Cinzia Carota

Bayesian nonparametric disclosure risk estimation via mixed effects log-linear models

Statistical agencies and other institutions collect data under the promise to protect the confidentiality of respondents. When releasing microdata samples, the risk that records can be identified must be assessed. To this aim, a widely adopted approach is to isolate categorical variables key to the identification and analyze multi-way contingency tables of such variables. Common disclosure risk measures focus on sample unique cells in these tables and adopt parametric log-linear models as the standard statistical tools for the problem. Such models have often to deal with large and extremely sparse tables that pose a number of challenges to risk estimation. This paper proposes to overcome these problems by studying nonparametric alternatives based on Dirichlet process random effects. The main finding is that the inclusion of such random effects allows us to reduce considerably the number of fixed effects required to achieve reliable risk estimates. This is studied on applications to real data, suggesting in particular that our mixed models with main effects only produces roughly equivalent estimates compared to the all-two way interactions models, and is effective in defusing potential shortcomings of traditional log-linear models. This paper adopts a fully Bayesian approach that accounts for all sources of uncertainty, including that about the population frequencies, and supplies unconditional (posterior) variances and credible intervals.

Some faults of the Bayes factor in nonparametric model selection

We compare different Bayesian strategies for testing a parametric model versus a nonparametric alternative on the ground of their ability to solve the inconsistency problems arising when using the Bayes factor under certain conditions. A preliminary critical discussion of such an inconsistency is provided.

An Application of Graphical Models to the Innobarometer Survey: A Map of Firms’ Innovative Behaviour

Probabilistic graphical models successfully combine probability with graph theory and therefore provide applied statisticians with a powerful data mining engine. Graphical models are a good framework for formal analysis, allowing the researcher to obtain a quick overview of the structure of association among variables in a system. This paper is the first attempt to apply high-dimensional graphical models in innovation studies, since the i ncreasing availability of data in the field and the complexity of the underlying processes are calling for new techniques which can handle not only a large amount of observations, but also rich datasets in terms of number and relations among variables. In this context, the process of variables and model selection became more arduous, influenced by biases of the scientist and, in the worst case scenario, subject to scientific malpractices such as the p-hacking behavior. On the contrary, high-dimensional graphical models allow for bottom-up, hypotheses free, data-driven, and see-through approach.

A diagnostic for autocorrelation of the disturbances in regression models

Consider a standard regression model whose disturbances are Gaussian white noise and embed it in a larger model with error terms generated by a first-order autoregressive process. We derive a Bayesian diagnostic of the adequacy of the standard model where the autoregressive parameter, sayr, is zero against the altermative thatr differs from zero. It is shown that it is closely related to the Bayes factor and the Durbin-Watson statistic for the same hypotheses. This formal result is taken as a starting point for some applications where the Bayes factor and the Durbin-Watson statistic play an auxiliary role.

Bayesian Methods for Microsimulation Models

This article proposes Bayesian methods for microsimulation models and for policy evaluations. In particular, the Bayesian Multinomial Logit and the Bayesian Multinomial Mixed Logit models are presented. They are applied to labour-market choices by single females and single males, enriched with EUROMOD microsimulated information, to evaluate fiscal policy effects. Estimates using the two Bayesian models are reported and compared to the results stemming from a standard approach to the analysis of the phenomenon under consideration. Improvements in model performances, when Bayesian methods are introduced and when random effects are included, are outlined. Finally, ongoing work, based on nonparametric model extensions and on analysis of work choices by couples is briefly described.