Efficient Estimation of Fixed and Time-varying Covariate Effects in Multiplicative Intensity Models

The proportional hazards assumption of the Cox model does sometimes not hold in practise. An example is a treatment effect that decreases with time. We study a general multiplicative intensity model allowing the influence of each covariate to vary non-parametrically with time. An efficient estimation procedure for the cumulative parameter functions is developed. Its properties are studied using the martingale structure of the problem. Furthermore, we introduce a partly parametric version of the general non-parametric model in which the influence of some of the covariates varies with time while the effects of the remaining covariates are constant. This semiparametric model has not been studied in detail before. An efficient procedure for estimating the parametric as well as the non-parametric components of this model is developed. Again the martingale structure of the model allows us to describe the asymptotic properties of the suggested estimators. The approach is applied to two different data sets, and a Monte Carlo simulation is presented.     

BEHAVIOURAL SCIENCE RESEARCH AND THE RIGHTS OF HUMAN SUBJECTS*

L'examen de la question des libertés civiles des sujets expérimentaux humains est placé dans le cadre de l'accroissement de l'analyse au moyen dordinateurs et l'augmentation des banques de données recouvrables. Les chercheurs doivent se justifier devant un public de plus en plus grand. Une préoccupation se manifeste en particulier pour les implications cumulatives de I'accumulation de données et pour les questions impliquées quant à la distribution du pouvoir. Des exemples de difficultés de plus en plus sérieuses à obtenir la collaboration de sujets humains expérimentaux sont fournies comme base d'une plus grande préoccupation concernant toute la question des ordinateurs, de la vie privée et de la recherche dans les sciences du comportement. Une série de mesures appropriées sont recommendées. The issue of the civil liberties of human experimental subjects is examined in the light of the growth of computer analysis and retrievable storage of data. Researchers are becoming accountable to an increasing audience. Particular concern is demonstrated for both the cumulative implications of data-gathering and the underlying power-distribution issues involved. Examples of increasing difficulty in obtaining cooperation from human experimental subjects, whether in the laboratory or the field, are cited as a base for greater concern surrounding the entire issue of computers, privacy, and behavioural science research. A series of recommendations for appropriate action are reported.     

Simulating alternative organizational structures of production systems

Changes in international competition lead to changes of the requirements on production enterprises. The introduction of new production technologies does not seem to be exclusively an adequate reaction to the increasing problems. Therefore tools of computer-integrated production (CIM) and new organizational concepts have to be conceived. Because these organizational and technical means are linked on the one hand with severe changes in the production system and, on the other hand, with rather higher economic risks in their realization, the effects have to be estimated in advance in the planning stage. Consideration of dynamical system behaviour plays an important part because the main goal is the improvement of order processing. This paper presents a simulation program, which can be used during the design phase of the organizational structure of production systems as a powerful tool for predicting the effects of new computer-aided tools and structures.     

The Cause-Deleted Index: Estimating Cause of Death Contributions to Mortality

Causes underlying mortality disparities are often determined by causal decomposition. This method is based on the decomposition of differences in mortality or life expectancy into parameters representing the contribution of underlying causes. It quantifies disparities as differences in mortality rates and does not account for the fact that many underprivileged groups are more likely to die from nearly all causes. Results are driven by the frequency of cause of death. Alternatively, the cause deleted index quantifies the role of underlying causes in mortality disparities as the change in the relative risk of dying that is related to deleting a specific cause. The consistency between the methods in estimating cause of death contributions is analyzed using 2000 U.S. mortality data and simulated mortality profiles. The two methods often produce divergent results because causal decomposition relies on the prevalence of causes of death.