Geoff Cumming
The New Statistics: Estimation for Better Research
13th June 2013, 3:00 am - 4:30 pm
to be announced
Abstract
Research aims to build cumulative quantitative science. For this we need research methods and, most notably, data analytic methods that encourage researchers to think in terms of quantitative theories, ask quantitative research questions, and draw quantitative conclusions from data. Null hypothesis significance testing (NHST), which overwhelmingly dominates data analysis in many disciplines, fails all these requirements—and has other deep flaws. Among a number of more promising approaches, estimation based on effect sizes and confidence intervals is most immediately available and promising. I refer to estimation--and its extension, meta-analysis--asThe New Statistics. The techniques themselves are not new, but switching from NHST to these techniques would for many researchers be new, and highly beneficial. I will illustrate some quantitative advantages of estimation over NHST and describe practical ways to use the new statistics. I will also discuss the hot topic of replication, failure to replicate, and selective publication. I will use ESCI (Exploratory Software for Confidence Intervals), which runs under Excel, to illustrate concepts and calculate confidence intervals. ESCI is a free download from www.thenewstatistics.com