Which Wald statistic? Choosing a parameterization of the Wald statistic to maximize power in k-sample generalized estimating equations

The Wald statistic is known to vary under reparameterization. This raises the question: which parameterization should be chosen, in order to optimize power of the Wald statistic? We specifically consider k-sample tests of generalized linear models (GLMs) and generalized estimating equations (GEEs) in which the alternative hypothesis contains only two parameters. An example is presented in which such an alternative hypothesis is of interest. Amongst a general class of parameterizations, we find the parameterization that maximizes power via analysis of the non-centrality parameter, and show how the effect on power of reparameterization depends on sampling design and the differences in variance across samples. There is no single parameterization with optimal power across all alternatives. The Wald statistic commonly used under the canonical parameterization is optimal in some instances but it performs very poorly in others. We demonstrate results by example and by simulation, and describe their implications for likelihood ratio statistics and score statistics. We conclude that due to poor power properties, the routine use of score statistics and Wald statistics under the canonical parameterization for GEEs is a questionable practice.     

FAMILY THERAPISTS' RESPONSE TO AIDS: AN EXAMINATION OF ATTITUDES,KNOWLEDGE, AND CONTACT*

As the number of individuals and families impacted by AIDS continue to multiply, family therapists will increasingly be asked to become engaged in the challenge of caring for those affected. To date, little has been written in family therpy journals regarding the response of family therapists to this crisis. This paper represents an initial attempt to examine the atitudes of clinical members of the American Association for Marriage and Family Therapy toward AIDS and persons with AIDS. Included in the study are data regarding the therapists' contact with persons with AIDS, gays, and lesbians, and the impact of such contact on attitudes. Implications for future research, training, and treatment of those affected by AIDS are also provided.     

Physician-nurse migration to the United States: regional and health status origins in relation to legislation and policy

This article examines the origins of physicians and nurses who were admitted as permanent immigrants to the US from 1962-1979. Data are mainly from the Immigration and Naturalization Service. Countries used in the developmental analysis are only those whose population was estimated at 1 million or more as of mid-1979, encompassing 99% of the physicians and 97% of the nurses. Life expectancy at birth is the criterion used to differentiate origin countries by developmental dimension of health status. During the study period, health workers constituted about 30% of immigrants admitted to the US; of these, nurses and physicians constituted 72-82% throughout the study period. The period 1962-1979 has 4 distinct phases, marked by important legislative and/or policy changes; 1) 1962-1965, when the McCarran-Walter Act prevailed; 2) ending in 1968, the 2nd phase covers the transition mandated under the 1965 Immigration Act, which encouraged physician immigration; 3) the 3rd phase, 1969-1976, covers the transition to the 1976 Immigration and Nationality Act amendments; and 4) the 4th stage is 1977-1979. Results show that 1) under the McCarran-Walter Act, North America became the dominant physician source; 2) from 1966-1968, Asia attained dominance as the physician source and became even more predominant after 1968; 3) North America produced relatively few physicians in the early 1970s; 4) Europe produced substantially fewer physicians in the 1970s than in the 1960s; 5) South America, Africa, and Oceania were the lowest contributors of physicians; 6) during the McCarran-Walter years, North America and Europe produced almost 90% of nurses admitted into the US; 7) the 1965 Immigration Act and its aftermath resulted in Asia becoming the dominant source of nurses; 8) prior to the 1965 Immigration Act, Canada generated 20% of the aggregate number of physicians; 9) the Philippines surpassed Canada during the transition and India led after the transition; and 10) Canada supplied 30% of the nurses up through the transition, with the Philippines in the lead 1969-1979. Low health status countries were a relatively minor nurse source. Health status at the origin was a far less significant determinant of physician immigration than that of nurses. English language high and low health status country groups produced substantially more physician and nurse immigrants that their corresponding non-English language counterparts. The US attracted more physicians and nurses from less developed countries than more developed countries after 1968.     

Child health,breast-feeding,and survival in Malaysia: a random-effects logit approach

This study examines the possibility that estimation of the effect of breast-feeding on infant survival is affected by selection bias, in that children who are healthier at birth may be more likely to be breast-fed. Data are from the 1976 Malaysian Family Life Survey. "Ordinary logit models for breast-feeding and survival are estimated, and the results suggest that selection is indeed present. For example, children of higher birth weight appear to be more likely to be breast-fed and likely to survive. In addition, weight at birth and the duration of breast-feeding appear to be linked." Using birth weight as an indicator for the child's health, the authors conclude that "the direct influence of breast-feeding on survival remains of overwhelming importance even after corrections for selection bias are made."     

General characteristics of the electronic data processing system for the national census of 1988

The author describes in detail the procedures used in transferring population census data from the individual questionnaires to machine-readable form and the interim controls employed in the Central Statistical Agency of Poland and its regional offices.     

The influence of income on family fertility in the light of the results of empirical studies

Based on data from the 1973-83 Polish Family Budget Survey, this study examines the influence of income as a source of maintenance and income as a determinant of childbearing on family parity. It is assumed that income influences parity as a source of maintenance and as an influence on lifestyle or reproductive behavior. As a source of maintenance, the influence will be positive. As a life-style factor, it could be positive or negative. The hypotheses are 1) that positive influences are characteristics of older couples with a lower educational level and living in villages and small towns and income is a source of maintenance, and 2) a negative influence on parity reflects younger couples with higher educational levels from larger cities and income is a life-style factor. Some examples of how the theory of income influences total fertility are given in addition to a further elaboration of the theory. Other concepts introduced, as related to childbearing, are the value of goods and services with unchanged preferences and an increase in income, with preferences changing with income increases, with income preferences for nonprocreational and for procreational needs, and income as a factor influencing procreational needs and behavior. Correlation analysis is conducted for 33 subpopulations based on duration of marriage, educational level of the household, and size of residence. The results indicate that, in years 1973, 1975, and 1978 when income and parity are significant, the sign is positive. In 1980, there was a decrease in the number of subpopulations with a significant relationship, and in 3 subpopulations the results were negative. The 13 positive subpopulations were similar to those in the 1970s. In 1983, income explained an even smaller amount of parity. The subpopulations with a postive relationship usually have longer marriage durations, which suggest income as a source of maintenance. Negative results were among subpopulations with shorter marriage durations, which suggests life-style factor effects of income.     

A research note on Keyfitz's ‘The demographics of unfunded pensions’

Keyfitz (1985), The demographics of unfunded pensions, discusses the performance of pay-as-you-go old age insurance plans under different financial structures. One arrangement is to pay a fixed pension to retirees and collect the necessary contributions from the working population. Another is to fix the contribution per worker and distribute whatever monies are collected to the retirees. Unfortunately, equity problems arise in a pay-as-you-go plan when the population is not stable. Generally, large cohorts will receive greater rates of return than small cohorts when the pension is fixed, while small cohorts will receive greater rates of return when the contribution is fixed. Social security analysts in the U.S. have been concerned with minimizing inequities in cohort rates of return. In considering the US population in 1980, Keyfitz shows that disparities in cohort rates of return are less under a fixed contribution scheme than under a fixed pension scheme. This research note points out that Keyfitz's finding is limited to the specific situation in 1980. When analyzing in a broader framework the mechanics of unfunded pensions and their interaction with nonstable populations, the reverse is true: fixing the pension yields less disparity in cohort rates of return than fixing the contribution.Keyfitz (1985) a déjà discuté dans cette revue des résultats d'un système de répartition pour le financement des pensions de retraite sous diverses conditions financières. Une solution consiste à payer une pension fixée aux retraités et à collecter les fonds nécessaires auprès de la population active. Une autre solution est de fixer la contribution par actif et de distribuer ce qui a été ainsi collecté aux retraités. Malheureusement des problèmes d'équité surviennent dans ces systèmes de répartition lorsque la population n'est pas stable. Généralement, les cohortes nombreuses bénéficieront de plus forts taux de rendement que des cohortes moins nombreuses lorsque la pension est fixée, tandis que l'inverse se produit lorsque la contribution par actif est fixée. Aux Etats-Unis les analystes des systèmes de sécurité sociale ont cherché à minimiser les inégalités dans les taux de rendement. Lorsque Keyfitz considérait la population des Etats-Unis en 1980, il avait montré que les disparités dans les taux de rendement des cohortes étaient moindres sous la condition d'une contribution par actif fixée que sous celle d'une pension par retraité fixée. Cette note the recherche indique que le résultat obtenu par Keyfitz est propre à la situation spécifique de l'année 1980. Lorsque l'on analyse d'un point de vue général le fonctionnement des pensions de retraite et leurs interactions avec des populations non-stables, l'inverse est vérifié : le fait de fixer la pension conduit à moins de disparités dans les taux de rendement des cohortes que le fait de fixer la contribution des actifs.

---

This research was undertaken when the author was completing her dissertation at the Graduate Group in Demography at the University of California, Berkeley. The author now operates Lapkoff Demographic Research, a private research consulting firm.     

Audience effects on solitary faces during imagery: Displaying to the people in your head

Subjects imagined situations in which they reported enjoying themselves either alone or with others. Electromyographic (EMG) activity was recorded bilaterally from regions overlying thezygomatic major muscles responsible for smiling. Controlling for equal rated happiness in the two conditions, subjects showed more smiling in high-sociality than low-sociality imagery. In confirming imaginary audience effects during imagery, these data corroborate hypotheses that solitary facial displays are mediated by the presence of imaginary interactants, and suggest caution in employing them as measures of felt emotion.Avery Gilbert and Amy Jaffey had compelling insights throughout the course of study. We thank Paul Ekman, Carroll Izard, and Paul Rozin for extensive comments on earlier drafts. We also thank Bernard Apfelbaum, Jon Baron, Janet Bavelas, John Cacioppo, Linda Camras, Dean Delis, Rob DeRubeis, Alan Fiske, Stephen Fowler, Greg McHugo, Harriet Oster, David Premack, W. John Smith, and David Williams for their valuable comments and suggestions.     

Mortality among three Puerto Rican populations: residents of Puerto Rico and migrants in New York City and in the balance of the United States, 1979-81

"This study attempts to explain similarities and differences in the mortality experience of three population groups: Puerto Ricans on the island commonwealth, Puerto Rican born persons in New York City and Puerto Rican born persons in the rest of mainland United States. Mortality is much higher among Puerto Ricans in New York City than among those residing elsewhere. Much of the difference is due to excess mortality caused by cirrhosis of the liver and homicide. Puerto Rican born persons living on the mainland but outside New York City generally have low mortality, even when compared with U.S. whites."