L’arbre de régression multivariable et les modèles linéaires généralisés revisités : applications à l’étude de la diversité bêta et à l’estimation de la biomasse d’arbres tropicaux

En écologie, dans le cadre par exemple d’études des services fournis par les écosystèmes, les modélisations descriptive, explicative et prédictive ont toutes trois leur place distincte. Certaines situations bien précises requièrent soit l’un soit l’autre de ces types de modélisation ; le bon choix s’impose afin de pouvoir faire du modèle un usage conforme aux objectifs de l’étude. Dans le cadre de ce travail, nous explorons dans un premier temps le pouvoir explicatif de l’arbre de régression multivariable (ARM). Cette méthode de modélisation est basée sur un algorithme récursif de bipartition et une méthode de rééchantillonage permettant l’élagage du modèle final, qui est un arbre, afin d’obtenir le modèle produisant les meilleures prédictions. Cette analyse asymétrique à deux tableaux permet l’obtention de groupes homogènes d’objets du tableau réponse, les divisions entre les groupes correspondant à des points de coupure des variables du tableau explicatif marquant les changements les plus abrupts de la réponse. Nous démontrons qu’afin de calculer le pouvoir explicatif de l’ARM, on doit définir un coefficient de détermination ajusté dans lequel les degrés de liberté du modèle sont estimés à l’aide d’un algorithme. Cette estimation du coefficient de détermination de la population est pratiquement non biaisée. Puisque l’ARM sous-tend des prémisses de discontinuité alors que l’analyse canonique de redondance (ACR) modélise des gradients linéaires continus, la comparaison de leur pouvoir explicatif respectif permet entre autres de distinguer quel type de patron la réponse suit en fonction des variables explicatives. La comparaison du pouvoir explicatif entre l’ACR et l’ARM a été motivée par l’utilisation extensive de l’ACR afin d’étudier la diversité bêta. Toujours dans une optique explicative, nous définissons une nouvelle procédure appelée l’arbre de régression multivariable en cascade (ARMC) qui permet de construire un modèle tout en imposant un ordre hiérarchique aux hypothèses à l’étude. Cette nouvelle procédure permet d’entreprendre l’étude de l’effet hiérarchisé de deux jeux de variables explicatives, principal et subordonné, puis de calculer leur pouvoir explicatif. L’interprétation du modèle final se fait comme dans une MANOVA hiérarchique. On peut trouver dans les résultats de cette analyse des informations supplémentaires quant aux liens qui existent entre la réponse et les variables explicatives, par exemple des interactions entres les deux jeux explicatifs qui n’étaient pas mises en évidence par l’analyse ARM usuelle. D’autre part, on étudie le pouvoir prédictif des modèles linéaires généralisés en modélisant la biomasse de différentes espèces d’arbre tropicaux en fonction de certaines de leurs mesures allométriques. Plus particulièrement, nous examinons la capacité des structures d’erreur gaussienne et gamma à fournir les prédictions les plus précises. Nous montrons que pour une espèce en particulier, le pouvoir prédictif d’un modèle faisant usage de la structure d’erreur gamma est supérieur. Cette étude s’insère dans un cadre pratique et se veut un exemple pour les gestionnaires voulant estimer précisément la capture du carbone par des plantations d’arbres tropicaux. Nos conclusions pourraient faire partie intégrante d’un programme de réduction des émissions de carbone par les changements d’utilisation des terres.

A Sensitivity Analysis of Cross-Country Growth Regressions: Is 1990-2010 Different?

Cet article étudie la sensibilité des estimations de certaines variables explicatives de la croissance économique dans des régressions en coupe transversale sur un ensemble de pays. Il applique un modèle modifié de l’analyse de sensibilité de Leamer (1983, 1985). Mes résultats confirment la conclusion de Levine and Renelt (1992), toutefois, je montre que plus de variables sont solidement corrélées à la croissance économique. Entre 1990-2010, je trouve que huit sur vingt cinq variables ont des coefficients significatifs et sont solidement corrélées à la croissance de long terme, notamment, les parts de l’investissement et des dépenses étatiques dans le PIB, la primauté du droit et une variable dichotomique pour les pays subsahariens. Je trouve aussi une preuve empirique solide de l'hypothèse de la convergence conditionnelle, ce qui est cohérent avec le modèle de croissance néoclassique.

The Double Pareto-Lognormal Distribution and its applications in actuarial science and finance

Le but de ce mémoire de maîtrise est de décrire les propriétés de la loi double Pareto-lognormale, de montrer comment on peut introduire des variables explicatives dans le modèle et de présenter son large potentiel d'applications dans le domaine de la science actuarielle et de la finance. Tout d'abord, nous donnons la définition de la loi double Pareto-lognormale et présentons certaines de ses propriétés basées sur les travaux de Reed et Jorgensen (2004). Les paramètres peuvent être estimés en utilisant la méthode des moments ou le maximum de vraisemblance. Ensuite, nous ajoutons une variable explicative à notre modèle. La procédure d'estimation des paramètres de ce mo-\\dèle est également discutée. Troisièmement, des applications numériques de notre modèle sont illustrées et quelques tests statistiques utiles sont effectués.

Régression logistique bayésienne : comparaison de densités a priori

La régression logistique est un modèle de régression linéaire généralisée (GLM) utilisé pour des variables à expliquer binaires. Le modèle cherche à estimer la probabilité de succès de cette variable par la linéarisation de variables explicatives. Lorsque l’objectif est d’estimer le plus précisément l’impact de différents incitatifs d’une campagne marketing (coefficients de la régression logistique), l’identification de la méthode d’estimation la plus précise est recherchée. Nous comparons, avec la méthode MCMC d’échantillonnage par tranche, différentes densités a priori spécifiées selon différents types de densités, paramètres de centralité et paramètres d’échelle. Ces comparaisons sont appliquées sur des échantillons de différentes tailles et générées par différentes probabilités de succès. L’estimateur du maximum de vraisemblance, la méthode de Gelman et celle de Genkin viennent compléter le comparatif. Nos résultats démontrent que trois méthodes d’estimations obtiennent des estimations qui sont globalement plus précises pour les coefficients de la régression logistique : la méthode MCMC d’échantillonnage par tranche avec une densité a priori normale centrée en 0 de variance 3,125, la méthode MCMC d’échantillonnage par tranche avec une densité Student à 3 degrés de liberté aussi centrée en 0 de variance 3,125 ainsi que la méthode de Gelman avec une densité Cauchy centrée en 0 de paramètre d’échelle 2,5.

Regional flows of domestic tourism in Spain = Flujos regionales del turismo doméstico en España

The paper analyses the regional flows of domestic tourism that took place in Spain in year 2000, contributing to the state of knowledge on tourism required by authorities and private firms when faced with decision making, for example, for regional infrastructure planning. Although tourism is one of the main income-generating economic activities in Spain, domestic tourism has received little attention in the literature compared to inbound tourism. The paper uses among others, gravitational model tools and concentration indices, to analyse regional concentration of both domestic demand and supply; tourism flows among regions, and the causes that may explain the observed flows and attractiveness between regions. Among the most remarkable results are the high regional concentration of demand and supply, and the role of population and regional income as explanatory variables. Also remarkable are the attractiveness of own region and neighbour ones, and that domestic tourism may be acting as a regional income redistributing activity

A Log-linear model of grain size influence on the geochemistry of sediments

Sediment composition is mainly controlled by the nature of the source rock(s), and chemical (weathering) and physical processes (mechanical crushing, abrasion, hydrodynamic sorting) during alteration and transport. Although the factors controlling these processes are conceptually well understood, detailed quantification of compositional changes induced by a single process are rare, as are examples where the effects of several processes can be distinguished. The present study was designed to characterize the role of mechanical crushing and sorting in the absence of chemical weathering. Twenty sediment samples were taken from Alpine glaciers that erode almost pure granitoid lithologies. For each sample, 11 grain-size fractions from granules to clay (ø grades <-1 to >9) were separated, and each fraction was analysed for its chemical composition. The presence of clear steps in the box-plots of all parts (in adequate ilr and clr scales) against ø is assumed to be explained by typical crystal size ranges for the relevant mineral phases. These scatter plots and the biplot suggest a splitting of the full grain size range into three groups: coarser than ø=4 (comparatively rich in SiO2, Na2O, K2O, Al2O3, and dominated by “felsic” minerals like quartz and feldspar), finer than ø=8 (comparatively rich in TiO2, MnO, MgO, Fe2O3, mostly related to “mafic” sheet silicates like biotite and chlorite), and intermediate grains sizes (4≤ø <8; comparatively rich in P2O5 and CaO, related to apatite, some feldspar). To further test the absence of chemical weathering, the observed compositions were regressed against three explanatory variables: a trend on grain size in ø scale, a step function for ø≥4, and another for ø≥8. The original hypothesis was that the trend could be identified with weathering effects, whereas each step function would highlight those minerals with biggest characteristic size at its lower end. Results suggest that this assumption is reasonable for the step function, but that besides weathering some other factors (different mechanical behavior of minerals) have also an important contribution to the trend. Key words: sediment, geochemistry, grain size, regression, step function

Experimental design on the simplex

Optimum experimental designs depend on the design criterion, the model and the design region. The talk will consider the design of experiments for regression models in which there is a single response with the explanatory variables lying in a simplex. One example is experiments on various compositions of glass such as those considered by Martin, Bursnall, and Stillman (2001). Because of the highly symmetric nature of the simplex, the class of models that are of interest, typically Scheff´e polynomials (Scheff´e 1958) are rather different from those of standard regression analysis. The optimum designs are also rather different, inheriting a high degree of symmetry from the models. In the talk I will hope to discuss a variety of modes for such experiments. Then I will discuss constrained mixture experiments, when not all the simplex is available for experimentation. Other important aspects include mixture experiments with extra non-mixture factors and the blocking of mixture experiments. Much of the material is in Chapter 16 of Atkinson, Donev, and Tobias (2007). If time and my research allows, I would hope to finish with a few comments on design when the responses, rather than the explanatory variables, lie in a simplex. References Atkinson, A. C., A. N. Donev, and R. D. Tobias (2007). Optimum Experimental Designs, with SAS. Oxford: Oxford University Press. Martin, R. J., M. C. Bursnall, and E. C. Stillman (2001). Further results on optimal and efficient designs for constrained mixture experiments. In A. C. Atkinson, B. Bogacka, and A. Zhigljavsky (Eds.), Optimal Design 2000, pp. 225–239. Dordrecht: Kluwer. Scheff´e, H. (1958). Experiments with mixtures. Journal of the Royal Statistical Society, Ser. B 20, 344–360. 1

Networks of PhD Students and Academic Performance: A Comparison across Countries

In this article we compare regression models obtained to predict PhD students’ academic performance in the universities of Girona (Spain) and Slovenia. Explanatory variables are characteristics of PhD student’s research group understood as an egocentered social network, background and attitudinal characteristics of the PhD students and some characteristics of the supervisors. Academic performance was measured by the weighted number of publications. Two web questionnaires were designed, one for PhD students and one for their supervisors and other research group members. Most of the variables were easily comparable across universities due to the careful translation procedure and pre-tests. When direct comparison was not possible we created comparable indicators. We used a regression model in which the country was introduced as a dummy coded variable including all possible interaction effects. The optimal transformations of the main and interaction variables are discussed. Some differences between Slovenian and Girona universities emerge. Some variables like supervisor’s performance and motivation for autonomy prior to starting the PhD have the same positive effect on the PhD student’s performance in both countries. On the other hand, variables like too close supervision by the supervisor and having children have a negative influence in both countries. However, we find differences between countries when we observe the motivation for research prior to starting the PhD which increases performance in Slovenia but not in Girona. As regards network variables, frequency of supervisor advice increases performance in Slovenia and decreases it in Girona. The negative effect in Girona could be explained by the fact that additional contacts of the PhD student with his/her supervisor might indicate a higher workload in addition to or instead of a better advice about the dissertation. The number of external student’s advice relationships and social support mean contact intensity are not significant in Girona, but they have a negative effect in Slovenia. We might explain the negative effect of external advice relationships in Slovenia by saying that a lot of external advice may actually result from a lack of the more relevant internal advice

Determinantes de la demanda de turismo en Colombia 2004 2007: seguridad, comercio y otros factores

El turismo es un sector de gran dinamismo en países en desarrollo, casos como México o Perú son modelos latinoamericanos que se han destacado por el desarrollo del sector turístico y en años recientes Colombia ha presentado cifras importantes respecto a los demás países de la región. Aquí se pueden practicar cuatro líneas de turismo potenciales según el Plan Sectorial de Turismo Nacional 2008 2011 Un Destino de Clase Mundial como son el Turismo Ecológico, Cultural, de Salud y de Convenciones y Eventos. El objetivo de este documento es hallar los principales factores que influyeron en la llegada mensual de viajeros extranjeros a Colombia en el periodo 2004 2007 a través de la estimación de modelos de panel de datos, mediante la utilización tanto de variables microeconómicas como macroeconómicas así como otras variables culturales y geográficas. Como resultado se puede resaltar que factores como el número de llegadas del periodo inmediatamente anterior y el índice de intercambio comercial influyen de manera positiva, mientras el índice de secuestros, reduce de forma significativa el número de llegadas de viajeros extranjeros a Colombia.

Determinantes de la estructura de capital en Colombia una aproximación de regresión por cuartiles

Con el objetivo de conocer la influencia de los diferentes niveles de apalancamiento en el crecimiento de las empresas colombianas, surge la necesidad de responder la siguiente pregunta: ¿qué determina la elección de estructura de capital de las firmas? La regresión por cuantiles permite examinar toda la distribución de las firmas y no solo una medida de la tendencia central de la distribución de la estructura de capital. De esta manera se puede evaluar la importancia relativa de las diferentes variables explicativas en diferentes puntos de la distribución del apalancamiento de las firmas. Razón por la cual se utilizará esta aproximación; sin embargo también se utilizará el método de regresión para datos de panel (también llamado datos longitudinales) con efectos aleatorios, para comparar resultados, esto teniendo en cuenta que los datos no sólo varían entre observaciones sino también en el tiempo. De esta manera, aplicar el método de regresión por cuantiles, permite darle una mirada más profunda a la elección del nivel de apalancamiento, pues permite discriminar el efecto de las variables entre firmas altamente apalancadas y bajamente apalancadas.

Academic achievement in sciences: the role of preferences and educative assets

This paper provides new evidence on the effect of pupil’s self-motivation and academic assets allocation on the academic achievement in sciences across countries. By using the Programme for International Student Assessment 2006 (PISA 2006) test we find that both explanatory variables have a positive effect on student’s performance. Self-motivation is measured through an instrument that allows us to avoid possible endogeneity problems. Quantile regression is used for analyzing the existence of different estimated coefficients over the distribution. It is found that both variables have different effect on academic performance depending on the pupil’s score. These findings support the importance of designing focalized programs for different populations, especially in terms of access to information and communication technologies such as internet.

How do inequality affect tax morale in Latin America and Caribbean?

Latin America is well known as an inequitable region. As it is recognized, inequality and corruption perception weaken the way that political institutions works and the democratic system. Focusing on Latin American and Caribbean countries, this paper analyzes what are the elements shaping tax morale. In particular, how the context influences ethical grounds decisions such as the predisposition to pay taxes is analyzed, using the survey carried out in 2005 by Latinobarometro. The objective is to analyze how country performance determines tax morale. To do so, four probitmodels are estimated using Gini index, Transparency International Corruption Perception Index and Gross Domestic Product per capita (GDPpc) as explanatory variables. As expected we found that some socio-demographic variables play a relevant role. Interestingly, we also found that, in this attitude, LAC countries do not register a gender bias. However, those are not our main contributions to the literature on the field. The most important results are linked to: 1) the levelmatters, GDPpc increases the probability of people having tax morale, 2) moreover, income distributionalso influence on tax morale but in opposite direction and 3) corruption perception also reduces tax morale. Those results show that the quality of institutions matters and therefore, the way that democracy works play a relevant role.

Breast cancer incidence and overdiagnosis in Catalonia (Spain)

Early detection of breast cancer (BC) with mammography may cause overdiagnosis and overtreatment, detecting tumors which would remain undiagnosed during a lifetime. The aims of this study were: first, to model invasive BC incidence trends in Catalonia (Spain) taking into account reproductive and screening data; and second, to quantify the extent of BC overdiagnosis. We modeled the incidence of invasive BC using a Poisson regression model. Explanatory variables were: age at diagnosis and cohort characteristics (completed fertility rate, percentage of women that use mammography at age 50, and year of birth). This model also was used to estimate the background incidence in the absence of screening. We used a probabilistic model to estimate the expected BC incidence if women in the population used mammography as reported in health surveys. The difference between the observed and expected cumulative incidences provided an estimate of overdiagnosis.Incidence of invasive BC increased, especially in cohorts born from 1940 to 1955. The biggest increase was observed in these cohorts between the ages of 50 to 65 years, where the final BC incidence rates more than doubled the initial ones. Dissemination of mammography was significantly associated with BC incidence and overdiagnosis. Our estimates of overdiagnosis ranged from 0.4% to 46.6%, for women born around 1935 and 1950, respectively.Our results support the existence of overdiagnosis in Catalonia attributed to mammography usage, and the limited malignant potential of some tumors may play an important role. Women should be better informed about this risk. Research should be oriented towards personalized screening and risk assessment tools

Comparing meta-analysis and ecological-longitudinal analysis in time-series studies: a case study of the effects of air pollution on mortality in three Spanish cities

The objective of this paper is to introduce a diVerent approach, called the ecological-longitudinal, to carrying out pooled analysis in time series ecological studies. Because it gives a larger number of data points and, hence, increases the statistical power of the analysis, this approach, unlike conventional ones, allows the complementation of aspects such as accommodation of random effect models, of lags, of interaction between pollutants and between pollutants and meteorological variables, that are hardly implemented in conventional approaches. Design—The approach is illustrated by providing quantitative estimates of the short-termeVects of air pollution on mortality in three Spanish cities, Barcelona,Valencia and Vigo, for the period 1992–1994. Because the dependent variable was a count, a Poisson generalised linear model was first specified. Several modelling issues are worth mentioning. Firstly, because the relations between mortality and explanatory variables were nonlinear, cubic splines were used for covariate control, leading to a generalised additive model, GAM. Secondly, the effects of the predictors on the response were allowed to occur with some lag. Thirdly, the residual autocorrelation, because of imperfect control, was controlled for by means of an autoregressive Poisson GAM. Finally, the longitudinal design demanded the consideration of the existence of individual heterogeneity, requiring the consideration of mixed models. Main results—The estimates of the relative risks obtained from the individual analyses varied across cities, particularly those associated with sulphur dioxide. The highest relative risks corresponded to black smoke in Valencia. These estimates were higher than those obtained from the ecological-longitudinal analysis. Relative risks estimated from this latter analysis were practically identical across cities, 1.00638 (95% confidence intervals 1.0002, 1.0011) for a black smoke increase of 10 μg/m3 and 1.00415 (95% CI 1.0001, 1.0007) for a increase of 10 μg/m3 of sulphur dioxide. Because the statistical power is higher than in the individual analysis more interactions were statistically significant,especially those among air pollutants and meteorological variables. Conclusions—Air pollutant levels were related to mortality in the three cities of the study, Barcelona, Valencia and Vigo. These results were consistent with similar studies in other cities, with other multicentric studies and coherent with both, previous individual, for each city, and multicentric studies for all three cities

Métodos de estadística espacial para evaluar la influencia de factores medioambientales sobre la incidencia y mortalidad por cáncer

Los objetivos de la tesis son: 1.- Estudiar la relación entre la incidencia y mortalidad por cáncer y los factores medioambientales, en particular la contaminación atmosférica, controlando por factores socioeconómicos. 2.- Utilizar aquellos métodos de estadística espacial apropiados para cada tipo de diseño. 3.- Distinguir en los modelos las diferentes fuentes de extra-variabilidad espacial. 4.- Controlar el problema de exceso de ceros inherente a alguna de las neoplasias de interés medioambientales. Conclusiones: - Tanto la incidencia como la mortalidad de las neoplasias, presentaron dos fuentes de extravariación. La extravariaicón espacial, por la que unidades vecinas tienden a presentar razones de incidencia/mortalidad similares, y la heterogeneidad no espacial. En general la extravariabilidad espacial ha resultado ser mucho mayor que la no espacial. - Para suavizar las RIE/RME correspondientes a variables con un porcentaje de ceros superior al40-50% debe utilizarse un modelo que capture este comportamiento. - El mejor modelo en términos de ajuste para recoger el exceso de ceros en las variables de interés ha resultado ser el modelo mixto de riesgo relativo. - Las RIE/RME suavizadas presentan un patrón geográfico claro sólo en algunas neoplasias de interés medioambiental. - Parte de la variabilidad remanente en las RIE/RME suavizadas pudo ser explicada mediante la introducción de variables explicativas, en particular la contaminación atmosférica y variables socioeconómicas. -Como los contaminantes atmosféricos fueron observados en un diseño geoestadístico y las neoplasias de interés mediambiental lo fueron en un diseño en rejilla se modelizó la superficie de exposición. - El efecto del contaminante en cada municipio/sección censal se aproximó introduciendo en el modelo el valor promedio en cada área y la variabilidad intra-área. - El efecto del contaminante se consideró aleatorio, en el sentido de que podría ser diferente en cada una de las áreas. - Las condiciones socioeconómicas fueron otra de las variables que redujeron la variabilidad remanente en las RIE/RME suavizadas. -Las variables explicativas observadas con un diseño en rejilla, como el índice de privación, se introdujeron en el modelo como efectos fijos. - El efecto de la privación sobre la incidencia y/o mortalidad por cáncer de tráquea, bronquios y pulmón, controlando por contaminantes atmosféricos, fue mayor en las mujeres que en los hombres. -Altas concentraciones de contaminantes atmosféricos aumentan el riesgo de padecer neoplasias de interés medioambiental, controlando por condiciones socioeconómicas.