Improved Transferability of Data‐Driven Damage Models Through Sample Selection Bias Correction

Damage models for natural hazards are used for decision making on reducing and transferring risk. The damage estimates from these models depend on many variables and their complex sometimes nonlinear relationships with the damage. In recent years, data‐driven modeling techniques have been used to capture those relationships. The available data to build such models are often limited. Therefore, in practice it is usually necessary to transfer models to a different context. In this article, we show that this implies the samples used to build the model are often not fully representative for the situation where they need to be applied on, which leads to a “sample selection bias.” In this article, we enhance data‐driven damage models by applying methods, not previously applied to damage modeling, to correct for this bias before the machine learning (ML) models are trained. We demonstrate this with case studies on flooding in Europe, and typhoon wind damage in the Philippines. Two sample selection bias correction methods from the ML literature are applied and one of these methods is also adjusted to our problem. These three methods are combined with stochastic generation of synthetic damage data. We demonstrate that for both case studies, the sample selection bias correction techniques reduce model errors, especially for the mean bias error this reduction can be larger than 30%. The novel combination with stochastic data generation seems to enhance these techniques. This shows that sample selection bias correction methods are beneficial for damage model transfer.     

A new generalized Weibull distribution in income economic inequality curves

Researchers have been developing various extensions and modified forms of the Weibull distribution to enhance its capability for modeling and fitting different data sets. In this note, we investigate the potential usefulness of the new modification to the standard Weibull distribution called odd Weibull distribution in income economic inequality studies. Some mathematical and statistical properties of this model are proposed. We obtain explicit expressions for the first incomplete moment, quantile function, Lorenz and Zenga curves and related inequality indices. In addition to the well-known stochastic order based on Lorenz curve, the stochastic order based on Zenga curve is considered. Since the new generalized Weibull distribution seems to be suitable to model wealth, financial, actuarial and especially income distributions, these findings are fundamental in the understanding of how parameter values are related to inequality. Also, the estimation of parameters by maximum likelihood and moment methods is discussed. Finally, this distribution has been fitted to United States and Austrian income data sets and has been found to fit remarkably well in compare with the other widely used income models.     

Properties of the weighted log‐rank test in the design of confirmatory studies with delayed effects

Proportional hazards are a common assumption when designing confirmatory clinical trials in oncology. This assumption not only affects the analysis part but also the sample size calculation. The presence of delayed effects causes a change in the hazard ratio while the trial is ongoing since at the beginning we do not observe any difference between treatment arms, and after some unknown time point, the differences between treatment arms will start to appear. Hence, the proportional hazards assumption no longer holds, and both sample size calculation and analysis methods to be used should be reconsidered. The weighted log‐rank test allows a weighting for early, middle, and late differences through the Fleming and Harrington class of weights and is proven to be more efficient when the proportional hazards assumption does not hold. The Fleming and Harrington class of weights, along with the estimated delay, can be incorporated into the sample size calculation in order to maintain the desired power once the treatment arm differences start to appear. In this article, we explore the impact of delayed effects in group sequential and adaptive group sequential designs and make an empirical evaluation in terms of power and type‐I error rate of the of the weighted log‐rank test in a simulated scenario with fixed values of the Fleming and Harrington class of weights. We also give some practical recommendations regarding which methodology should be used in the presence of delayed effects depending on certain characteristics of the trial.     

The context of employment discrimination: interpreting the findings of a field experiment

Although field experiments have documented the contemporary relevance of discrimination in employment, theories developed to explain the dynamics of differential treatment cannot account for differences across organizational and institutional contexts. In this article, I address this shortcoming by presenting the main empirical findings from a multi‐method research project, in which a field experiment of ethnic discrimination in the Norwegian labour market was complemented with forty‐two in‐depth interviews with employers who were observed in the first stage of the study. While the experimental data support earlier findings in documenting that ethnic discrimination indeed takes place, the qualitative material suggests that theorizing in the field experiment literature have been too concerned with individual and intra‐psychic explanations. Discriminatory outcomes in employment processes seems to be more dependent on contextual factors such as the number of applications received, whether requirements are specified, and the degree to which recruitment procedures are formalized. I argue that different contexts of employment provide different opportunity structures for discrimination, a finding with important theoretical and methodological implications.     

Wage and Non-Wage Compensation Among Young Alcoholic and Heavy Drinking Women: A Preliminary Analysis

The association between drinking and selected job characteristics among women aged 24 to 31 is examined. Using the 1989 NLSY, women are classified as alcohol abusers or dependent based on DSM-III-R criteria or as heavy drinkers based on reported frequency of six or more drinks. Heavy drinking is negatively associated with wage and non-wage compensation. These effects diminish when human capital measures are controlled. Current alcoholism and current and past heavy drinking are strongly negatively associated with years of schooling. The association between alcoholism and job compensation and characteristics is not as strong as that seen for heavy drinking. It is not known if this is a consequence of errors in identifying alcoholic women in population-based surveys.     

The identification of logistic regression models with errors in the variables

The simple logistic regression model with normal measurement error and normal regressor is shown to be identifiable without any extra information about the measurement error. The multiple logistic regression model with more than one regressor variable measured with error is not identifiable. If the covariance matrix of the measurement error is known up to a scalar factor, the model is identified. Further we discuss why in spite of the identifiability the models cannot be estimated in a reasonable way without extra information about the measurement error.