Modelos estadísticos y evaluación: tres estudios en educación

The educational system in Spain is undergoing a reorganization. At present, high-school graduates who want to enroll at a public university must take a set of examinations Pruebas de Aptitud para el Acceso a la Universidad (PAAU). A "new formula" (components, weights, type of exam,...) for university admission is been discussed. The present paper summarizes part of the research done by the author in her PhD. The context for this thesis is the evaluation of large-scale and complex systems of assessment. The main objectives were: to achieve a deep knowledge of the entire university admissions process in Spain, to discover the main sources of uncertainty and topromote empirical research in a continual improvement of the entire process. Focusing in the suitable statistical models and strategies which allow to high-light the imperfections of the system and reduce them, the paper develops, among other approaches, some applications of multilevel modeling.

Further insights on the French WISC-IV factor structure through Bayesian structural equation modeling

The interpretation of the Wechsler Intelligence Scale for Children-Fourth Edition (WISC-IV) is based on a 4-factor model, which is only partially compatible with the mainstream Cattell-Horn-Carroll (CHC) model of intelligence measurement. The structure of cognitive batteries is frequently analyzed via exploratory factor analysis and/or confirmatory factor analysis. With classical confirmatory factor analysis, almost all crossloadings between latent variables and measures are fixed to zero in order to allow the model to be identified. However, inappropriate zero cross-loadings can contribute to poor model fit, distorted factors, and biased factor correlations; most important, they do not necessarily faithfully reflect theory. To deal with these methodological and theoretical limitations, we used a new statistical approach, Bayesian structural equation modeling (BSEM), among a sample of 249 French-speaking Swiss children (8-12 years). With BSEM, zero-fixed cross-loadings between latent variables and measures are replaced by approximate zeros, based on informative, small-variance priors. Results indicated that a direct hierarchical CHC-based model with 5 factors plus a general intelligence factor better represented the structure of the WISC-IV than did the 4-factor structure and the higher order models. Because a direct hierarchical CHC model was more adequate, it was concluded that the general factor should be considered as a breadth rather than a superordinate factor. Because it was possible for us to estimate the influence of each of the latent variables on the 15 subtest scores, BSEM allowed improvement of the understanding of the structure of intelligence tests and the clinical interpretation of the subtest scores.

Modeling water movement in horizontal columns using fractal theory

Fractal mathematics has been used to characterize water and solute transport in porous media and also to characterize and simulate porous media properties. The objective of this study was to evaluate the correlation between the soil infiltration parameters sorptivity (S) and time exponent (n) and the parameters dimension (D) and the Hurst exponent (H). For this purpose, ten horizontal columns with pure (either clay or loam) and heterogeneous porous media (clay and loam distributed in layers in the column) were simulated following the distribution of a deterministic Cantor Bar with fractal dimension H" 0.63. Horizontal water infiltration experiments were then simulated using Hydrus 2D software. The sorptivity (S) and time exponent (n) parameters of the Philip equation were estimated for each simulation, using the nonlinear regression procedure of the statistical software package SAS®. Sorptivity increased in the columns with the loam content, which was attributed to the relation of S with the capillary radius. The time exponent estimated by nonlinear regression was found to be less than the traditional value of 0.5. The fractal dimension estimated from the Hurst exponent was 17.5 % lower than the fractal dimension of the Cantor Bar used to generate the columns.

Multi-level modeling of aspects associated with poor mental health in a sample of prehospital emergency professionals.

BACKGROUND: The goal of this paper is to investigate the respective influence of work characteristics, the effort-reward ratio, and overcommitment on the poor mental health of out-of-hospital care providers. METHODS: 333 out-of-hospital care providers answered a questionnaire that included queries on mental health (GHQ-12), demographics, health-related information and work characteristics, questions from the Effort-Reward Imbalance Questionnaire, and items about overcommitment. A two-level multiple regression was performed between mental health (the dependent variable) and the effort-reward ratio, the overcommitment score, weekly number of interventions, percentage of non-prehospital transport of patients out of total missions, gender, and age. Participants were first-level units, and ambulance services were second-level units. We also shadowed ambulance personnel for a total of 416 hr. RESULTS: With cutoff points of 2/3 and 3/4 positive answers on the GHQ-12, the percentages of potential cases with poor mental health were 20% and 15%, respectively. The effort-reward ratio was associated with poor mental health (P < 0.001), irrespective of age or gender. Overcommitment was associated with poor mental health; this association was stronger in women (β = 0.054) than in men (β = 0.020). The percentage of prehospital missions out of total missions was only associated with poor mental health at the individual level. CONCLUSIONS: Emergency medical services should pay attention to the way employees perceive their efforts and the rewarding aspects of their work: an imbalance of those aspects is associated with poor mental health. Low perceived esteem appeared particularly associated with poor mental health. This suggests that supervisors of emergency medical services should enhance the value of their employees' work. Employees with overcommitment should also receive appropriate consideration. Preventive measures should target individual perceptions of effort and reward in order to improve mental health in prehospital care providers.

Selecting statistical models to study the relationship between soybean yield and soil physical properties

Statistical models allow the representation of data sets and the estimation and/or prediction of the behavior of a given variable through its interaction with the other variables involved in a phenomenon. Among other different statistical models, are the autoregressive state-space models (ARSS) and the linear regression models (LR), which allow the quantification of the relationships among soil-plant-atmosphere system variables. To compare the quality of the ARSS and LR models for the modeling of the relationships between soybean yield and soil physical properties, Akaike's Information Criterion, which provides a coefficient for the selection of the best model, was used in this study. The data sets were sampled in a Rhodic Acrudox soil, along a spatial transect with 84 points spaced 3 m apart. At each sampling point, soybean samples were collected for yield quantification. At the same site, soil penetration resistance was also measured and soil samples were collected to measure soil bulk density in the 0-0.10 m and 0.10-0.20 m layers. Results showed autocorrelation and a cross correlation structure of soybean yield and soil penetration resistance data. Soil bulk density data, however, were only autocorrelated in the 0-0.10 m layer and not cross correlated with soybean yield. The results showed the higher efficiency of the autoregressive space-state models in relation to the equivalent simple and multiple linear regression models using Akaike's Information Criterion. The resulting values were comparatively lower than the values obtained by the regression models, for all combinations of explanatory variables.

Modeling changes in organic carbon stocks for distinct soils in southeastern brazil after four eucalyptus rotations using the century model

Soil organic matter (SOM) plays an important role in carbon (C) cycle and soil quality. Considering the complexity of factors that control SOM cycling and the long time it usually takes to observe changes in SOM stocks, modeling constitutes a very important tool to understand SOM cycling in forest soils. The following hypotheses were tested: (i) soil organic carbon (SOC) stocks would be higher after several rotations of eucalyptus than in low-productivity pastures; (ii) SOC values simulated by the Century model would describe the data better than the mean of observations. So, the aims of the current study were: (i) to evaluate the SOM dynamics using the Century model to simulate the changes of C stocks for two eucalyptus chronosequences in the Rio Doce Valley, Minas Gerais State, Brazil; and (ii) to compare the C stocks simulated by Century with the C stocks measured in soils of different Orders and regions of the Rio Doce Valley growing eucalyptus. In Belo Oriente (BO), short-rotation eucalyptus plantations had been cultivated for 4.0; 13.0, 22.0, 32.0 and 34.0 years, at a lower elevation and in a warmer climate, while in Virginópolis (VG), these time periods were 8.0, 19.0 and 33.0 years, at a higher elevation and in a milder climate. Soil samples were collected from the 0-20 cm layer to estimate C stocks. Results indicate that the C stocks simulated by the Century model decreased after 37 years of poorly managed pastures in areas previously covered by native forest in the regions of BO and VG. The substitution of poorly managed pastures by eucalyptus in the early 1970´s led to an average increase of C of 0.28 and 0.42 t ha-1 year-1 in BO and VG, respectively. The measured C stocks under eucalyptus in distinct soil Orders and independent regions with variable edapho-climate conditions were not far from the values estimated by the Century model (root mean square error - RMSE = 20.9; model efficiency - EF = 0.29) despite the opposite result obtained with the statistical procedure to test the identity of analytical methods. Only for lower soil C stocks, the model over-estimated the C stock in the 0-20 cm layer. Thus, the Century model is highly promising to detect changes in C stocks in distinct soil orders under eucalyptus, as well as to indicate the impact of harvest residue management on SOM in future rotations.

Modeling a processive molecular motor: the conventional kinesin

We present a model that allows for the derivation of the experimentally accesible observables: spatial steps, mean velocity, stall force, useful power, efficiency and randomness, etc. as a function of the [adenosine triphosphate] concentration and an external load F. The model presents a minimum of adjustable parameters and the theoretical predictions compare well with the available experimental results.

Preisach modeling of hysteresis for a pseudoelastic Cu-Zn-Al single crystal

Stress-strain trajectories associated with pseudoelastic behavior of a Cu¿19.4 Zn¿13.1 Al (at.%) single crystal at room temperature have been determined experimentally. For a constant cross-head speed the trajectories and the associated hysteresis behavior are perfectly reproducible; the trajectories exhibit memory properties, dependent only on the values of return points, where transformation direction is reverted. An adapted version of the Preisach model for hysteresis has been implemented to predict the observed trajectories, using a set of experimental first¿order reversal curves as input data. Explicit formulas have been derived giving all trajectories in terms of this data set, with no adjustable parameters. Comparison between experimental and calculated trajectories shows a much better agreement for descending than for ascending paths, an indication of a dissymmetry between the dissipation mechanisms operative in forward and reverse directions of martensitic transformation.

Cost-effective experimental design to support modeling of concentration-response functions

Modeling concentration-response function became extremely popular in ecotoxicology during the last decade. Indeed, modeling allows determining the total response pattern of a given substance. However, reliable modeling is consuming in term of data, which is in contradiction with the current trend in ecotoxicology, which aims to reduce, for cost and ethical reasons, the number of data produced during an experiment. It is therefore crucial to determine experimental design in a cost-effective manner. In this paper, we propose to use the theory of locally D-optimal designs to determine the set of concentrations to be tested so that the parameters of the concentration-response function can be estimated with high precision. We illustrated this approach by determining the locally D-optimal designs to estimate the toxicity of the herbicide dinoseb on daphnids and algae. The results show that the number of concentrations to be tested is often equal to the number of parameters and often related to the their meaning, i.e. they are located close to the parameters. Furthermore, the results show that the locally D-optimal design often has the minimal number of support points and is not much sensitive to small changes in nominal values of the parameters. In order to reduce the experimental cost and the use of test organisms, especially in case of long-term studies, reliable nominal values may therefore be fixed based on prior knowledge and literature research instead of on preliminary experiments

Symmetries and fixed point stability of stochastic differential equations modeling self-organized criticality

A stochastic nonlinear partial differential equation is constructed for two different models exhibiting self-organized criticality: the Bak-Tang-Wiesenfeld (BTW) sandpile model [Phys. Rev. Lett. 59, 381 (1987); Phys. Rev. A 38, 364 (1988)] and the Zhang model [Phys. Rev. Lett. 63, 470 (1989)]. The dynamic renormalization group (DRG) enables one to compute the critical exponents. However, the nontrivial stable fixed point of the DRG transformation is unreachable for the original parameters of the models. We introduce an alternative regularization of the step function involved in the threshold condition, which breaks the symmetry of the BTW model. Although the symmetry properties of the two models are different, it is shown that they both belong to the same universality class. In this case the DRG procedure leads to a symmetric behavior for both models, restoring the broken symmetry, and makes accessible the nontrivial fixed point. This technique could also be applied to other problems with threshold dynamics.

Analysis and spatial modeling of the indoor radon Swiss data

The present research deals with an important public health threat, which is the pollution created by radon gas accumulation inside dwellings. The spatial modeling of indoor radon in Switzerland is particularly complex and challenging because of many influencing factors that should be taken into account. Indoor radon data analysis must be addressed from both a statistical and a spatial point of view. As a multivariate process, it was important at first to define the influence of each factor. In particular, it was important to define the influence of geology as being closely associated to indoor radon. This association was indeed observed for the Swiss data but not probed to be the sole determinant for the spatial modeling. The statistical analysis of data, both at univariate and multivariate level, was followed by an exploratory spatial analysis. Many tools proposed in the literature were tested and adapted, including fractality, declustering and moving windows methods. The use of Quan-tité Morisita Index (QMI) as a procedure to evaluate data clustering in function of the radon level was proposed. The existing methods of declustering were revised and applied in an attempt to approach the global histogram parameters. The exploratory phase comes along with the definition of multiple scales of interest for indoor radon mapping in Switzerland. The analysis was done with a top-to-down resolution approach, from regional to local lev¬els in order to find the appropriate scales for modeling. In this sense, data partition was optimized in order to cope with stationary conditions of geostatistical models. Common methods of spatial modeling such as Κ Nearest Neighbors (KNN), variography and General Regression Neural Networks (GRNN) were proposed as exploratory tools. In the following section, different spatial interpolation methods were applied for a par-ticular dataset. A bottom to top method complexity approach was adopted and the results were analyzed together in order to find common definitions of continuity and neighborhood parameters. Additionally, a data filter based on cross-validation was tested with the purpose of reducing noise at local scale (the CVMF). At the end of the chapter, a series of test for data consistency and methods robustness were performed. This lead to conclude about the importance of data splitting and the limitation of generalization methods for reproducing statistical distributions. The last section was dedicated to modeling methods with probabilistic interpretations. Data transformation and simulations thus allowed the use of multigaussian models and helped take the indoor radon pollution data uncertainty into consideration. The catego-rization transform was presented as a solution for extreme values modeling through clas-sification. Simulation scenarios were proposed, including an alternative proposal for the reproduction of the global histogram based on the sampling domain. The sequential Gaussian simulation (SGS) was presented as the method giving the most complete information, while classification performed in a more robust way. An error measure was defined in relation to the decision function for data classification hardening. Within the classification methods, probabilistic neural networks (PNN) show to be better adapted for modeling of high threshold categorization and for automation. Support vector machines (SVM) on the contrary performed well under balanced category conditions. In general, it was concluded that a particular prediction or estimation method is not better under all conditions of scale and neighborhood definitions. Simulations should be the basis, while other methods can provide complementary information to accomplish an efficient indoor radon decision making.

Dosage optimization of treatments using population pharmacokinetic modeling and simulation.

Pharmacokinetic variability in drug levels represent for some drugs a major determinant of treatment success, since sub-therapeutic concentrations might lead to toxic reactions, treatment discontinuation or inefficacy. This is true for most antiretroviral drugs, which exhibit high inter-patient variability in their pharmacokinetics that has been partially explained by some genetic and non-genetic factors. The population pharmacokinetic approach represents a very useful tool for the description of the dose-concentration relationship, the quantification of variability in the target population of patients and the identification of influencing factors. It can thus be used to make predictions and dosage adjustment optimization based on Bayesian therapeutic drug monitoring (TDM). This approach has been used to characterize the pharmacokinetics of nevirapine (NVP) in 137 HIV-positive patients followed within the frame of a TDM program. Among tested covariates, body weight, co-administration of a cytochrome (CYP) 3A4 inducer or boosted atazanavir as well as elevated aspartate transaminases showed an effect on NVP elimination. In addition, genetic polymorphism in the CYP2B6 was associated with reduced NVP clearance. Altogether, these factors could explain 26% in NVP variability. Model-based simulations were used to compare the adequacy of different dosage regimens in relation to the therapeutic target associated with treatment efficacy. In conclusion, the population approach is very useful to characterize the pharmacokinetic profile of drugs in a population of interest. The quantification and the identification of the sources of variability is a rational approach to making optimal dosage decision for certain drugs administered chronically.