We describe anytime search procedures that (1) find disjoint subsets of recorded variables for which the members of each subset are d-separated by a single common unrecorded cause, if such exists; (2) return information about the causal relations among the latent factors so identified. We prove the procedure is point-wise consistent assuming (a) the causal relations can be represented by a directed acyclic graph (DAG) satisfying the Markov Assumption and the Faithfulness Assumption; (b) unrecorded variables are not caused by recorded variables; and (c) dependencies are linear. We compare the procedure with standard approaches over a variety of simulated structures and sample sizes, and illustrate its practical value with brief studies of social science data sets. Finally, we consider generalizations for non-linear systems.

Given a set of random variables, it is often the case that their associations can be explained by hidden common causes. We present a set of well-defined assumptions and a provably correct algorithm that allow us to identify some of such hidden common causes. The assumptions are fairly general and sometimes weaker than those used in practice by, for instance, econometricians, psychometricians, social scientists and in many other fields where latent variable models are important and tools such as factor analysis are applicable. The goal is automated knowledge discovery: identifying latent variables that can be used across diferent applications and causal models and throw new insights over a data generating process. Our approach is evaluated throught simulations and three real-world cases.

Security returns are heteroscedastic both cross-sectionally and over time, which affects the accuracy of standard factor extraction methods. In order to reduce the impact of such heterogeneity and to preserve the true factor structure, this paper studies the performance of a factor extracting method based on maximizing the explanatory power of the extracted factors. The implementation of the methodology is largely based on the principal components analysis on a correlation structure of asset returns. However, such a simple extension allows us to improve the finite sample performance over other popular approaches when returns are heteroscedastic both across individual assets and over time. Moreover, the out-of-sample study suggests that the extracted factors are not only stable across different sample groups, but also more pervasive in explaining the out-of-sample individual stock returns than other methods. These factors even have better out-of-sample explanatory power than the Fama and French factors. In addition, we shed light on the issue of choosing the correct number of factors.

The R package ltm has been developed for the analysis of multivariate dichotomous and polytomous data using latent variable models, under the Item Response Theory approach. For dichotomous data the Rasch, the Two-Parameter Logistic, and Birnbaum’s Three-Parameter models have been implemented, whereas for polytomous data Semejima’s Graded Response model is available. Parameter estimates are obtained under marginal maximum likelihood using the Gauss-Hermite quadrature rule. The capabilities and features of the package are illustrated using two real data examples.

Abstract not found

Abstract. The mining of association rules can provide relevant and novel information to the data analyst. However, current techniques do not take into account that the observed associations may arise from variables that are unrecorded in the database. For instance, the pattern of answers in a large marketing survey might be better explained by a few latent traits of the population than by direct association among measured items. Techniques for mining association rules with hidden variables are still largely unexplored. This paper provides a sound methodology for finding association rules of the type H ⇒ A1,..., Ak, where H is a hidden variable inferred to exist by making suitable assumptions and A1,..., Ak are discrete binary or ordinal variables in the database. 1

Please do not quote without permission

We estimate a status order for contemporary Norway using register data on married and cohabiting couples. By applying multidimensional scaling to contingency tables which cross-classify the occupation of spouses and cohabiting partners, we are able to extract a dimension which could reasonably be interpreted as reflecting social status in the classical Weberian sense. The general contour of the Norwegian status order and that of the UK are remarkably similar, as is the way in which social status relates to education, income and social class in the two countries. But social status is more equitably distributed in Norway than in the UK.

Abstract not found

Communication as a managerial competency – the “glue ” that keeps South African mainstream media newsrooms together? Author’s name, affiliation, contact details here Paper read in the Media Management and Economics Division of the Association for