A longitudinal study of newspaper and wire service coverage of nanotechnology risks

This study reviewed coverage of nanotechnology risks in 20 U.S. and 9 U.K. newspapers and 2 wire services from 2000 to 2009. It focused on information that citizens could come across in daily newspaper reading that could highlight the salience of these issues and alert readers to potential risks. Few articles about nanotechnology health, environmental, and societal risks were found in these publications during this period, averaging only 36.7 per year for both countries. The coverage emphasized three main narratives over time: runaway technology, science-based studies, and regulation. Health risks were covered most frequently, followed by environmental and societal risk issues. Regulation coverage was not as frequent but increased over time. The majority of the coverage focused on news events and 10 events drew modest media attention. Scientific uncertainty discussions appeared in about half of the articles, and scientists and engineers were the dominant information sources in both countries. Some significant differences between U.S. and U.K. coverage were found: U.K. coverage emphasized more societal concerns, while U.S. coverage paid more attention to environmental risks. Because the volume of coverage was not extensive and was counterbalanced by many more articles extolling nanotechnology's benefits, it is questionable whether this coverage alerted readers about potential nanotechnology risks. Coupled with citizens' minimal knowledge about nanotechnology, this type of coverage could create public distrust of nanotechnology applications should a dangerous risk event occur.     

Clustering objects on subsets of attributes (with discussion)

Summary.  A new procedure is proposed for clustering attribute value data. When used in conjunction with conventional distance-based clustering algorithms this procedure encourages those algorithms to detect automatically subgroups of objects that preferentially cluster on subsets of the attribute variables rather than on all of them simultaneously. The relevant attribute subsets for each individual cluster can be different and partially (or completely) overlap with those of other clusters. Enhancements for increasing sensitivity for detecting especially low cardinality groups clustering on a small subset of variables are discussed. Applications in different domains, including gene expression arrays, are presented.     

Profiles of behavioral problems in children who witness domestic violence

Unlike previous investigations of shelter-based samples, our study examined whether profiles of adjustment problems occurred in a community-program-based sample of 175 school-aged children exposed to domestic violence. Cluster analysis revealed three stable profiles/clusters. The largest cluster (69%) consisted of children below clinical thresholds for any internalizing or externalizing problem. Children in the next largest cluster (18%) were characterized as having externalizing problems with or without internalizing problems. The smallest cluster (13%) consisted of children with internalizing problems only. Comparison across demographic and violence characteristics revealed that the profiles differed by child gender, mother's education, child's lifetime exposure to violence, and aspects of the event precipitating contact with the community program. Clinical and future research implications of study findings are discussed.     

Measuring changes in interagency collaboration: an examination of the Bridgeport Safe Start Initiative

This paper describes the evaluation of interagency collaboration in a network of child-serving providers as part of the evaluation of the Bridgeport Safe Start Initiative (BSSI). In line with the system of care approach, the objectives of BSSI included reducing fragmentation of efforts and delivering integrated services to families of young children exposed to or at risk of exposure to family violence. Interagency collaboration was examined via social network and focus group data collected at three time points starting at baseline. Network analysis findings suggest that over time the network structure became consistent with BSSI's vision of an ideal collaborative network structure. Focus group findings, however, present a more complex picture of the status of collaboration. This paper sheds light on approaches and challenges to measuring interagency collaboration in a service delivery system and communicating social network analysis findings to stakeholders in a way that is accessible and useful.     

THE EFFECTS OF ENDORSEMENTS ON PRODUCT EVALUATION

This paper examines the effect of product endorsement claims on a consumer's decision making processes. A simulated endorsement claim was made for a product after which a consumer was presented with the product. Endorsements tested were the three most common types in use: celebrity, expert, and typical consumer, with two control groups for experimental design. Three scales of measurement were used to assess the effect of the endorsement: a consumer's intent to purchase, his overall attitude towards the product, and his expected price of the product. The results show the strength of celebrity and typical consumer endorsers. They also show that the endorser effect is dramatically weaker than hypothesized. The Lavidge-Steiner hierarchy of advertising effects model is introduced in an attempt to explain this phenomenon. Guidelines for further research are provided.     

On bagging and nonlinear estimation

We propose an elementary model for the way in which stochastic perturbations of a statistical objective function, such as a negative log-likelihood, produce excessive nonlinear variation of the resulting estimator. Theory for the model is transparently simple, and is used to provide new insight into the main factors that affect performance of bagging. In particular, it is shown that if the perturbations are sufficiently symmetric then bagging will not significantly increase bias; and if the perturbations also offer opportunities for cancellation then bagging will reduce variance. For the first property it is sufficient that the third derivative of a perturbation vanish locally, and for the second, that second and fourth derivatives have opposite signs. Functions that satisfy these conditions resemble sinusoids. Therefore, our results imply that bagging will reduce the nonlinear variation, as measured by either variance or mean-squared error, produced in an estimator by sinusoid-like, stochastic perturbations of the objective function. Analysis of our simple model also suggests relationships between the results obtained using different with-replacement and without-replacement bagging schemes. We simulate regression trees in settings that are far more complex than those explicitly addressed by the model, and find that these relationships are generally borne out.