Using the Delphi process to analyze social policy implementation: A post hoc case from vocational rehabilitation

This study uses a Policy Delphi to discern differences in perspective among and within groups responsible for formulating and implementing vocational rehabilitation policy. Four groups of players were chosen for our analysis: government officials, academics, directors of rehabilitation centers, and the staff who interface with program participants. Significant differences were found between the groups regarding the relative importance of possible legislative goals. This suggests that the failure of vocational rehabilitation policy to promote a work agenda may be attributed to a lack of consensus among policy implementors. The Delphi technique could help policy planners understand the different perspectives within the implementation community, and hence craft more realistic legislation.     

Estimation and Inference Are Missing Data Problems: Unifying Social Science Statistics via Bayesian Simulation

Bayesian simulation is increasingly exploited in the socialsciences for estimation and inference of model parameters. Butan especially useful (if often overlooked) feature of Bayesiansimulation is that it can be used to estimate any function ofmodel parameters, including "auxiliary" quantities such as goodness-of-fitstatistics, predicted values, and residuals. Bayesian simulationtreats these quantities as if they were missing data, samplingfrom their implied posterior densities. Exploiting this principlealso lets researchers estimate models via Bayesian simulationwhere maximum-likelihood estimation would be intractable. Bayesiansimulation thus provides a unified solution for quantitativesocial science. I elaborate these ideas in a variety of contexts:these include generalized linear models for binary responsesusing data on bill cosponsorship recently reanalyzed in PoliticalAnalysis, item—response models for the measurement ofrespondent's levels of political information in public opinionsurveys, the estimation and analysis of legislators' ideal pointsfrom roll-call data, and outlier-resistant regression estimatesof incumbency advantage in U.S. Congressional elections     

Federalism, Political Structure, and Public Policy in the United States and Canada

Two of the three large countries on the North American continent—the United States and Canada—share a number of similarities that often make it difficult for the untrained observer to differentiate between the two nations. On the surface, the two are structured similarly as federal systems that, by definition, exhibit shared power between the national government and provincial or state political entities.Although there are other important social and economic characteristics of the two countries that help explain differences in policy processes and outcomes, it is the contention of this article that one gets the clearest sense of what Elazar has called thinking federal by utilizing an analytical approach that joins questions related to federalism with some conceptual frameworks of the public policy field. Two frameworks undergird the argument in this article—the Lowi typology of different types of policies and Deil Wright's typology of different models that describe the American inter-governmental system.In both countries, policies must be sensitive to the greater interdependencies between units of government as well as to linkages between policy areas. The mechanisms or instrumentalities for dealing with policy issues are intrinsically complex. It is also clear that the intergovernmental networks that exist in both the U.S. and Canada are composed of an array of actors. The differing political structures of the systems do impact the types of intergovernmental policies that have emerged in the two countries. The executive dominance so imbedded in Canadian governments has contributed to their ability to adopt and implement certain controversial redistributive policies, such as a national health insurance program. By contrast, the fragmentation of the U.S. system makes redistributive policies more difficult.     

Systematic differences in electropherogram peak heights reported by different versions of the GeneScan software

DNA profiling using STRs on the 310 and 3100 Genetic Analyzers routinely generates electropherograms that are analyzed with the GeneScan software available from the instrument's manufacturer, Applied Biosystems. Users have been able to choose from three different smoothing options that have been known to result in significant differences in the peak heights that are reported. Improvements in the underlying algorithm of the most recent version of the software also result in significant and somewhat predictable differences in peak height values. Laboratories that have performed validation studies using older versions of GeneScan should either reanalyze the data generated in those validation studies with the newest version of the software or otherwise take into consideration the systematically higher peak height values obtained as they begin following the recommendation of the manufacturer and use the new algorithm.     

Selection Bias in Comparative Research: The Case of Incomplete Data Sets

Selection bias is an important but often neglected problem incomparative research. While comparative case studies pay someattention to this problem, this is less the case in broadercross-national studies, where this problem may appear throughthe way the data used are generated. The article discusses threeexamples: studies of the success of newly formed political parties,research on protest events, and recent work on ethnic conflict.In all cases the data at hand are likely to be afflicted byselection bias. Failing to take into consideration this problemleads to serious biases in the estimation of simple relationships.Empirical examples illustrate a possible solution (a variationof a Tobit model) to the problems in these cases. The articlealso discusses results of Monte Carlo simulations, illustratingunder what conditions the proposed estimation procedures leadto improved results.