Jouni Kuha

AIC and BIC Comparisons of Assumptions and Performance

The two most commonly used penalized model selection criteria, the Bayesian information criterion (BIC) and Akaike’s information criterion (AIC), are examined and compared. Their motivations as approximations of two different target quantities are discussed, and their performance in estimating those quantities is assessed. Despite their different foundations, some similarities between the two statistics can be observed, for example, in analogous interpretations of their penalty terms. The behavior of the criteria in selecting good models for observed data is examined with simulated data and also illustrated with the analysis of two well-known data sets on social mobility. It is argued that useful information for model selection can be obtained from using AIC and BIC together, particularly from trying as far as possible to find models favored by both criteria.

Incidence of sciatic pain among men in machine operating, dynamic physical work, and sedentary work. A three-year follow-up.

The incidence of sciatic pain among 25-49 year-old men in three different types of work: machine operating, dynamic physical work (construction carpenters), and office work was assessed prospectively. Postal questionnaires both at the beginning and the end of the 3-year follow-up were answered by 1149 men who at baseline had no history of sciatic pain. The crude risk ratio of the incidence of sciatic pain was 1.6 (95% confidence interval 1.2-2.2) for the machine operators and 1.7 (1.3-2.4) for the carpenters when the office workers were referents. The adjusted risk ratios were 1.4 (1.0-1.9) and 1.5 (1.1-2.1), respectively. Previous history of severe lumbago or other low-back pain increased the risk fourfold. Frequent physical exercise and smoking were of borderline significance as predictors.

Developing European indicators of trust in justice

A social indicators approach to trust in justice recognizes that the police and criminal courts need public support and institutional legitimacy if they are to operate effectively and fairly. In order to generate public cooperation and compliance, these institutions must demonstrate to citizens that they are trustworthy and that they possess the authority to govern. In this paper we first outline the conceptual roadmap for a current comparative analysis of trust in justice. We then describe the methodological development process of a 45-item module in Round 5 of the European Social Survey, which fields the core survey indicators. After presenting the findings from a quantitative pilot of the indicators, we consider the policy implications of a procedural justice model of criminal justice.

Ethnic diversity, segregation, and the social cohesion of neighbourhoods in London

The question of whether and how ethnic diversity affects the social cohesion of communities has become an increasingly prominent and contested topic of academic and political debate. In this paper we focus on a single city: London. As possibly the most ethnically diverse conurbation on the planet, London serves as a particularly suitable test-bed for theories about the effects of ethnic heterogeneity on prosocial attitudes. We find neighbourhood ethnic diversity in London to be positively related to the perceived social cohesion of neighbourhood residents, once the level of economic deprivation is accounted for. Ethnic segregation within neighbourhoods, on the other hand, is associated with lower levels of perceived social cohesion. Both effects are strongly moderated by the age of individual residents: diversity has a positive effect on social cohesion for young people but this effect dissipates in older age groups; the reverse pattern is found for ethnic segregation.

ESTIMATION BY DATA AUGMENTATION IN REGRESSION MODELS WITH CONTINUOUS AND DISCRETE COVARIATES MEASURED WITH ERROR

Estimation methods are considered for regression models which have both misclassified discrete covariates and continuous covariates measured with error. Adjusted parameter estimates are obtained using the method of data augmentation, where the true values of the covariates measured with error are regarded as missing data. Validation data on the covariates are assumed to be available. The distinction between internal and external validation data is emphasized, and its effects on the analysis are examined. The method is illustrated with simulated data.

A Nonparametric Approach to Matched Pairs with Missing Data

The matched-pairs t-statistic on the overall ranks is extended to data with observations missing at random. Either one of the two variables is allowed to be missing. The procedure is completely nonparametric. Comparisons with the likelihood ratio test for normal data indicate that the proposed method fares well when the data are normal and outperforms it in other cases. Simulations also confirm that the proposed method has higher power than common nonparametric complete-pairs tests for observations missing completely at random. Finally, a data set on the delinquent values of boys released from correctional institutions is analyzed and discussed.

Confounding the commentators : how the 2010 exit poll got it (more or less) right

Abstract An exit poll conducted on behalf of the three main UK broadcasting organizations for the 2010 general election predicted that the Liberal Democrats would win fewer seats than in 2005, a suggestion that was met with widespread disbelief amongst commentators. Not only did this forecast prove correct, but the poll’s prediction that the Conservatives would be the largest party in a hung parliament with 307 seats also proved to be spot on. This paper describes and evaluates the methodology of the poll. The key features were a focus on changes in party vote shares facilitated by conducting the poll in the same places as at the previous election, modelling of the variation in those estimated changes, and a probabilistic approach to identifying the likely outcome in seats. All three features contributed to the accuracy of the 2010 poll. The data provided a highly accurate estimate of the change in the Liberal Democrat vote. Meanwhile, a slight overestimation of the swing from Labour to the Conservatives was corrected by the modelling of changes in vote shares and the use of a probabilistic approach to seat prediction.

Empirical Legitimacy as Two Connected Psychological States

In this chapter we consider the idea that legitimacy judgments involve two connected beliefs. The first relates to consent and authorization: do people believe that an authority has the right to dictate appropriate behavior? The second relates to moral validity: do people believe that an authority exercises its power in ways that accord with prevailing norms of appropriate conduct? Marshalling data from Round 5 of the European Social Survey, we first assess the scaling properties of measures of police legitimacy using data from the UK. We then examine the utility of three different ways of representing legitimacy within in a larger model of public cooperation with the police.

On the index of dissimilarity for lack of fit in loglinear and log-multiplicative models

The index of dissimilarity, often denoted by Delta, is commonly used, especially in social science and with large datasets, to describe the lack of fit of models for categorical data. The definition and sampling properties of the index for general loglinear and log-multiplicative models are investigated. It is argued that in some applications a standardized version of the index is appropriate for interpretation. A simple, approximate variance formula is derived for the index, whether standardized or not. A simple bias reduction formula is also given. The accuracy of these formulae and of confidence intervals based upon them is investigated in a simulation study based on large-scale social mobility data.

Trust and Legitimacy Across Europe: A FIDUCIA Report on Comparative Public Attitudes Towards Legal Authority

FIDUCIA (New European Crimes and Trust-based Policy) seeks to shed light on a number of distinctively ‘new European’ criminal behaviours which have emerged in the last decade as a consequence of both technology developments and the increased mobility of populations across Europe. A key objective of FIDUCIA is to propose and proof a ‘trust-based’ policy model in relation to emerging forms of criminality – to explore the idea that public trust and institutional legitimacy are important for the social regulation of the trafficking of human beings, the trafficking of goods, the criminalisation of migration and ethnic minorities, and cybercrimes. In this paper we detail levels of trust and legitimacy in the 26 countries, drawing on data from Round 5 of the European Social Survey. We also conduct a sensitivity analysis that investigates the effect of a lack of measurement equivalence on national estimates.