Change and stability of migration intentions. Evidence from Italy

Due to a lack of relevant data, very few empirical studies have examined the changes in and stability of secondary migration intentions. We aim to fill this gap by analysing return migration intentions among international migrants in Italy. Data are drawn from the cross-sectional SCIF survey conducted by ISTAT in 2011–2012. Our findings reveal that migration intentions at the beginning of the migratory experience tend to differ from those measured at more advanced migration stages (i.e. at the time of the survey). In particular, intentions to return seem less stable than intentions to stay. When confirming intentions to return or remain, critical factors include financial stability, family situation and ties with the country of origin and destination. Additionally, having an Italian partner, a partner living in Italy, and a positive self-assessed family financial condition are positively associated with transitioning from a temporary plan to a permanent settlement intention.     

The independent factor analysis approach to latent variable modelling

Independent factor analysis (IFA) has recently been proposed in the signal processing literature as a way to model a set of observed variables through linear combinations of latent independent variables and a noise term. A peculiarity of the method is that it defines a probability density function for the latent variables by mixtures of Gaussians. The aim of this paper is to cast the method into a more rigorous statistical framework and to propose some developments. In the first part, we present the IFA model in its population version, address identifiability issues and draw some parallels between the IFA model and the ordinary factor analysis (FA) one. Then we show that the IFA model may be reinterpreted as an independent component analysis-based rotation of an ordinary FA solution. We also give evidence that the IFA model represents a special case of mixture of factor analysers. In the second part, we address inferential issues, also deriving the standard errors for the model parameter estimates and providing model selection criteria. Finally, we present some empirical results on real data sets.     

Model Assessment for Predictive Classification Models

In this article, we present a novel methodology to assess predictive models for a binary target. In our opinion, the main weakness of the criteria proposed in the literature is not to take the financial costs of a wrong decision into account.

The objective of this article is to derive the optimal cut-off in predictive classification models and to improve model assessment on the basis of a general class of loss functions. We describe how our proposal performs in a real application on credit scoring.     

Fiducial distribution in a power series family

Abstract

In this article the interest is on finding the fiducial distribution of the parameter, when the probability distribution belongs to the power series family, as in Johnson et al. (1992 Johnson, N. L., S. Kotz, and A. W. Kemp. 1992. Univariate discrete distributions. New York, NY: John Wiley and Sons. [Google Scholar]). Recently in Nájera and O’Reilly (2017 Nájera, E., and F. O’Reilly. 2017. On fiducial generators. Communications Statistics - Theory Methods 46 (5):2232–2248. doi:10.1080/03610926.2015.1040505.[Taylor & Francis Online], [Web of Science ®] , [Google Scholar]) an argument is given to obtain a unique fiducial in the Bernoulli case. An attempt is made here to define some sort of invariance in a power series distribution so that, as was done in the Bernoulli case, one may find a unique invariant fiducial for the parameter. The Bernoulli case is reviewed in detail and the Poisson and negative binomial cases are addressed.     

Nonparametric rank-based test procedures for non-additive models in the two-way layout I. no replications

One of the major unresolved problems in the area of nonparametric statistics is the need for satisfactory rank-based test procedures for non-additive models in the two-way layout, especially when there is only one observation on each combination of the levels of the experimental factors. In this paper we consider an arbitrary non-additive model for the two-way layout with n levels of each factor. We utilize both alignment and ranking of the data together with basic properties of Latin squares to develop rank tests for interaction (non-additivity). Our technique involves first aligning within one of the main effects, ranking within the other main effects (columns and rows) and then adding the resulting ranks within “interaction bands” corresponding to orthogonal partitions of the interaction for the model, as denoted by the letters of an n × n Latin square. A Friedman-type statistic is then computed on the resulting sums. This is repeated for each of (n?1) mutually orthogonal Latin squares (thus accounting for all the interaction degrees of freedom). The resulting (n?1) Friedman-type statistics are finally combined to obtain an overall test statistic. The necessary null distribution tables for applying the proposed test for non-additivity are presented and we discuss the results of a Monte Carlo simulation study of the relative powers of this new procedure and other (parametric and nonparametric) procedures designed to detect interaction in a two-way layout with one observation per cell.