SPSS command: SPSS BOOTSTRAP

Bootstrap confidence intervals in your browser

Bootstrapping resamples your cases with replacement many times and reads the bias, standard error and confidence interval of a statistic from how it varies across resamples.

Try with example data Run on your own file

Free · No account · Files stay on your device · Updated 18 October 2026

Worked example result

Bootstrap for Descriptive Statistics · Mean

Statistic = 39.21

Statistic	N	Statistic	Bias
Mean	30	39.21	0.210
Median	30	31.66	1.293
Std. Deviation	30	24.88	−0.716

Bootstrap for Descriptive Statistics

See the full worked example

What it tells you

Crosstabs runs the SPSS BOOTSTRAP approach on your device for descriptives, Pearson correlations and linear regression coefficients, with percentile or BCa intervals, optional strata and a reproducible seed.

A standard error and interval for statistics without a simple formula, such as the median or a standard deviation.
Whether a correlation or regression coefficient is clearly away from zero without assuming normal errors.
The bootstrap bias: how far the average resampled statistic sits from the original estimate.
How much an asymmetric (BCa) interval differs from the simple percentile interval for skewed statistics.

When to use it

Use it when

Skewed data such as spend or time, where a normal-theory interval for the mean is doubtful.
Medians and standard deviations, whose sampling distributions have no convenient closed form.
Checking that a regression or correlation result does not depend on normality or equal-variance assumptions.

Use something else when

Very small samples (say, under 20 cases): resamples can only recombine the few values you have, so intervals are too narrow.
Survey data with a sampling weight, strata and clusters: case resampling ignores the design and is blocked; use Complex Samples.
Time series or other dependent observations, where resampling single cases breaks the dependence.

How to run it in Crosstabs

01
Open your data
Open a .sav, CSV or Excel file in the workspace and switch to Statistics mode.
02
Choose the statistic to bootstrap
In the Utilities and charts group pick Bootstrapping, then Statistic to bootstrap: Descriptives (mean, median, standard deviation), Pearson correlations, or Linear regression coefficients.
03
Pick the variables
Tick Variables (two or more for correlations) and the statistics to compute, or for regression choose the Dependent and tick the Predictors.
04
Set the resampling
Number of samples (default 1000), Confidence level (%) (default 95), Confidence interval (Percentile or Bias corrected accelerated (BCa)), Sampling (Simple, or a strata variable) and Random seed.
05
Run
Declare an active weight as a frequency weight (a sampling weight blocks the run), then press Run; resampling happens in a background worker.

Run Bootstrapping in your browser

Worked example

Computed by the Crosstabs engine from the synthetic data below when this page was built — not typed in.

Try with example data

Data

30 synthetic online basket values (currency units) drawn from a right-skewed lognormal distribution with a fixed seed. Bootstrapping uses 1,000 simple resamples, the SPSS default, and seed 20261017.

Bootstrap for Descriptive Statistics

Statistic	N	Statistic	Bias	Std. Error	95% CI Lower (percentile)	95% CI Upper (percentile)	95% CI (BCa)
Mean	30	39.21	0.210	4.599	31.34	49.53	32.06 to 50.67
Median	30	31.66	1.293	4.352	25.24	43.00	22.67 to 42.90
Std. Deviation	30	24.88	−0.716	4.734	14.46	32.93	16.98 to 36.04

Based on 1000 simple bootstrap samples (seed 20261017). Bias and standard error are the same for both intervals because both runs use the same resamples.

How to read it

The mean basket is 39.21 with a bootstrap standard error of 4.60; its 95% BCa interval, which corrects for bias and skewness and is the one the worked example runs, is 32.06 to 50.67, against 31.34 to 49.53 for the plain percentile interval (the SPSS default). The median (31.66) has no simple textbook standard error, which is where the bootstrap earns its keep: its BCa interval is 22.67 to 42.90.

Output tables explained

Bootstrap for Descriptive Statistics: For each variable and statistic: N, the statistic, bootstrap bias, standard error and the lower and upper confidence limits.
Bootstrap for Correlations: Each pair of variables with r, its bias, standard error and interval; the summary line says whether the interval includes 0.
Bootstrap for Coefficients: The constant and each predictor's B with bootstrap bias, standard error and interval; missing values are excluded listwise.

Assumptions and notes

Cases are independent and the sample represents the population; the bootstrap cannot repair a biased sample.
Percentile limits are read from the (B + 1)-order statistics as in SPSS; if an end point falls outside the bootstrap distribution the extreme value is shown and footnoted.
With a frequency weight each case stands for that many cases when resampling; strata keep each stratum's size fixed.
The same seed reproduces the same resamples here; SPSS's random stream differs, so values will not match SPSS draw for draw.

How the results are validated

The independent references use a Python port of the same seeded generator, so resample indices are identical. On 1,000 resamples, bias and standard errors match to 1e-9 and percentile and BCa limits to 1e-10 against an independent implementation of the SPSS rules ((B + 1)-order statistics, jackknife acceleration), and scipy.stats.bootstrap fed the same resamples agrees to 1e-9. Regression coefficients, stratified resampling and frequency weights (against expanded data) are also tested. Validation evidence

Questions

Percentile or BCa?

The percentile interval is simple and is the SPSS default. BCa adjusts for bias and skewness of the bootstrap distribution and usually has better coverage, especially for skewed statistics; it costs one extra jackknife pass.

How many bootstrap samples do I need?

1,000 (the SPSS default) is fine for standard errors and 95% percentile intervals. For BCa or 99% intervals, 2,000–10,000 gives more stable end points.

Why does the interval change when I change the seed?

Each seed gives a different set of resamples, so end points move by a small Monte Carlo amount. Keep the seed in your report so others can reproduce it, and add samples if the change matters.

Can I bootstrap any procedure, as SPSS can?

No. This panel bootstraps descriptives (mean, median, standard deviation), Pearson correlations and linear regression coefficients. Other procedures do not offer a bootstrap option here.

Why is a sampling weight not allowed?

Resampling cases as if they were independent would ignore strata and clusters, giving the wrong standard errors. Use Complex Samples, which estimates variance from the design.

Run Bootstrapping on your own data

Open a .sav, CSV or Excel file. Everything runs in your browser; the file never leaves your device.

Open the workspace

SPSS is a trademark of IBM Corp. Crosstabs is not affiliated with or endorsed by IBM.