The discursive turn in policy analysis and the validation of policy stories

This paper is concerned with the language of policy documents in the field of health care, and how ‘readings’ of such documents might be validated in the context of a narrative analysis. The substantive focus is on a comparative study of UK health policy documents (N=20) as produced by the various assemblies, governments and executives of England, Scotland, Wales and Northern Ireland during the period 2000-2009. Following an identification of some key characteristics of narrative structure the authors indicate how text-mining strategies allied with features of semantic and network analysis can be used to unravel the basic elements of policy stories and to facilitate the presentation of data in such a way that readers can verify the strengths (and weaknesses) of any given analysis – with regard to claims concerning, say, the presence, absence, or relative importance of key ideas and concepts. Readers can also ‘see’ how the different components of any one story might fit together, and to get a sense of what has been excluded from the narrative as well as what has been included, and thereby assess the reliability and validity of interpretations that have been placed upon the data.

A comparison of two measures of facial activity during pain in the newborn child

Facial activity is strikingly visible in infants reacting to noxious events. Two measures that reduce this activity to composite events, the Neonatal Facial Coding System (NFCS) and the Facial Action Coding System (FACS), were used to examine facial expressions of 56 neonates responding to routine heel lancing for blood sampling purposes. The NFCS focuses upon a limited subset of all possible facial actions that had been identified previously as responsive to painful events, whereas the FACS is a comprehensive system that is inclusive of all facial actions. Descriptions of the facial expressions obtained from the two measurement systems were very similar, supporting the convergent validity of the shorter, more readily applied system. As well, the cluster of facial activity associated with pain in this sample, using either measure, was similar to the cluster of facial activity associated with pain in adults and other newborns, both full-term and preterm, providing construct validity for the position that the face encodes painful distress in infants and adults.

Citizenship and Structured Dependency: the implications of policy design for senior political power

This paper argues that the structured dependency thesis must be extended to incorporate political power. It outlines a political framework of analysis with which to identify who gains and who loses from social policy. I argue that public policy for older people is a product not only of social structures but also of political decision-making. The Schneider and Ingram (1993) ‘ target populations’ model is used to investigate how the social construction of groups as dependent equates with lower levels of influence on policy making. In United Kingdom and European research, older people are identified as politically quiescent, but conversely in the United States seniors are viewed as one of the most influential and cohesive interest groups in the political culture. Why are American seniors perceived as politically powerful, while older people in Europe are viewed as dependent and politically weak? This paper applies the ‘target populations’ model to senior policy in the Republic of Ireland to investigate how theoretical work in the United States may be used to identify the significance of senior power in policy development. I conclude that research must recognise the connections between power, politics and social constructions to investigate how state policies can influence the likelihood that seniors will resist structured dependency using political means.

Statistical Tests for Joint Analysis of Performance Measures

Recently there has been an increasing interest in the development of new methods using Pareto optimality to deal with multi-objective criteria (for example, accuracy and architectural complexity). Once one has learned a model based on their devised method, the problem is then how to compare it with the state of art. In machine learning, algorithms are typically evaluated by comparing their performance on different data sets by means of statistical tests. Unfortunately, the standard tests used for this purpose are not able to jointly consider performance measures. The aim of this paper is to resolve this issue by developing statistical procedures that are able to account for multiple competing measures at the same time. In particular, we develop two tests: a frequentist procedure based on the generalized likelihood-ratio test and a Bayesian procedure based on a multinomial-Dirichlet conjugate model. We further extend them by discovering conditional independences among measures to reduce the number of parameter of such models, as usually the number of studied cases is very reduced in such comparisons. Real data from a comparison among general purpose classifiers is used to show a practical application of our tests.