Exact short poisson confidence intervals

The authors propose a new method for constructing a confidence interval for the expectation θ of a Poisson random variable. The interval they obtain cannot be shortened without the infimum over θ of the coverage probability falling below 1 ‐ α. In addition, the endpoints of the interval are strictly increasing functions of the observed variable. An easy‐to‐program algorithm is provided for computing this interval.     

GASOLINE PRICES AND ROAD FATALITIES: INTERNATIONAL EVIDENCE

This study utilizes data for 144 countries from 1991 to 2010 to present the first international estimates of the gasoline price elasticity of road fatalities. We instrument each country's gasoline price with that country's oil reserves and the yearly international crude oil price to address potential endogeneity concerns. Our findings suggest that the average reduction in road fatalities resulting from a 10% increase in the gasoline pump price is in the order of 3%–6%. Around 35,000 road deaths per year could be avoided by the removal of global fuel subsidies. (JEL R41, H23, O18, Q43)     

On cross‐validation of Bayesian models

The authors examine several aspects of cross‐validation for Bayesian models. In particular, they propose a computational scheme which does not require a separate posterior sample for each training sample.     

A Partial Likelihood Estimator of Vaccine Efficacy

A partial likelihood method is proposed for estimating vaccine efficacy for a general epidemic model. In contrast to the maximum likelihood estimator (MLE) which requires complete observation of the epidemic, the suggested method only requires information on the sequence in which individuals are infected and not the exact infection times. A simulation study shows that the method performs almost as well as the MLE. The method is applied to data on the infectious disease mumps.     

The Organizational Dynamics of the U.S. Environmental Movement: Legitimation,Resource Mobilization,and Political Opportunity*

Abstract We analyze the relationship between legitimation, resource mobilization, and political opportunity and the founding rate of national environmental organizations between 1895 and 1994. We address recent criticisms that organizational ecologists' reliance on the density dependence model—which treats legitimation as an unmeasured intervening variable—has failed to capture the active sociopolitical character of this process. We advocate a more historical approach to legitimation which relies on print media to construct direct measures of legitimation. Specifically, we use the yearly count of environmental books published as a measure of the legitimacy of environmentalism in the U.S. This count remains statistically significant and positively related to foundings even when other variables are added to the model. A negative second‐order effect of total books suggests that the recent proliferation of state and local environmental organizations is beginning to depress the founding rate of national organizations. Variations in resource mobilization and political opportunity also have affected foundings, though variables measuring the latter were not significant in the full model.     

Choice of Social Work as a Career: Possible Influences

Abstract

Little is known about why students choose social work for a career. Certain life events in the family of origin may exert a compelling influence, though other factors are certainly involved. This study examines the influence of selected life events and such factors as prior employment, volunteering, and influential persons via a survey of 415 social work students and a comparison group of 203 nonsocial work students. Social work students were more likely to report problems such as alcoholism and emotional illness within their families of origin and to attribute these experiences to their choice of career.     

The asymptotic covariance structure of log-linear model estimated parameters for the multiple recapture census

This paper presents a matrix formulation for log-linear model analysis of the incomplete contingency table which arises from multiple recapture census data. Explicit matrix product expressions are given for the asymptotic covariance structure of the maximum likelihood estimators of both the log-linear model parameter vector and the predicted value vector for the observed and missing cells. These results are illustrated for data pertaining to a population of children possessing a common congenital anomaly.     

Valuing Informal Care Experience: Does Choice of Measure Matter?

Well-being equations are often estimated to generate monetary values for non-marketed activities. In such studies, utility is often approximated by either life satisfaction or General Health Questionnaire scores. We estimate and compare monetary valuations of informal care for the first time in the UK employing both measures, using longitudinal data on well-being and informal care provision. The choice of well-being measure has some effect on the estimated parameters and resulting monetary valuations, but any differences are not statistically significant. Further research is needed to confirm the comparability of these measures if researchers are to continue to use them interchangeably.