Robust statistics for personality data

Summer School 10 | 2011, July 16-21 | Bertinoro, Italy


Information

 

Featured topics

  • Robust statistics for personality and individual differences

Background

  • Traditional parametric statistical procedures such as the Pearson correlation,  regression, and tests of group differences by t-tests and analysis of variance depend much more on unrealistic assumptions than most psychologists believe. Biased results due to extreme cases such as outliers or mixed distributions of a small extreme group and a much larger normal group are common in psychology and may be one of the major reasons for the embarrassingly low replicability of findings in psychological research. In recent years numerous alternatives to parametric statistics have been developed called robust statistics (see overview by Erceg-Hurn et al.  American Psychologist 2008) and have been implemented in freely available statistical packages such as R. In addition, there is a recent increase in applying bootstrapping for robust estimations of confidence intervals (e.g., replacement of the Sobel test in mediation analyses by bootstrapping procedures; new bootstrapping option for most major statistical tests from SPSS 18.0 on) and in controlling significance levels in correlational matrices through randomization. The aim of the summer school on robust statistics is to make participants familiar with major robust statistical methods and their implementation in R. Participants will be encouraged to bring own data for analyses under the supervision of faculty members.

Teaching Faculty

  • Rand R. Wilcox (University of Southern California, U.S.A.)
  • Jens B. Asendorpf (Humboldt University Berlin, Germany)
  • Felix Schönbrodt (University of Munich, Germany)
  • Ryne A. Sherman (the University of California at Riverside, U.S.A.)

Organizers

  • Jens B. Asendorpf
  • Marco Perugini

Sponsors

  • The European Association of Personality Psychology (EAPP)
  • The International Society for the Study of Individual Differences (ISSID)