Sharon L. Lohr

Design and methods of the Population Assessment of Tobacco and Health (PATH) Study

Background This paper describes the methods and conceptual framework for Wave 1 of the Population Assessment of Tobacco and Health (PATH) Study data collection. The National Institutes of Health, through the National Institute on Drug Abuse, is partnering with the Food and Drug Administrations (FDA) Center for Tobacco Products to conduct the PATH Study under a contract with Westat. Methods The PATH Study is a nationally representative, longitudinal cohort study of 45 971 adults and youth in the USA, aged 12 years and older. Wave 1 was conducted from 12 September 2013 to 15 December 2014 using Audio Computer-Assisted Self-Interviewing to collect information on tobacco-use patterns, risk perceptions and attitudes towards current and newly emerging tobacco products, tobacco initiation, cessation, relapse behaviours and health outcomes. The PATH Studys design allows for the longitudinal assessment of patterns of use of a spectrum of tobacco products, including initiation, cessation, relapse and transitions between products, as well as factors associated with use patterns. Additionally, the PATH Study collects biospecimens from consenting adults aged 18 years and older and measures biomarkers of exposure and potential harm related to tobacco use. Conclusions The cumulative, population-based data generated over time by the PATH Study will contribute to the evidence base to inform FDAs regulatory mission under the Family Smoking Prevention and Tobacco Control Act and efforts to reduce the Nations burden of tobacco-related death and disease.

A comparison of weighted and unweighted analyses in the national Crime Victimization Survey

The National Crime Victimization Survey (NCVS) data tapes include several variables of weights that reflect how many households or persons are represented by a given data record. While these sampling weights are crucial for estimating overall victimization rates, they do not have as much of an effect on methodological models, because weights in the NCVS are used primarily in ratio estimation adjustments and to compensate for a relatively low nonresponse rate. The general use of weights is discussed, and several examples are given that indicate that although there may be some differences in the coefficients, the basic conclusions drawn from the models are the same with or without weights. It is recommended that weighted analyses in the NCVS be used primarily as a tool in model development.

A longitudinal analysis of factors associated with reporting violent crimes to the police

The use of generalized estimating equations and time-series methods for fitting longitudinal models in the National Crime Victimization Survey (NCVS) is discussed, with reference to the relation between the reporting of a violent crime to the police and previous victimizations. Two longitudinal models are fit to NCVS data to predict the likelihood of reporting a violent crime to the police based on characteristics of the victim and the incident and based on previous victimization experiences. In both models, it is found that higher reporting rates are associated with positive results accruing from reporting previous victimization to the police.

Inference from Dual Frame Surveys

Abstract In a dual frame survey, samples are drawn independently from two overlapping frames that together cover the population of interest. Several estimators for population totals in dual frame surveys are discussed and compared under a unified setup. We propose jackknife variance estimators, establish their design consistency, and explore their finite-sample performance in a simulation study.

Dissolving the relationship between divorce laws and divorce rates

Law and other causes influence the divorce rate. Thus much is axiomatic. But when an attempt is made to go further and determine the relative influence and effect of law and the sum of other causes, then the controversy opens.1

Estimation in Multiple-Frame Surveys

Multiple-frame surveys are commonly used to decrease costs of sampling or to reduce undercoverage that could occur if only one sampling frame were used. We describe potential uses and examples of multiple-frame surveys. We then derive optimal linear estimators and pseudo–maximum likelihood estimators for the population total when samples are taken independently from each frame using probability sampling designs. We explore the properties of these estimators theoretically and through a simulation study. We also derive variance estimators and discuss some practical problems that may be encountered in multiple-frame surveys.

Estimates of Repeat Victimization Using the National Crime Victimization Survey

Many crime victims experience multiple victimizations over time. Estimating the rate of repeat victimization from a longitudinal survey, however, is difficult because individuals often have missing data for some of the interviews. We use data from the 1996–98 U.S. National Crime Victimization Survey to explore potential effects of missing data on estimated rates of repeat violent victimizations in individuals. We introduce two algorithms for estimating repeat victimization rates, using logistic models to impute values for individuals who have partial data. These models are applied to estimate rates of repeat victimization for all violent crimes, and separately for crimes of domestic violence. Estimates show substantial sensitivity to the form of the model used.

Small area estimation with auxiliary survey data

Large governmental surveys typically provide accurate national statistics. To decrease the mean squared error of estimates for small areas, i.e., domains in which the sample size is small, auxiliary variables from administrative records are often used as covariates in a mixed linear model. It is generally assumed that the auxiliary information is available for every small area. In many cases, though, such information is available for only some of the small areas, either from another survey or from a previous administration of the same survey. The authors propose and study small area estimators that use multivariate models to combine information from several surveys. They discuss computational algorithms, and a simulation study indicates that if quantities in the different surveys are sufficiently correlated, substantial gains in efficiency can be achieved.

Interpreting Crime Statistics.

Preface List of abbreviations Notes on contributors Introduction 1. Statistics of offences 2. Criminal justice and offenders 3. The Police 4. Community sentences and the work of the probation service 5. Prison 6. Parole 7. Homicide 8. Gender, crime and the criminal justice system 9. Ethnic differences 10. On using crime statistics for prediction 11. Survey design and interpretation of the British Crime Survey

Hasse Diagrams in Statistical Consulting and Teaching

Abstract Hasse diagrams summarize the structure of mixed models and can be used by a statistical consultant to help design a complicated experiment or to help clarify the structure of data to be analyzed. They are also useful in the classroom as an aid for obtaining expected mean squares or deciding which denominator should be used in an F statistic.