Publication date: 04/12/2021

To perform a bootstrap analysis, right-click a numeric column of sample statistics in a table in a report window and select Bootstrap. The selected column is highlighted, and the Bootstrapping window appears. After you select options and click OK in the Bootstrapping window, bootstrap results for every statistic in the column appear in the default results table.

Note: The Bootstrap option is not available in reports that use a By variable.

The Bootstrapping window contains the following options:

Number of Bootstrap Samples

Sets the number of times that you want to resample the data and compute the statistics. A larger number results in more precise estimates of the statistics’ properties. By default, the number of bootstrap samples is set to 2500.

Random Seed

Sets a random seed that you can re-enter in subsequent runs of the bootstrap analysis to duplicate your current results. By default, no seed is set.

Fractional Weights

Performs a Bayesian bootstrap analysis. In each bootstrap iteration, each observation is assigned a weight that is calculated as described in Calculation of Fractional Weights. The weighted observations are used in computing the statistics of interest. By default, the Fractional Weights option is not selected and a simple bootstrap analysis is conducted.

Tip: Use the Fractional Weights option if the number of observations that are used in your analysis is small or if you are concerned about separation in a logistic regression setting.

Suppose that the Fractional Weights option is selected. For each bootstrap iteration, each observation that is used in the report is assigned a nonzero weight. These weights sum to n, the number of observations used in the calculations of the statistics of interest. For more information about how the weights are calculated and used, see Calculation of Fractional Weights.

Split Selected Column

Places bootstrap results for each statistic in the column that you selected for bootstrapping into a separate column in the Bootstrap Results table. Each row of the Bootstrap Results table (other than the first) corresponds to a single bootstrap sample.

If you deselect this option, a Stacked Bootstrap Results table appears. For each bootstrap iteration, this table contains results for the entire report table that contains the column that you selected for bootstrapping. Results for each row of the report table appear as rows in the Stacked Bootstrap Results table. Each column in the report table defines a column in the Stacked Bootstrap Results table. For an example, see Stacked Bootstrap Results Table.

Discard Stacked Table if Split Works

(Applicable only if the Split Selected Column option is selected.) Determines the number of results tables produced by Bootstrap.

If the Discard Stacked Table if Split Works option is not selected, then two Bootstrap tables are shown:

– The Stacked Bootstrap Results table, which contains bootstrap results for each row of the table containing the column that you selected for bootstrapping. This table gives bootstrap results for every statistic in the report, where each column is defined by a statistic.

– The unstacked Bootstrap Results table, which is obtained by splitting the stacked table. This table provides results only for the column that is selected in the original report.

If the Discard Stacked Table if Split Works option is selected and if the Split Selected Column operation is successful, the Stacked Bootstrap Results table is not shown.

Want more information? Have questions? Get answers in the JMP User Community (community.jmp.com).