# Procedures that support bootstrapping

The following procedures support bootstrapping.

- Bootstrapping does not work with multiply imputed datasets. If there is an
*Imputation_*variable in the dataset, the Bootstrap dialog is disabled. - Bootstrapping does not work if there are non-integer weight values.
- Bootstrapping uses listwise deletion to determine the case basis; that is, cases with missing values on any of the analysis variables are deleted from the analysis, so when bootstrapping is in effect, listwise deletion is in effect even if the analysis procedure specifies another form of missing value handling.

Note: Charts are generally not provided when bootstrapping is in effect for an
analysis.

## Statistics Base Edition

- Frequencies
- The following features are supported:
- The Statistics table supports bootstrap estimates for the mean, standard deviation, variance, median, skewness, kurtosis, and percentiles.
- The Frequencies table supports bootstrap estimates for percent.

- Descriptives
- The following features are supported:
- The Descriptive Statistics table supports bootstrap estimates for the mean, standard deviation, variance, skewness, and kurtosis.

- Explore
- The following features are supported:
- The Descriptives table supports bootstrap estimates for the mean, 5% Trimmed Mean, standard deviation, variance, median, skewness, kurtosis, and interquartile range.
- The M-Estimators table supports bootstrap estimates for Huber's M-Estimator, Tukey's Biweight, Hampel's M-Estimator, and Andrew's Wave.
- The Percentiles table supports bootstrap estimates for percentiles.

- Crosstabs
- The following features are supported:
- The Directional Measures table supports bootstrap estimates for Lambda, Goodman and Kruskal Tau, Uncertainty Coefficient, and Somers' d.
- The Symmetric Measures table supports bootstrap estimates for Phi, Cramer's V, Contingency Coefficient, Kendall's tau-b, Kendall's tau-c, Gamma, Spearman Correlation, and Pearson's R.
- The Risk Estimate table supports bootstrap estimates for the odds ratio.
- The Mantel-Haenszel Common Odds Ratio table supports bootstrap estimates and significance tests for ln(Estimate).

- Means
- The following features are supported:
- The Report table supports bootstrap estimates for the mean, median, grouped median, standard deviation, variance, kurtosis, skewness, harmonic mean, and geometric mean.

- One-Sample T Test
- The following features are supported:
- The Statistics table supports bootstrap estimates for the mean and standard deviation.
- The Test table supports bootstrap estimates and significance tests for the mean difference.

- Independent-Samples T Test
- The following features are supported:
- The Group Statistics table supports bootstrap estimates for the mean and standard deviation.
- The Test table supports bootstrap estimates and significance tests for the mean difference.

- Paired-Samples T Test
- The following features are supported:
- The Statistics table supports bootstrap estimates for the mean and standard deviation.
- The Correlations table supports bootstrap estimates for correlations.
- The Test table supports bootstrap estimates for the mean.

- One-Way ANOVA
- The following features are supported:
- The Descriptive Statistics table supports bootstrap estimates for the mean and standard deviation.
- The Multiple Comparisons table supports bootstrap estimates for the mean difference.
- The Contrast Tests table supports bootstrap estimates and significance tests for value of contrast.

- GLM Univariate
- The following features are supported:
- The Descriptive Statistics table supports bootstrap estimates for the Mean and standard deviation.
- The Parameter Estimates table supports bootstrap estimates and significance tests for the coefficient, B.
- The Contrast Results table supports bootstrap estimates and significance tests for the difference.
- The Estimated Marginal Means: Estimates table supports bootstrap estimates for the mean.
- The Estimated Marginal Means: Pairwise Comparisons table supports bootstrap estimates for the mean difference.
- The Post Hoc Tests: Multiple Comparisons table supports bootstrap estimates for the Mean Difference.

- Bivariate Correlations
- The following features are supported:
- The Descriptive Statistics table supports bootstrap estimates for the mean and standard deviation.
- The Correlations table supports bootstrap estimates and significance tests for correlations.

Note: If nonparametric correlations (Kendall's tau-b or Spearman) are requested in addition to Pearson correlations, the dialog pastes`CORRELATIONS`

and`NONPAR CORR`

commands with a separate`BOOTSTRAP`

command for each. The same bootstrap samples will be used to compute all correlations.Prior to pooling, the Fisher

*Z*transform is applied to the correlations. After pooling, the inverse*Z*transform is applied. - Partial Correlations
- The following features are supported:
- The Descriptive Statistics table supports bootstrap estimates for the mean and standard deviation.
- The Correlations table supports bootstrap estimates for correlations.

- Linear Regression
- The following features are supported:
- The Descriptive Statistics table supports bootstrap estimates for the mean and standard deviation.
- The Correlations table supports bootstrap estimates for correlations.
- The Model Summary table supports bootstrap estimates for Durbin-Watson.
- The Coefficients table supports bootstrap estimates and significance tests for the coefficient, B.
- The Correlation Coefficients table supports bootstrap estimates for correlations.
- The Residuals Statistics table supports bootstrap estimates for the mean and standard deviation.

- Ordinal Regression
- The following features are supported:
- The Parameter Estimates table supports bootstrap estimates and significance tests for the coefficient, B.

- Discriminant Analysis
- The following features are supported:
- The Standardized Canonical Discriminant Function Coefficients table supports bootstrap estimates for standardized coefficients.
- The Canonical Discriminant Function Coefficients table supports bootstrap estimates for unstandardized coefficients.
- The Classification Function Coefficients table supports bootstrap estimates for coefficients.

## SPSS® Statistics Premium Edition and Advanced Statistics Option

- GLM Multivariate
- The following features are supported:
- The Parameter Estimates table supports bootstrap estimates and significance tests for the coefficient, B.

- Linear Mixed Models
- The following features are supported unless a REPEATED subcommand is
specified:
- The Estimates of Fixed Effects table supports bootstrap estimates and significance tests for the estimate.
- The Estimates of Covariance Parameters table supports bootstrap estimates and significance tests for the estimate.

- Generalized Linear Models
- The following features are supported:
- Cox Regression
- The following features are supported:
- The Variables in the Equation table supports bootstrap estimates and significance tests for the coefficient, B.

## SPSS Statistics Standard Edition and Regression Option

- Binary Logistic Regression
- The following features are supported:
- The Variables in the Equation table supports bootstrap estimates and significance tests for the coefficient, B.

- Multinomial Logistic Regression
- The following features are supported: