Statistics and Policy Decisions: Issues in Statistical Analyses
Pages 162-171
Helena Chmura Kraemer
DOI: http://dx.doi.org/10.6000/1929-6029.2015.04.02.1
Published: 21 May 2015

Abstract: When national policy decisions are to be guided by the results of statistical analyses, it is important, to avoid being misled to look beyond the authors’ conclusions and first to assess the study design, measurement and analytic methods, in order to decide whether a study’s conclusions rest on a solid foundation. In particular, observational studies must be carefully and critically evaluated. Using a study widely cited concerning the effects of low-level lead exposure and IQ, we illustrate several methodological errors, long known but often ignored. The goal is not to settle the controversies about the effect of lead on IQ, nor to disparage observational studies, for they are the foundation of all studies done to guide policy, but to encourage additional care in the use of such studies to address policy questions.

Keywords: Policy decisions, Statistical Significance, Practical or Policy Significance, Methodological Errors, Lead/IQ Association.

Measurement and Mismeasurement of Social Development in Infants Later Diagnosed with Autism Spectrum Disorder
Pages 180-187
Ami Klin and Warren Jones
DOI: http://dx.doi.org/10.6000/1929-6029.2015.04.02.3
Published: 21 May 2015

Abstract: Autism spectrum disorder (autism) is a common and heterogeneous neurodevelopmental disorder of genetic origins defined by challenges in social communication and clusters of restrictive and repetitive behaviors. An emerging hypothesis of autism pathogenesis describes symptoms as the results from deviations from normative developmental processes. In this account, symptoms represent the outcome of variable instantiation of genetic liabilities – in terms of dosage and timing – leading to disruptions in the developmental trajectories of foundational social adaptive skills. Given the fast pace of change in behavior and brain development in the first two years of life, we pose that the currently prevalent cross-sectional experimental designs are ill-suited to capture changes from normative benchmarks that might be small at any data point but which inexorably and cumulatively increase divergences in developmental trajectories that ultimately culminate in the unmistakable cluster of atypical behaviors we now call autism. We argue that only densely-sampled longitudinal experimental designs can capture the underlying dynamic processes moving the individual child’s development towards or away from normative benchmarks. We illustrate this phenomenon via a detailed example in which a cross-sectional comparison between a clinical and a control cohort failed to find differences, which could only be detected by ascertaining that the developmental trajectory of one cohort was moving upwards while the other was moving downwards, with the developmental lines intersecting at the cross-sectional data point. We conclude by magnifying Karmiloff-Smith’s assertion, oft-quoted but seldom followed, that “development itself is the key to understanding developmental disorders” [1].

Keywords: Autism, Autism Spectrum Disorder, Social Visual Engagement, Eye Fixation, Infancy, Prodromal, developmental trajectories, growth curve, growth charts.