On Variable Importance in Linear Regression

The paper examines in detail one particular measure of variable importance for linear regression that was theoretically justified by Pratt (1987), but which has since been criticized by Bring (1996) for producing “counterintuitive” results in certain situations, and by other authors for failing to guarantee that importance be non-negative. In the article, the “counterintuitive” result is explored and shown to be a defensible characteristic of an importance measure. It is also shown that negative importance of large magnitude can only occur in the presence of multicollinearity of the explanatory variables, and methods for applying Pratt's measure in such cases are described. The objective of the article is to explain and to clarify the characteristics of Pratt's measure, and thus to assist practitioners who have to choose from among the many methods available for assessing variable importance in linear regression.     

How can business cope with terrorism?

“What can business do to cope more successfully with terrorism?” The policy against terrorism available to business is a neglected issue in the scholarly literature especially in so far as individual firms rather than the business sector as a whole are concerned.Two sets of proposals are advanced, based on an economic analysis of terrorism. The first set discusses possibilities to reduce terrorists’ incentives to attack business premises; the second part outlines proposals designed to minimize the costs to businesses once a terrorist attack has taken place, hence reducing the impact.     

Unemployment, Wage Pressure and Sectoral Shifts: Permanent and Temporary Consequences of Intersectoral Shocks

The main purpose of this study is to examine the effect of allocative disturbances using Italian labour market data and a VECM (vector error correction model). In order to measure sectoral shifts in labour demand, we use Neumann-Topel's [Neumann, G.R., & Topel, R.H. (1991). Employment risk, diversification, and unemployment. Quarterly Journal of Economics 106:1341–1365] employment-based dispersion index. In searching for a restricted cointegration space, we set down, and identify a stylized version of the battle of the mark-ups model. Lilien [Lilien, D. (1982). Sectoral shifts and cyclical unemployment. Journal of Political Economy 90:777–793] argued that an important part of cyclical unemployment comes from permanent shifts in the composition of labour demand. Our findings emphasize that an important source of the persistence of unemployment comes from sectoral shifts.     

Bayesian non-parametric modeling for integro-difference equations

Integro-difference equations (IDEs) provide a flexible framework for dynamic modeling of spatio-temporal data. The choice of kernel in an IDE model relates directly to the underlying physical process modeled, and it can affect model fit and predictive accuracy. We introduce Bayesian non-parametric methods to the IDE literature as a means to allow flexibility in modeling the kernel. We propose a mixture of normal distributions for the IDE kernel, built from a spatial Dirichlet process for the mixing distribution, which can model kernels with shapes that change with location. This allows the IDE model to capture non-stationarity with respect to location and to reflect a changing physical process across the domain. We address computational concerns for inference that leverage the use of Hermite polynomials as a basis for the representation of the process and the IDE kernel, and incorporate Hamiltonian Markov chain Monte Carlo steps in the posterior simulation method. An example with synthetic data demonstrates that the model can successfully capture location-dependent dynamics. Moreover, using a data set of ozone pressure, we show that the spatial Dirichlet process mixture model outperforms several alternative models for the IDE kernel, including the state of the art in the IDE literature, that is, a Gaussian kernel with location-dependent parameters.     

Risk Perception and Personality Facets

The present study examined the relationship between personality facets and risk perception using the Big Five model. A broad range of hazards was considered: energy production, pollutants, sex, deviance, addictions, weapons, common individual hazards, outdoor activities, medical care, and psychotropic drugs. Key personality facets that were most predictive of risk perception compared to (or in association with) age, gender, educational level, and personality factors were identified. They were moderation and tranquility (associated with energy production or pollutants), rationality and efficiency (associated with pollutants, sex, deviance, addictions, or weapons), creativity, imagination, and reflection (associated with energy production, pollutants, or common individual hazards), self-disclosure (associated with outdoor activities), and nurturance and tenderness (associated with sex, deviance, addictions, or medical care). These facets may be recommended for use in future studies on risk perception.     

Criminal Victimization and the Quality of Life

The aim of this investigation was to explain theimpact of crime-related issues on satisfaction withthe quality of life, satisfaction with life as a wholeand happiness in the city of Prince George, BritishColumbia. As explanatory variables, we had measures ofrespondent fears of and actual cases of victimization, Indexes of Neighbourhood Problems, PolicePerformance, Neighbourhood Worries, DefensiveBehaviour, beliefs about increases in local crime,satisfaction with personal and family safety, andsatisfaction with a variety of domains of life (e.g.,friendships, financial security, health). Collectivelysuch variables could explain only 5% of the variationin happiness scores, 7% of the variation in lifesatisfaction scores and 9% of the variation insatisfaction with the quality of life scores. However,they could explain 38% of the variation in overallneighbourhood satisfaction scores. When measures ofsatisfaction with family life, health, self-esteem,etc. were added, we found that crime related issueswere simply displaced by the other measures and thatwe could explain 31% of the variation in overallhappiness scores, 58% of the variation in lifesatisfaction scores and 59% of the variation insatisfaction with the overall quality of life scores. We conclude, therefore, that crime-related issues haverelatively little impact on peoples satisfaction withthe quality of their lives, with life satisfaction orhappiness here.     

Testosterone deficiency and the metabolic syndrome

Evidence is presented to link components of the metabolic syndrome to testosterone deficiency and obesity. Testosterone deficiency in hypogonadism or testosterone deprivation in normo-gonadotropic men increases fat mass as well as fasting insulin levels. Testosterone supplementation (TS) in a dose dependent manner, increase lean body mass (LBM), reduces fat mass, body mass index (BMI) and waist hip ratio in both young and elderly hypogonadal men. A negative association between T and insulin resistance as well as impaired glucose intolerance has been demonstrated and in type 2 diabetic men TS improves metabolic parameters. TS improves most components of the metabolic syndrome and also reduces inflammatory cytokines.     

Differential Reliance on the Duchenne Marker During Smile Evaluations and Person Judgments

When evaluating the smiles of other people (regarding amusement, authenticity, spontaneity, or intensity), perceivers typically rely on Orbicularis oculi activity that causes wrinkles around a target’s eyes. But does this so-called Duchenne marker also impact more generalized judgments of person characteristics (e.g., regarding a target’s attractiveness, intelligence, dominance, and trustworthiness)? To address this issue, the current study asked participants to provide the above smile evaluations and person judgments for a series of Duchenne and non-Duchenne smiles. The results showed that smile evaluations uniformly increased during Duchenne marker presence. The marker’s effect on person judgments, in contrast, was judgment dependent. While attractiveness, dominance and intelligence ratings showed the expected enhancement, trustworthiness ratings remained unaffected by the facial cue of interest. The findings suggest that the Duchenne marker’s role as a cue of social relevance during target perception depends on the type of person inference under consideration.     

Public Services and the Quality of Life

This is a report of the results of a survey of citizen beliefs and attitudes about public services and the quality of life in Prince George, British Columbia, Canada in the summer of 1997. Information is provided about the perceived frequency with which various services were used, the perceived satisfaction and value for tax dollars spent on the services as well as on levels of government officials generally, preferences for the provision of more or fewer services and for spending relatively more or less revenue on different services, views about user-fees, and views about smoke-free public places and the likely impact of bylaw changes on people's behaviour. Using such information, we examined correlations among perceived satisfaction, perceived value for money, use, spending preferences and demand, and, using multiple regression analysis, explained 66% of the variance in life satisfaction scores, 57% of the variance in satisfaction with the quality of life scores and 37% of the variance in happiness scores. Applying LISREL 8.14, it was shown that a model in which our three global indicators were explained by 13 domain indicators was superior to a model in which the latter indicators were explained by the former, i.e., a Bottom-Up model was superior to a Top-Down model. A simple linear model was also used to explain 32%, 20% and 19%, respectively, of the variance in satisfaction with municipal, provincial and federal government officials.