An omnibus lack of fit test in logistic regression with sparse data

The usefulness of logistic regression depends to a great extent on the correct specification of the relation between a binary response and characteristics of the unit on which the response is recoded. Currently used methods for testing for misspecification (lack of fit) of a proposed logistic regression model do not perform well when a data set contains almost as many distinct covariate vectors as experimental units, a condition referred to as sparsity. A new algorithm for grouping sparse data to create pseudo replicates and using them to test for lack of fit is developed. A simulation study illustrates settings in which the new test is superior to existing ones. Analysis of a dataset consisting of the ages of menarche of Warsaw girls is also used to compare the new and existing lack of fit tests.     

A framework for progressively improving small area population estimates

Summary.  The paper presents a framework for small area population estimation that enables users to select a method that is fit for the purpose. The adjustments to input data that are needed before use are outlined, with emphasis on developing consistent time series of inputs. We show how geographical harmonization of small areas, which is crucial to comparisons over time, can be achieved. For two study regions, the East of England and Yorkshire and the Humber, the differences in output and consequences of adopting different methods are illustrated. The paper concludes with a discussion of how data, on stream since 1998, might be included in future small area estimates.     

Interval estimation for log-linear models with one variable subject to non-ignorable non-response

Summary.  Log-linear models for multiway contingency tables where one variable is subject to non-ignorable non-response will often yield boundary solutions, with the probability of non-respondents being classified in some cells of the table estimated as 0. The paper considers the effect of this non-standard behaviour on two methods of interval estimation based on the distribution of the maximum likelihood estimator. The first method relies on the estimator being approximately normally distributed with variance equal to the inverse of the information matrix. It is shown that the information matrix is singular for boundary solutions, but intervals can be calculated after a simple transformation. For the second method, based on the bootstrap, asymptotic results suggest that the coverage properties may be poor for boundary solutions. Both methods are compared with profile likelihood intervals in a simulation study based on data from the British General Election Panel Study. The results of this study indicate that all three methods perform poorly for a parameter of the non-response model, whereas they all perform well for a parameter of the margin model, irrespective of whether or not there is a boundary solution.     

Maximum entropy estimation for survey sampling

Calibration methods have been widely studied in survey sampling over the last decades. Viewing calibration as an inverse problem, we extend the calibration technique by using a maximum entropy method. Finding the optimal weights is achieved by considering random weights and looking for a discrete distribution which maximizes an entropy under the calibration constraint. This method points a new frame for the computation of such estimates and the investigation of its statistical properties.     

The Agree Consultant:more agreeable software?

This paper first describes a program AGREE calculating many variants of coefficients for interobserver agreement. A pilot program MOCK was written, aimed at helping unitiated users of AGREE to select the most appropriate coeflcient, given the data type and the research goal. It is a mock-up version of the data entrance and analysis sections of AGREE, to which are added some menus and a knowledge-based CONSULTANT system questioning the user. Results of a small experiment with four variants of the CONSULTANT are presented. This leads to a discussion of desirable features for this kind of help programs and preprocessors for specialized statistical software.     

The behavioural wedge

When nations fail to agree, can individual citizens make a difference? The third of our post-Copenhagen features is by Jonathan Gilligan, Thomas Dietz, Gerald T. Gardner, Paul C. Stern , and Michael P. Vandenbergh . They look at the effects that voluntary actions by individuals can have, and at the policies that can best encourage such actions.     

The cross-product ratio in bivariate lognormal and gamma distributions,with an application to non-randomized trials

Non-randomized trials can give a biased impression of the effectiveness of any intervention. We consider trials in which incidence rates are compared in two areas over two periods. Typically, one area receives an intervention, whereas the other does not. We outline and illustrate a method to estimate the bias in such trials under two different bivariate models. The illustrations use data in which no particular intervention is operating. The purpose is to illustrate the size of the bias that could be observed purely due to regression towards the mean (RTM). The illustrations show that the bias can be appreciably different from zero, and even when centred on zero, the variance of the bias can be large. We conclude that the results of non-randomized trials should be treated with caution, as interventions which show small effects could be explained as artefacts of RTM.     

ON THE COVERAGE PROBABILITY OF CONFIDENCE INTERVALS IN REGRESSION AFTER VARIABLE SELECTION

This paper considers a linear regression model with regression parameter vector β. The parameter of interest is θ= a^Tβ where a is specified. When, as a first step, a data‐based variable selection (e.g. minimum Akaike information criterion) is used to select a model, it is common statistical practice to then carry out inference about θ, using the same data, based on the (false) assumption that the selected model had been provided a priori. The paper considers a confidence interval for θ with nominal coverage 1 ‐ α constructed on this (false) assumption, and calls this the naive 1 ‐ α confidence interval. The minimum coverage probability of this confidence interval can be calculated for simple variable selection procedures involving only a single variable. However, the kinds of variable selection procedures used in practice are typically much more complicated. For the real‐life data presented in this paper, there are 20 variables each of which is to be either included or not, leading to 2²⁰ different models. The coverage probability at any given value of the parameters provides an upper bound on the minimum coverage probability of the naive confidence interval. This paper derives a new Monte Carlo simulation estimator of the coverage probability, which uses conditioning for variance reduction. For these real‐life data, the gain in efficiency of this Monte Carlo simulation due to conditioning ranged from 2 to 6. The paper also presents a simple one‐dimensional search strategy for parameter values at which the coverage probability is relatively small. For these real‐life data, this search leads to parameter values for which the coverage probability of the naive 0.95 confidence interval is 0.79 for variable selection using the Akaike information criterion and 0.70 for variable selection using Bayes information criterion, showing that these confidence intervals are completely inadequate.     

Multiple imputation of missing data with ante-dependence covariance structure

A controlled clinical trial was conducted to investigate the efficacy effect of a chemical compound in the treatment of Premenstrual Dysphoric Disorder (PMDD). The data from the trial showed a non-monotone pattern of missing data and an ante-dependence covariance structure. A new analytical method for imputing the missing data with the ante-dependence covariance is proposed. The PMDD data are analysed by the non-imputation method and two imputation methods: the proposed method and the MCMC method.