Humboldt-Universität zu Berlin - Statistik

Datenanalyse I (VL+UE)

The lecture covers classical statistical topics and we will work half of the time with SPSS and/or R in a PC-Pool. Note: The lecture will be hold in german, but slides and exercise will be in english.


  • Survey design
    • Operationalization
    • Validity 
  • Short repetition of Statistics I+II
    • Sampling
    • Scaling
    • Test theory
    • Estimation theory
  • Identification of outliers
  • Treatment of missing values
    • Types (MAR, MCAR, MNAR)
    • Imputation methods (single, multiple)
  • Univariate statistics
    • Graphical representation (stem-and-leaf, strip plot, kernel density, violin plot)
    • Coeffcients (quantile, entropy, higher moments, hinges and spreads)
    • Tests (t-Tests, Mann-Withney-U, Median, Wilcoxon, Kruskal-Wallis, ANOVA, Friedman)
    • Transformations (Power, Box-Cox)
  • Bivariate statistics
    • Graphical representation (sunflower plot, mosaic plot, trellis display)
    • Voeffcients and tests (association and PRE- measure, Cohen's kappa, relative risk, odds ratio)
  • Robust statistics
    • Robust coefficients for location and dispersion
    • L- and M-estimators for location
  • Effect sizes
  • Multivariate statistics
    • Graphical representation (mosaic plot, scagnostics, parallel coordinates, Chernoff faces, interactive graphics)
    • Principal component analysis
    • Exploratory factor analysis (reliability for sum scores)
    • Cluster analysis