A test for crossing receiver operating characteristic (roc) curves

The receiver operating characteristic (ROC) curve gives a graphical representation of sensitivity and specificity of a prediction model when varying the decision treshold on a diagnostic criterion. A classical test for comparing the overall accuracies for two models -1 and 2- is based on the difference between ROC curves areas - related to its standard error. This test is designed for the situation where ROC curve 1 caps ROC curve 2. Often both curves cross :in this paper, a new test, based on the integrated difference between the curves, is proposed to deal with this situation. In a simulation experiment, the new test was less powerful than the old test for detecting an overall superiority, but much more powerfull against the crossing alternative.     

Preserving dominance relations through disaggregation: the evil and the saint

This article investigates the circumstances in which stochastic dominance relations at any finite degree at the household level can be assumed to be preserved at the individual level. We find necessary and sufficient conditions on the common sharing function adopted by households to divide the cake among a “strong” and a “weak” individual. The sharing function which maps the household income into the outcome of the weak individual must belong to the class of utility functions which supports the stochastic order. In addition, the household must follow a compensating rule, meaning that the share of resources devoted to the weak individual increases with household income. Applications to fiscal federalism are also proposed. “All inequality is a source of evil - for by the inferior more is lost in the account of happiness than is gained by the superior” J. Bentham, First Principle preparatory to Constitutional Code, 1822.      

Analysis of organizational conditions for risk management: the case study of a petrochemical site

This paper presents an ergonomic intervention in the petrochemical sector. The scheduled shutdown of one of the gas production sites has led the management to reduce the number of personnel on site, and then to get new recruits and experienced technicians from other sites as the policy for leaving personnel had not been properly planned, resulting in understaffing on site. Workers with seniority on the site, and who are also the most experienced do not accept the way newcomers are induced on site, whereas the management accuses them of resisting change. The intervention consisted in reconnecting local and corporate management through making the work activity visible and linking two sets of data that they held separately. Different types of analyses were made, work demography, decision making processes and tools used by the management, analysis of the building of career and work logics. Those different levels of analysis are gathered in macro-ergonomics, while showing the possible combinations between top down and bottom up approaches. The intervention resulted in concrete changes: HR simulation tool, training organisation, feedback.     

Longitudinal Links Between Gambling Participation and Academic Performance in Youth: A Test of Four Models

Gambling participation and low academic performance are related during adolescence, but the causal mechanisms underlying this link are unclear. It is possible that gambling participation impairs academic performance. Alternatively, the link between gambling participation and low academic performance could be explained by common underlying risk factors such as impulsivity and socio-family adversity. It could also be explained by other current correlated problem behaviors such as substance use. The goal of the present study was to examine whether concurrent and longitudinal links between gambling participation and low academic performance exist from age 14 to age 17 years, net of common antecedent factors and current substance use. A convenience sample of 766 adolescents (50.6% males) from a longitudinal twin sample participated in the study. Analyses revealed significant, albeit modest, concurrent links at both ages between gambling participation and academic performance. There was also a longitudinal link between gambling participation at age 14 and academic performance at age 17, which persisted after controlling for age 12 impulsivity and socio-family adversity as well as current substance use. Gambling participation predicts a decrease in academic performance during adolescence, net of concurrent and antecedent personal and familial risk factors.     

Encompassing and indirect inference

Summary In the literature on encompassing [see e.g. Mizon-Richard (1986), Hendry-Richard (1990), Florens-Hendry-Richard (1987)] there is a basic contradiction: on the one hand it is said that it is not possible to assume that the true distribution belongs to one of two competing modelM ₁ andM ₂, but, on the other hand, this assumption is made in the study of encompassing tests. In this paper we first propose a formal definition of encompassing, we then briefly examine the properties of this notion and we propose encompassing tests which do not assume that the true distribution belongs toM ₁ orM ₂; these tests are based on simulations. Finally, generalizing an idea used in the definition of an encompassing test (the GET test) we propose a new kind of inference, called indirect inference, which allows for estimation and test procedures when the model is too complicated to be treated by usual methods (for instance maximum likelihood methods); the only assumption made on the model is that it can be simulated, which seems to be a minimal requirement. This new class of inference methods can be used in a large number of domains and some examples are given. The present paper is based on Gouriéroux-Monfort (1992), and Gouriéroux-Monfort-Renault (1993), respectively GM and GMR hereafter. Invited paper at the Conference on ?Statistical Tests: Methodology and Econometric Applications?, held in Bologna, Italy, 27–28 May 1993.     

Work and mental health: the experience of the Quebec workforce between 1987 and 1998

This study examines the distribution of psychological distress in twelve occupational groups over the decade 1987-1998 in the Quebec workforce. Cross-sectional data from the three phases of the Quebec Health and Social Survey are used with n = 9,450 in 1987, n = 10,947 in 1992 and n = 10,960 in 1998, totalling 31,357 workers aged 15 and over. Occupations are classified according to the Canadian Socio-economic Classification of Occupations. Prevalence estimates for occupational groups are computed and logistic regression analyses are conducted controlling for gender, age and marital status. The results show that the prevalence of workers with psychological distress increased sharply between 1987 and 1992 and declined back in 1998 but still increased compared to 1987. However, only non-qualified white collars, semi-qualified blue collars and male non-qualified blue collars show a significant increment in psychological distress over time. Analysis of the differentials in the prevalence of psychological distress gives greater odds of distress for supervisors, semi-qualified white and blue collar workers compared to upper managers. The odds for occupations are stable over time, gender, age and marital status. The odds of female workers significantly decreased in the three phases. It appears that the restructuring of the work environment and the perturbations in the larger society promoted an increase of psychological distress within definite segments of the workforce. The specific contribution of occupation is limited but supervisors and occupations requiring lower qualifications are more at risk regarding mental health at work.     

Importance sampling with the generalized exponential power density

In this paper, the generalized exponential power (GEP) density is proposed as an importance function in Monte Carlo simulations in the context of estimation of posterior moments of a location parameter. This density is divided in five classes according to its tail behaviour which may be exponential, polynomial or logarithmic. The notion of p-credence is also defined to characterize and to order the tails of a large class of symmetric densities by comparing their tails to those of the GEP density.The choice of the GEP density as an importance function allows us to obtain reliable and effective results when p-credences of the prior and the likelihood are defined, even if there are conflicting sources of information. Characterization of the posterior tails using p-credence can be done. Hence, it is possible to choose parameters of the GEP density in order to have an importance function with slightly heavier tails than the posterior. Simulation of observations from the GEP density is also addressed.     

Social security: Indispensable solidarity

Social security must be considered not as an economic burden, but as a collection of mechanisms for collective solidarity based on the principle of income redistribution. The economic and social transformations that the various regions of the world are going through, in particular the globalization of the economy, are making social security increasingly necessary. Instead of yielding to proposals for the privatization of schemes — a formula that is less efficient and more expensive — we should be protecting and improving social security; developing it in those countries where its position is still weak; and stressing that, while its objectives are humane and social, it brings economic benefits too.     

Insularity and Governance: The Case of French Polynesia

Insularity, smallness and remoteness help explain specific tendencies noted in the governance of French Polynesia, a partially self-governing French overseas territory in the South Pacific, with a population of 220,000. Professional relations tend to become personalized, and impersonal and impartial procedures are hard to apply. The market for experts is small: some people are able to claim a monopoly on some type of expertise and then defend it by every means. Solidarity is vertical, not horizontal, which makes for a fragile democracy.     

Optimal Ordering Policies for Inventory Problems With Dynamic Information Delays

Information delays exist when the most recent inventory information available to the Inventory Manager (IM) is dated. In other words, the IM observes only the inventory level that belongs to an earlier period. Such situations are not uncommon, and they arise when it takes a while to process the demand data and pass the results to the IM. We introduce dynamic information delays as a Markov process into the standard multiperiod stochastic inventory problem with backorders. We develop the concept of a reference inventory position. We show that this position along with the magnitude of the latest observed delay and the age of this observation are sufficient statistics for finding the optimal order quantities. Furthermore, we establish that the optimal ordering policy is of state‐dependent base‐stock type with respect to the reference inventory position (or state‐dependent (s, S) type if there is a fixed ordering cost). The optimal base stock and (s, S) levels depend on the magnitude of the latest observed delay and the age of this observation. Finally, we study the sensitivity of the optimal base stock and the optimal cost with respect to the sufficient statistics.