Multivariate forests with missing mixed outcomes

In this article, we propose a multivariate random forest method for multiple responses of mixed types with missing responses. Imputation is performed for each bootstrap sample used to build the individual trees that form the forest. The individual trees are built using a weighted splitting rule allowing downweighting of imputed observations. A simulation study shows the benefits of this approach over complete case analysis when missing responses are missing completely at random and missing at random (MAR). In particular, the gain in prediction accuracy of the proposed method is larger in the MAR case and also increases as the proportion of missing increases.     

Adolescent Electronic Media Use: Instead of Doing What?

Adolescents like to watch TV and listen to the radio, even on school days, but their day contains, like for all other people, only 24 hours. What do those who use electronic media frequently neglect? Data from 14- and 16-year-old adolescents from 12 European countries and from the USA (EURONET sample) were analyzed (N = 3,250). The correlations between time spent using electronic media (EM) and 15 other daily activities were all negative, which is not surprising given the fact that the sum of all activities is bound to be 24 hours. We, therefore, calculated correlations between electronic media use and the proportions of each other activity within the remaining time. A clear pattern emerged. The method allowed us to identify “mandatory” activities that always have high priority, such as sleeping and attending school, and “optional” activities that enter into competition with electronic media use, including engaging in sports, playing music, leisure reading, and being with friends without media use. As to the mandatory activities, the correlations between media-using time and the proportions of remaining time were positive; relative to the optional activities, the correlations between media-using time and the proportions of remaining time were negative. While one might expect school homework to be regarded as mandatory, the correlation patterns identified it as optional. As to the average electronic media-using times, there were marked differences not only between the individual participants, but also between the countries. While Bulgarian adolescents spent an average of 2.37 hours of a regular school day using electronic media, French adolescents averaged only 1.72 hours.     

Behavioral evidence for framing effects in the resolution of the doctrinal paradox

The aggregation of individual sets of judgments over interconnected propositions can yield inconsistent collective sets of judgments, even when the individual sets of judgments are themselves consistent. A doctrinal paradox occurs when majority voting on a compound proposition (such as a conjunction or disjunction) yields a different result than majority voting on each of the elements of the proposition. For example, when most individuals accept proposition X; most individuals accept proposition Y ; but only a minority of individuals accept the compound proposition ‘X and Y’. Conducting two elemental votes would lead to accept X and Y , but conducting one compound vote would lead to reject X and Y . In such a situation, do people manifest a stable preference as to which voting procedure should be applied? This research reports the results of two behavioral experiments using a within-participant design, which show that procedural preferences can be upturned by framing either positively or negatively the set of judgments to be aggregated. This shift in procedural preference leads to large swings in the final collective judgment endorsed by participants.     

Grandchild Care Transfers by Ageing Immigrants in France: Intra-household Allocation and Labour Market Implications

Using a unique sample on ageing first generation immigrants in France, we explore the distribution of grandchild care across offspring and the impact of this transfer on the labour supply of the mother. Our results indicate that grandchild care is spread unequally across siblings, and this inequality is driven more strongly by better labour market potential than by weaker financial status of the recipient. There is further evidence that even after accounting for potential endogeneity and unobserved heterogeneity, grandchild care has a positive impact on the labour supply of the mother. The pattern of intergenerational time transfers differs across people with different national and religions backgrounds.     

Environmental Sustainability and Social Work: A Rural Australian Evaluation of Incorporating Eco-Social Work in Field Education

Climate change poses significant threat to the wellbeing of global society. Addressing this change has as yet generated no fixed blueprint for social work practice and education. This paper reports on a formative evaluation of one Australian initiative to address this transformative opening in social work field education. Prompted by service users' and workers' experience of the impact of drought, a rurally located social work course team amended the field education curriculum to include a focus on Environment and Sustainability. This learning goal was added to the existing learning goals derived from the Australian Association of Social Workers (AASW) Practice Standards. Students and field supervisors were surveyed on their experience of meeting this new learning goal. While most expressed confidence in understanding the concepts involved, they clearly lacked assurance in interpreting these in practice encounters. Considering their qualitative input suggests that this topic is making a transition from being on the margins of social work to becoming mainstream. Their open-ended responses indicate that the incorporation of environmental sustainability into practice is at a threshold stage of development. Further enactment of eco-social work at the local level is concluded to be supported by using a transformative learning framework in facilitating critical reflection and collaborative dialogue for effective change.     

The impact of risk aversion and ambiguity aversion on annuity and saving choices

Journal of Risk and Uncertainty - We analyze the impact of risk aversion and ambiguity aversion on the competing demands for annuities and bequeathable savings using a lifecycle recursive utility...     

Five legitimate definitions of correlated equilibrium in games with incomplete information

Aumann's (1987) theorem shows that correlated equilibrium is an expression of Bayesian rationality. We extend this result to games with incomplete information.First, we rely on Harsanyi's (1967) model and represent the underlying multiperson decision problem as a fixed game with imperfect information. We survey four definitions of correlated equilibrium which have appeared in the literature. We show that these definitions are not equivalent to each other. We prove that one of them fits Aumann's framework; the agents normal form correlated equilibrium is an expression of Bayesian rationality in games with incomplete information.We also follow a universal Bayesian approach based on Mertens and Zamir's (1985) construction of the universal beliefs space. Hierarchies of beliefs over independent variables (states of nature) and dependent variables (actions) are then constructed simultaneously. We establish that the universal set of Bayesian solutions satisfies another extension of Aumann's theorem.We get the following corollary: once the types of the players are not fixed by the model, the various definitions of correlated equilibrium previously considered are equivalent.     

Conspicuous conservatism in risk choice

We analyze the risk levels chosen by agents who have private information regarding their quality, and whose performance will be judged and rewarded by outsiders. Assume that risk choice is observable. Agents will choose risk strategically to enhance their expected reputations. We show that conspicuous conservatism results: agents of different qualities choose levels below those that would be chosen if quality were observable. This happens because bad agents must cloak their identity by choosing the same risk level as good agents, and good agents are more likely to distinguish themselves if they reduce the risk level. Our results contrast starkly with those for the case when risk choice cannot be observed.

Variance estimation when donor imputation is used to fill in missing values

Donor imputation is frequently used in surveys. However, very few variance estimation methods that take into account donor imputation have been developed in the literature. This is particularly true for surveys with high sampling fractions using nearest donor imputation, often called nearest‐neighbour imputation. In this paper, the authors develop a variance estimator for donor imputation based on the assumption that the imputed estimator of a domain total is approximately unbiased under an imputation model; that is, a model for the variable requiring imputation. Their variance estimator is valid, irrespective of the magnitude of the sampling fractions and the complexity of the donor imputation method, provided that the imputation model mean and variance are accurately estimated. They evaluate its performance in a simulation study and show that nonparametric estimation of the model mean and variance via smoothing splines brings robustness with respect to imputation model misspecifications. They also apply their variance estimator to real survey data when nearest‐neighbour imputation has been used to fill in the missing values. The Canadian Journal of Statistics 37: 400–416; 2009 © 2009 Statistical Society of Canada