Description
Using SPSS EXAMINE to inspect distributions, normality, outliers and homogeneity of variance by group before choosing parametric or non-parametric tests.
EXAMINE produces box plots, stem-and-leaf plots, Q-Q plots, normality tests (Kolmogorov-Smirnov with Lilliefors correction, Shapiro-Wilk), extreme values and Levene's test for homogeneity of variance, overall or split by group. Its purpose is to decide whether assumptions of t-tests, ANOVA or regression are plausible. Rigour comes from interpreting plots rather than relying on significance tests of normality, which are underpowered in small samples and over-sensitive in large ones, checking assumptions within groups rather than overall, and choosing robust alternatives (Welch's test, rank tests, bootstrapping, log transforms) when assumptions fail.
Attributes
- Discipline
- Statistics & Modelling
- Loss type
- Vestigial
- Stage
- Evaluation, Post-launch
- Data
- Quantitative
- Effort
- Hours
Steps
- Run EXAMINE with the outcome as the dependent list and the grouping variable as the factor, requesting plots, normality tests and Levene's test.
- Inspect Q-Q plots and box plots within each group, giving more weight to plots than to normality p-values.
- Note extreme values flagged by box plots (beyond 1.5 or 3 interquartile ranges) and trace them back to raw data.
- Check homogeneity of variance; if the largest group variance is more than about 4 times the smallest or Levene's test is significant, use Welch-type tests.
- For skewed task times, log-transform and re-examine, or switch to non-parametric or bootstrap methods.
- Document the assumption decisions in the analysis log.
Sources
- SPSS Syntax Editor
- Python integration (SPSS Python plugin)
- pyreadstat for Python ↔ SPSS bridge
- Field, A. (2024) Discovering Statistics Using IBM SPSS Statistics, 6th ed.
- Tukey, J. W. (1977) Exploratory Data Analysis
- Sauro, J. & Lewis, J. R. (2016) Quantifying the User Experience, 2nd ed.
- The Inquiry Series, C6 — SPSS for UX Researchers — The Right Way
Also known as
- Explore procedure
- Assumption checking
Categories
- Analytics & Behaviour
- Statistics & Modelling