"SECRET BALLOTS" AND SELF-REPORTS IN AN EXIT-POLL EXPERIMENT

A controlled exit-poll experiment on Election Day (November3, 1992) shows that refusals to answer questions and other evasiveforms of responding were significantly lower when respondentswere given a self-administered "secret-ballot" questionnairethan when they were interviewed face-to-face. While there weresome suggestive interactions of this mode-of-data collectioneffect with age and sex, and with the partisan atmosphere ofthe precinct in which the interviews were conducted, they wereborderline in significance, inconsistent in pattern, or difficuitto interpret. More important, comparisons with official electionreturns (the truth) indicated that the secret-ballot techniquewas more accurate than were face-to-face interviews in estimatingthe final outcome on the most socially sensitive of three self-reportedvotes: a vote for or against a tax levy for elderly services.Using a genuine secret ballot in the tradition of the olderGallup preelection polls can thus increase the validity of self-reportsin exit polls today.     

Batch sampling to improve power in a community trial. Experience from the Pawtucket Heart Health Program

Experiments involving large social units, such as schools, work sites, or whole cities, are commonly limited in statistical power because the number of randomized units is small, leaving few degrees of freedom for residual (between-unit) error. The authors describe a method for increasing residual degrees of freedom in a community experiment without substantially increasing cost or difficulty. In brief, they propose that the experimental units should be divided into random subsamples (batches). Batch sampling can improve statistical power if the community endpoint means are stable over time or if their temporal variation is comparable in period to the batch-sampling schedule. The authors demonstrate the theoretical advantages of the batch system and illustrate its use with data from the Pawtucket Heart Health Program, in which such a design was implemented.     

Residential segregation in Australian cities: a literature review

"In this review of literature dealing with the postwar immigrant experience in urban Australia, some of the key interpretations of residential segregation are assessed. The article focuses on studies which have examined ethnic clusters formed by southern Europeans in Sydney and Melbourne and more recently by Indochinese refugees. Much of the analysis to date has been based on measuring static residential patterns rather than social interaction, and the need to question the significance of ethnic concentrations which sometimes characterize the early stages of immigrant adaptation is suggested."     

Solving the part families problem in discrete-parts manufacture by simulated annealing

In the production planning and control of discrete-parts manufacture, aggregation of parts into families, on the basis of similarity, is carried out to ease both long-horizon planning and short-horizon scheduling. Additional advantages are related to those of group technology (GT), such as simplifying the flow of parts and tools and reducing both set-up and production costs. The problem of formally forming part families is presented and discussed. Previous work is reviewed and assessed. Two solution approaches, one based on a location model, the other on simulated annealing, are presented and compared along with test results. The location formulation, which results in an integer programming model solved by Lagrangian relaxation, proved capable of producing solutions of excellent quality, but only for relatively small problem instances. In contrast, simulated annealing, which is a general heuristic approach to combinatorial optimization, produced solutions of comparable quality and could handle realistically large problem instances. However, careful design of the simulated annealing algorithm is crucially important. An effective design is presented.     

Analysis of Competing Risks by Using Bayesian Smoothing

We consider the competing risks set-up. In many practical situations, the conditional probability of the cause of failure given the failure time is of direct interest. We propose to model the competing risks by the overall hazard rate and the conditional probabilities rather than the cause-specific hazards. We adopt a Bayesian smoothing approach for both quantities of interest. Illustrations are given at the end.     

Wild foods and household food security responses to AIDS: evidence from South Africa

Wild foods may offer unique benefits to households afflicted by AIDS, providing a nutritious and freely available food source at minimal labour and financial costs. This article presents the results of food security assessments in two rural South African sites. Detailed household dietary recalls from 227 households, combined with qualitative work, explored the association of household AIDS proxies (recent morbidity, mortality and orphan fostering) with household food security and dietary composition. The study found that AIDS-proxy households were significantly more food insecure, and households fostering orphans were both poorer and more food insecure. Wild foods were evident in 40.3% of the 48 h recalls, with significantly greater likelihood of use in households with fostering paternal orphans, and/or with at least one AIDS proxy. Only paternal orphans were significantly associated with likelihood of using wild foods when controlling for household socio-economic status. Qualitative data suggests that households afflicted by AIDS might curtail their use of wild foods due to household labour shortages and stigma. This is unfortunate, as regressions indicate that households using wild foods may be more economically resilient. This may be particularly important for households registering AIDS proxies, due to a demonstrated negative correlation between accumulated household AIDS proxies and household income.     

A new heteroscedastic regression to analyze mass loss of wood in civil construction in Brazil

A heteroscedastic regression based on the odd log-logistic Marshall–Olkin normal (OLLMON) distribution is defined by extending previous models. Some structural properties of this distribution are presented. The estimation of the parameters is addressed by maximum likelihood. For different parameter settings, sample sizes and some scenarios, various simulations investigate the performance of the heteroscedastic OLLMON regression. We use residual analysis to detect influential observations and to check the model assumptions. The new regression explains the mass loss of different wood species in civil construction in Brazil.