Search

In this article we introduce randomized repeated significance tests (RSTs), based on scan statistics, for detecting a local change in a parameter of a distribution for one and two dimensional discrete data. When the size of the window where a local change in the parameter has occurred is known, a randomized RST based on a fixed window scan statistic is proposed. When the size of the window where a local change in the parameter has occurred is unknown, a randomized RST based on the minimum P-value scan statistic is developed. Simulation based methods are used to implement these randomized RSTs. Numerical results for one and two dimensional data, generated from Bernoulli and Poisson distributions, for selected values of model parameters, demonstrate the effectiveness of the randomized RSTs in detecting a local change in the parameter of the respective model. [ABSTRACT FROM AUTHOR]

Purpose: Taking into account the large area of citrus in Spain and the impacts generated by agriculture, reducing the environmental impact of this crop represents an important goal. This study attempts to compare the environmental impact of two citrus cropping systems, organic and conventional, in the region of Valencia (Spain), and to assess the variability within both farming systems in order to highlight the influence of management practices on the environmental performance. Methods: A survey was carried out on citrus farmers, 145 corresponding to organic production and 122 to conventional. Life cycle assessment (LCA) was used to estimate the environmental impacts of farms and the contribution of each production stage to impacts. Two functional units (FUs), mass- and area-based, were chosen. The variability and confidence intervals of the average impact results were assessed by means of a bootstrap technique. Finally, a k-means cluster analysis was performed to identify groups of farms with comparable impact levels. Results and discussion: The mean impact values of the conventional farm sample were higher than those of the organic farms, when using 1 ha year as FU, whereas for the FU of 1 kg no differences were found for some impact categories. Most of the impact results were also observed to be highly variable. The distribution of the mean after the bootstrap confirmed both the variability of the impacts and the differences between the farming systems. The cluster analysis determined two groups via their impact categories. Cluster-1, which showed higher impacts, was made up of part of the conventional farms while cluster-2 included the remaining conventional farms and all the organic ones. No difference in yield was found between the conventional farms of both clusters. Conclusions: Bootstrapped LCA offers the ability to assess the variability of the impacts, regardless of the sample size and the non-normal impact distributions. The cluster analysis shows that conventional farms can attain similar impacts than the organic ones, while maintaining the yield. FU selection is crucial, since a mass-based FU reduces the difference in the environmental performance between conventional and organic farms. To attain a more sustainable citrus farming, a careful selection of the management practices is needed. [ABSTRACT FROM AUTHOR]

This work assumes that the small area quantities of interest follow a Fay–Herriot model with spatially correlated random area effects. Under this model, parametric and nonparametric bootstrap procedures are proposed for estimating the mean squared error of the empirical best linear unbiased predictor (EBLUP). A simulation study based on the Italian Agriculture Census 2000 compares bootstrap and analytical estimates of the MSE and studies their robustness to non-normality. Results indicate lower bias for the non-parametric bootstrap under specific departures from normality. [ABSTRACT FROM AUTHOR]

In this paper we suggest a simple graphical device for assessing multivariate normality. The method is based on the characteristic that linear combinations of the sample mean and sample covariance matrix are independent if and only if the random variable is normally distributed. We demonstrate the usage of the suggested method and compare it to the classical Q-Q plot by using some multivariate data sets. [ABSTRACT FROM AUTHOR]

To improve the reduction of metrological data, that are typically grouped in series and cannot be considered as replicated data, a modelling procedure has been obtained by adding to the model representing the physical behaviour, common to all data, a specific term for each series. Such a procedure combines both the advantages of preserving the individuality of each series and of improving the variance estimate which arises from fitting the overall data. A non-parametric bootstrap method for the error analysis has been developed, which does not imply the assumption of the Normal distribution in the least squares estimation. Two examples of application of the method to thermodynamic data series are reported. [ABSTRACT FROM AUTHOR]