mean square error of approximation (RMSEA) is that under certain assumptions, the sample estimate has a known sampling distribution that allows for the computation of confidence intervals. However, little is known about the finite sampling behaviors of this measure under violations of these ideal asymptotic conditions. This information is critical for developing optimal criteria for using the RMSEA to evaluate model fit in practice. Using data generated from a computer simulation study, the authors empirically tested a set of theoretically generated research hypotheses about the sampling characteristics of the RMSEA under conditions commonly encountered in applied social science research. The results suggest that both the sample estimates and confidence intervals are accurate for sample sizes of n = 200 and higher, but caution is warranted in the use of these measures at smaller sample sizes, at least for the types of models considered here.

The noncentral chi-square distribution in misspecified structural equation models : Finite sample results from a Monte Carlo simulation

The noncentral chi-square distribution plays a key role in structural equation modeling (SEM). The likelihood ratio test statistic that accompanies virtually all SEMs asymptotically follows a noncentral chi-square under certain assumptions relating to misspecification and multivariate distribution. Many scholars use the noncentral chi-square distribution in the construction of fit indices, such as Steiger and Linds (1980) Root Mean Square Error of Approximation (RMSEA) or the family of baseline fit indices (e.g., RNI, CFI), and for the computation of statistical power for model hypothesis testing. Despite this wide use, surprisingly little is known about the extent to which the test statistic follows a noncentral chi-square in applied research. Our study examines several hypotheses about the suitability of the noncentral chi-square distribution for the usual SEM test statistic under conditions commonly encountered in practice. We designed Monte Carlo computer simulation experiments to empirically test these research hypotheses. Our experimental conditions included seven sample sizes ranging from 50 to 1000, and three distinct model types, each with five specifications ranging from a correct model to the severely misspecified uncorrelated baseline model. In general, we found that for models with small to moderate misspecification, the noncentral chi-square distribution is well approximated when the sample size is large (e.g., greater than 200), but there was evidence of bias in both mean and variance in smaller samples. A key finding was that the test statistics for the uncorrelated variable baseline model did not follow the noncentral chi-square distribution for any model type across any sample size. We discuss the implications of our findings for the SEM fit indices and power estimation procedures that are based on the noncentral chi-square distribution as well as potential directions for future research.

Latent Variable Models Under Misspecification: Two-Stage Least Squares (2SLS) and Maximum Likelihood (ML) Estimators

This article compares maximum likelihood (ML) estimation to three variants of two-stage least squares (2SLS) estimation in structural equation models. The authors use models that are both correctly and incorrectly specified. Simulated data are used to assess bias, efficiency, and accuracy of hypothesis tests. Generally, 2SLS with reduced sets of instrumental variables performs similarly to ML when models are correctly specified. Under correct specification, both estimators have little bias except at the smallest sample sizes and are approximately equally efficient. As predicted, when models are incorrectly specified, 2SLS generally performs better, with less bias and more accurate hypothesis tests. Unless a researcher has tremendous confidence in the correctness of his or her model, these results suggest that a 2SLS estimator should be considered.

The influence of parental separation on smoking initiation in adolescents

Most adult smokers start smoking when they are adolescents and, the prevalence of smoking declines less than other unhealthy behaviors as people mature. Understanding why adolescents start smoking is, therefore, key to developing effective policy aimed at lowering the prevalence of smoking in both children and adults. In this study, I suggest that parental separation is one possible risk factor for smoking initiation. I use a nationally representative sample of American adolescents interviewed at two points in time to examine the influence of parental separation on smoking initiation. Two questions are addressed. First, is there a relationship between parental separation and the likelihood that an adolescent will initiate smoking? Second, if there is a relationship, through what factors does parental separation operate to influence the initiation of smoking in adolescents? My findings suggest that parental separation increases the likelihood that adolescents will start smoking. It does so in part by raising depressive symptoms and rebelliousness in adolescents. Despite the significance of these indirect effects, however, the bulk of the effect of parental separation on smoking initiation is direct.

Unhealthy and Uninsured: Exploring Racial Differences in Health and Health Insurance Coverage Using a Life Table Approach

Millions of people in the United States do not have health insurance, and wide racial and ethnic disparities exist in coverage. Current research provides a limited description of this problem, focusing on the number or proportion of individuals without insurance at a single time point or for a short period. Moreover, the literature provides no sense of the joint risk of being uninsured and in need of medical care. In this article, we use a life table approach to calculate health- and insurance-specific life expectancies for whites and blacks, thereby providing estimates of the duration of exposure to different insurance and health states over a typical lifetime. We find that, on average, Americans can expect to spend well over a decade without health insurance during a typical lifetime and that 40% of these years are spent in less-healthy categories. Findings also reveal a significant racial gap: despite their shorter overall life expectancy, blacks have a longer uninsured life expectancy than whites, and this racial gap consists entirely of less-healthy years. Racial disparities in insurance coverage are thus likely more severe than indicated by previous research.

Access to Health Care: Does Neighborhood Residential Instability Matter?

Many Americans do not have access to adequate medical care. Previous research on this problem focuses primarily on individual-level determinants of access such as income and insurance coverage. The role of community-level factors in helping or hindering individuals in obtaining needed medical care, however, has not received much attention. We address this gap in the literature by investigating the association between neighborhood residential instability and access to health care. Using individual-level data from the 2000 Medical Expenditure Panel Survey and block-group level data from the 2000 decennial census, we find that individuals who live in neighborhoods with high residential turnover have worse health care access than residents of other neighborhoods. This association persists even when the prevalence of poverty, the supply of health care, and a variety of individual characteristics are held constant. We offer explanations for these findings and suggest directions for future research.

Explaining racial and ethnic differences in children's use of stimulant medications.

Objectives:To document and explain racial/ethnic differences in the use of stimulant drugs among US children. Data and Methods:We use a nationally representative sample of children ages 5–17 years old from the Medical Expenditure Panel Survey (MEPS) for the years 2000–2002. We estimate race-specific means and regressions to highlight differences across groups in individual/family characteristics that may affect stimulant use and differences in responses to these characteristics. Then, we use Oaxaca-Blinder decomposition methods to quantify the portion of differential use explained by differences in individual/family characteristics. Finally, we use pooled regressions with race/ethnicity interactions to formally test the hypothesis that responses to perceived mental health and behavioral problems vary across groups. Results:White children are about twice as likely to use stimulants as either Hispanic or Black children. Differences in individual/family characteristics account for about 25% of the difference between whites and Hispanics, but for none of the difference between whites and blacks. Pooled regressions show that racial/ethnic gaps in stimulant use persist among children with otherwise similar reported mental health conditions. Conclusions:Our finding that the majority of racial/ethnic differences in childrens stimulant use is explained by differences in responses to individual/family characteristics highlights the importance of further research to examine the reasons for these differences. It is striking that children with otherwise similar reports of mental health problems have such different outcomes in terms of stimulant use. Potential explanations range from discrimination to cultural differences by race/ethnicity or community.

Access to health care for nonmetro and metro latinos of Mexican origin in the United States

Background:A growing number of Latinos are moving to nonmetro areas, but little research has examined how this trend might affect the Latino-disadvantage in access to healthcare. Objective:We investigate health care access disparities between non-Latino whites and Latinos of Mexican origin, and whether the disparities differ between metro and nonmetro areas. Methods:A series of logistic regression models provide insight on whether individuals have a usual source of care and whether they have had any physician visits in the past year. Our analyses focus on the interaction between Mexican origin descent and nonmetro residence. Subjects:Nationally representative data from the 2002–2003 Medical Expenditure Panel Survey are analyzed. The sample consists of working-aged adults age 18–64, yielding a sample size of 29,875. Results:The Mexican disadvantage in having a usual source of care is much greater among nonmetro residents than among those living in metro areas. The Mexican disadvantage in the likelihood of seeing a physician at least 1 time during the year does not differ across locations. Although general and ethnicity-specific predictors explain the disadvantage of Mexicans in having a usual source of care, they do not explain the added disadvantage of being Mexican and living in nonmetro areas. Conclusions:This study identifies a new challenge to the goal of eliminating health care disparities in the United States. The Latino population living in nonmetro areas is growing, and our findings suggest that Latinos in nonmetro areas face barriers to having a usual source of care that are greater than those faced by Latinos in other areas.