SPSS command: SPSS EXAMINE
Explore distributions and check normality
Explore (SPSS EXAMINE) is the procedure for looking hard at a scale variable before you test it: centre, spread, shape, outliers and normality, overall or for each level of a factor.
Free · No account · Files stay on your device · Updated 18 October 2026
What it tells you
Crosstabs runs it in your browser with robust M-estimators, Tukey's hinges, Shapiro-Wilk and Lilliefors-corrected Kolmogorov-Smirnov tests, and Levene's test with four centring options.
- Whether the mean is distorted by outliers, by comparing it with the 5% trimmed mean, the median and four M-estimators.
- Whether each group looks normal, via Shapiro-Wilk and Kolmogorov-Smirnov with the Lilliefors correction.
- Whether group variances are similar, with Levene's test based on the mean, median, median with adjusted df and trimmed mean.
- Which cases are extreme: the five highest and lowest values with their case numbers, and the cases beyond the boxplot fences.
When to use it
Use it when
- Before a t test or ANOVA, to check the normality and equal-variance assumptions group by group.
- When you suspect outliers or data-entry errors and need the case numbers to find them.
- To describe skewed variables such as income or response time with robust statistics.
Use something else when
- Many variables needing only means and SDs: Descriptives is more compact.
- Testing a difference between groups: Explore checks assumptions; use a t test, ANOVA or a nonparametric test for the comparison itself.
- Categorical variables: use Frequencies.
How to run it in Crosstabs
- 01
Load data and open Statistics
Open your file in the workspace and switch to Statistics mode.
- 02
Choose Explore
In the Describe group, pick Explore (EXAMINE).
- 03
Set the Dependent list and Factor list
Tick one or more numeric variables in the Dependent list. Under Factor list, keep None (whole sample) or choose a grouping variable with 2 to 30 categories.
- 04
Choose what to display
Under Display, tick Descriptives, M-estimators, Outliers (extreme values), Percentiles, Normality tests and Boxplot outlier rules as needed; set the Confidence interval for mean (%).
- 05
Set missing values and run
Keep Exclude cases listwise (SPSS default) or choose pairwise, pick the Percentile method and M-estimator iterations, then click Run.
Worked example
Computed by the Crosstabs engine from the synthetic data below when this page was built — not typed in.
Data
Page load times in seconds for 40 synthetic visits: 20 on desktop (normal, mean 2.1 s) and 20 on mobile (log-normal, so skewed to the right). Generated with a fixed seed; device is the factor.
| Statistic | desktop | mobile |
|---|---|---|
| Mean | 2.153 | 2.618 |
| 95% Confidence Interval for Mean: Lower Bound | 1.956 | 2.062 |
| 95% Confidence Interval for Mean: Upper Bound | 2.351 | 3.174 |
| 5% Trimmed Mean | 2.153 | 2.584 |
| Median | 2.095 | 2.400 |
| Std. Deviation | 0.422 | 1.188 |
| Interquartile Range | 0.660 | 1.342 |
| Skewness | 0.147 | 0.721 |
| device | Kolmogorov-Smirnov Statistic | df | Sig. | Shapiro-Wilk Statistic | df | Sig. |
|---|---|---|---|---|---|---|
| desktop | 0.153 | 20 | .200* | 0.962 | 20 | .584 |
| mobile | 0.153 | 20 | .200* | 0.934 | 20 | .185 |
Kolmogorov-Smirnov uses the Lilliefors significance correction. * This is a lower bound of the true significance.
| Item | Levene Statistic | df1 | df2 | Sig. |
|---|---|---|---|---|
| Based on Mean | 11.506 | 1 | 38 | .002 |
| Based on Median | 9.140 | 1 | 38 | .004 |
| Based on Median and with adjusted df | 9.140 | 1 | 23.116 | .006 |
| Based on trimmed mean | 10.979 | 1 | 38 | .002 |
How to read it
Mobile loads are slower and more spread out: median 2.40 s against 2.09 s, with an interquartile range of 1.34 versus 0.66. The mobile mean (2.62) sits above its 5% trimmed mean (2.58) and its skewness is 0.72. Shapiro-Wilk gives p = .185 for mobile and p = .584 for desktop, and Levene's test based on the median gives p = .004. Read these together before choosing a t test, Welch's test or a rank-based test.
Output tables explained
- Case Processing Summary
- Valid, missing and total N and percent for each dependent variable and group.
- Descriptives
- Mean with standard error and confidence interval, 5% trimmed mean, median, variance, SD, minimum, maximum, range, interquartile range, and skewness and kurtosis with standard errors.
- M-Estimators
- Huber's, Tukey's biweight, Hampel's and Andrews' wave estimates of the centre, which downweight outlying values.
- Percentiles
- The 5th to 95th percentiles by the chosen method, plus Tukey's hinges.
- Extreme Values
- The five highest and five lowest values with their case (row) numbers.
- Tests of Normality
- Kolmogorov-Smirnov with the Lilliefors correction and Shapiro-Wilk, with df and significance; a small p means the data depart from normality.
- Test of Homogeneity of Variance
- Shown when a factor is set: Levene's statistic based on the mean, the median, the median with adjusted df, and the trimmed mean.
- Boxplot Outlier Rules
- Tukey's hinges, the hinge spread, the 1.5× and 3× fences, and which cases are outliers (o) or extremes (*).
Assumptions and notes
- Shapiro-Wilk needs 3 to 5,000 cases per group; above that only Kolmogorov-Smirnov is reported.
- With large samples normality tests flag trivial departures; read them with the skewness, kurtosis and percentiles.
- Under a sampling weight, confidence intervals, normality tests and Levene's tests are not reported because they would treat weights as case counts.
How the results are validated
Checked on 13 reference samples from NumPy 2.5, SciPy 1.18 and statsmodels 0.15: confidence intervals match scipy t.interval within 1e-9, Shapiro-Wilk matches scipy.stats.shapiro within 1e-6 (W) and 1e-5 (p), the Lilliefors D statistic matches statsmodels within 1e-12, M-estimators solve the statsmodels psi equations within 1e-9, Tukey's hinges and percentiles agree within 1e-12, and Levene's four variants match SciPy within 1e-10 (F) and 1e-9 (p). Validation evidence
Questions
Shapiro-Wilk or Kolmogorov-Smirnov — which should I report?
Shapiro-Wilk has more power for small and moderate samples and is the usual choice up to a few thousand cases. The Lilliefors-corrected K-S test is shown alongside; its p value is reported as a lower bound (.200) when the statistic is small.
Why do my M-estimators differ slightly from SPSS output?
By default the estimators iterate to full convergence. SPSS stops at a 0.5% change or 30 steps; choose that stopping rule under M-estimator iterations to follow it.
Does Explore draw boxplots?
It reports the Boxplot Outlier Rules table — hinges, fences and the case numbers of outliers and extremes — which is what you need to decide what a boxplot would flag.
Which Levene row should I read?
Based on Median is the most robust to non-normal data and is a sensible default; Based on Mean is the classical version used by the t test and ONEWAY tables.
Can I explore several variables by one factor at once?
Yes. Tick several variables in the Dependent list; each is split by the same factor. Listwise exclusion keeps the same cases in every table.
Related procedures
- DescriptivesDESCRIPTIVES
- P-P and Q-Q PlotsPPLOT
- Independent-Samples T TestT-TEST GROUPS
- One-Way ANOVAONEWAY
- Mann-Whitney U TestNPAR TESTS /M-W
- FrequenciesFREQUENCIES
- Free SPSS alternative: full coverage list
Run Explore on your own data
Open a .sav, CSV or Excel file. Everything runs in your browser; the file never leaves your device.
SPSS is a trademark of IBM Corp. Crosstabs is not affiliated with or endorsed by IBM.