A Bayesian mixture of experts approach to covariate misclassification

This article considers misclassification of categorical covariates in the context of regression analysis; if unaccounted for, such errors usually result in mis-estimation of model parameters. With the presence of additional covariates, we exploit the fact that explicitly modelling non-differential misclassification with respect to the response leads to a mixture regression representation. Under the framework of mixture of experts, we enable the reclassification probabilities to vary with other covariates, a situation commonly caused by misclassification that is differential on certain covariates and/or by dependence between the misclassified and additional covariates. Using Bayesian inference, the mixture approach combines learning from data with external information on the magnitude of errors when it is available. In addition to proving the theoretical identifiability of the mixture of experts approach, we study the amount of efficiency loss resulting from covariate misclassification and the usefulness of external information in mitigating such loss. The method is applied to adjust for misclassification on self-reported cocaine use in the Longitudinal Studies of HIV-Associated Lung Infections and Complications.     

Rebates, inventories, and intertemporal price discrimination

We demonstrate that universally redeemed rebates can increase manufacturer profits by reducing the incentives of downstream retailers to hoard inventories when optimal wholesale prices vary predictably over time. By bypassing retailers and making direct contracts with buyers, the manufacturer can increase the variations in effective prices paid by consumers without concomitantly creating larger incentives for retailers to hold inventories. During profitable, high-demand periods, manufacturer revenues are ordinarily constrained by'competition'from retailer inventories, thus limiting profits. However, by selectively offering rebates to consumers while maintaining high wholesale prices, low-demand periods can be accommodated without inducing retailer hoarding.     

New perspective on the benefits of the gene–environment independence in case–control studies

We study the benefit of exploiting the gene–environment independence (GEI) assumption for inferring the joint effect of genotype and environmental exposure on disease risk in a case–control study. By transforming the problem into a constrained maximum likelihood estimation problem we derive the asymptotic distribution of the maximum likelihood estimator (MLE) under the GEI assumption (MLE‐GEI) in a closed form. Our approach uncovers a transparent explanation of the efficiency gained by exploiting the GEI assumption in more general settings, thus bridging an important gap in the existing literature. Moreover, we propose an easy‐to‐implement numerical algorithm for estimating the model parameters in practice. Finally, we conduct simulation studies to compare the proposed method with the traditional prospective logistic regression method and the case‐only estimator. The Canadian Journal of Statistics 47: 473–486; 2019 © 2019 Statistical Society of Canada     

Automated statistical methods for measuring the strength of block ciphers

A block cipher is one of the most common forms of algorithms used for data encryption. This paper describes an efficient set of statistical methods for analysing the security of these algorithms under the black-box approach. The procedures can be fully automated, which provides the designer or user of a block cipher with a useful set of tools for security analysis.     

Western voodoo: providing mental health care to Haitian refugees

This article described certain aspects of Haitian life, voodoo and its role in Haitian society, the quality and quantity of psychiatric and mental health care for Haitians in Haiti, and suggestions for providing appropriate mental health care to Haitian refugees in the United States. Conway and Buchanan (1985) described what has helped Haitian refugees adapt in the transition to life in the United States: the strengths from their cultural heritage, such as fortitude; perseverance in the most arduous circumstances; deep religious faith; high self-respect; reliance on the extended family; and the tradition of sharing. Building on these assets may assist Western mental health-care providers in offering culturally sensitive mental health care to Haitians.     

Educational efforts in the implementation of rural family planning programs in East Pakistan

The pilot family planning studies reported in this paper were conducted in a rural area adjacent to the city of Dacca in East Pakistan. It reports the preliminary findings of action-research in the implementation of educational efforts to reach rural villagers of a developing country.Preliminary analyses of the records identify two significant educational problems: (1) most of the villagers (85-90 percent of the couples) who initially accepted contraceptives do not truly adopt and become continuing users and (2) the continuing users (10-15 percent of the initial users) are generally characterized by large families. (Later data show an even lower percentage of continuing users.)The field activities in the development of various educational approaches to family planning are described. Three separate geographic areas (from 15,000 to 20,000 population) were each approached in a different way, varying in the number and educational qualifications of the workers and in the degree of involvement of village leadership. Preliminary analyses of field records indicate that these variations of approach apparently have little effect on the percentage of the population willing to accept contraceptive supplies.It is the opinion of the writers that more intensive educational efforts are necessary at the village level to develop social support for continuing use of contraception and to gain adoption of contraception by younger married couples primarily for spacing of births.The impact of introduction of the IUCD in populations where condoms and foam tablets have been available for one to two years has also been reported. Preliminary findings indicate that the IUCD encourages adoption by previous non-users and may increase the over-all percentage of contraceptive users.     

CharacteriStics and utilization of midwives in a selected rural area of East Pakistan

The purpose of this study was to obtain information about the utilization of midwives (dais) by village women and to learn more about the characteristics and practices of those dais. Two interview schedules were prepared. The first was used to interview 632 village women to determine who performed or assisted with their last two deliveries. The second was used to obtain information from 21 dais.The results of the first phase may be summarized as follows: (a) 38 per cent of the women were delivered by relatives; (b) 33 per cent of the women delivered their own children; (c) 14 per cent were delivered by neighbors; (d) 6 per cent were delivered by dais; (e) 2 percent were delivered in hospitals; and (f) the remaining 7 per cent were accounted for by several minor categories.The results of the interviews with dais are summarized as follows: (a) they are mainly widows and older women; (b) they have no formal training; (c) they work for friends, neighbors, and relatives and receive a sari as compensation; (d) they cannot handle complicated deliveries; (e) they deliver 3-4 children a year; (f) their sterilizing procedures depend upon soap, water, and folk beliefs; (g) most think midwifery is a worthwhile service; (h) about one-half have a general understanding of the reproduction process; (i) most do not know how to prevent conception; and (j) about half think that it is a good idea to participate actively in a family planning program.     

Comparing child protective investigation performance between law enforcement agencies and child welfare agencies

This study examines the comparative effectiveness of using law enforcement agencies for child protective investigation (CPI), in contrast with the traditional approach of CPI conducted by the public child welfare agency. The analysis uses 2006-2007 data from a natural experiment conducted in Florida to show modest differences in performance and cost-efficiency between the two approaches to CPI. These findings may have implications for other states considering outsourcing CPI to law enforcement.     

Methods for testing subblock patterns

A number of statistical tests have been recommended over the last twenty years for assessing the randomness of long binary strings used in cryptographic algorithms. Several of these tests include methods of examining subblock patterns. These tests are the uniformity test, the universal test and the repetition test. The effectiveness of these tests are compared based on the subblock length, the limitations on data requirements, and on their power in detecting deviations from randomness. Due to the complexity of the test statistics, the power functions are estimated by simulation methods. The results show that for small subblocks the uniformity test is more powerful than the universal test, and that there is some doubt about the parameters of the hypothesised distribution for the universal test statistic. For larger subblocks the results show that the repetition test is the most effective test, since it requires far less data than either of the other two tests and is an efficient test in detecting deviations from randomness in binary strings.     

An extension of the Dirichlet prior for the analysis of longitudinal multinomial data

Studies producing longitudinal multinomial data arise in several subject areas. This article suggests a Bayesian approach to the analysis of such data. Rather than infusing a latent model structure, we develop a prior distribution for the multinomial parameters which reflects the longitudinal nature of the observations. This distribution is constructed by modifying the prior that posits independent Dirichlet distributions for the multinomial parameters across time. Posterior analysis, which is implemented using Monte Carlo methods, can then be used to assess the temporal behaviour of the multinomial parameters underlying the observed data. We test this methodology on simulated data, opinion polling data, and data from a study concerning the development of moral reasoning.