On the Options tab of the Bootstrapping for 2-Sample Means dialog box, select the type of interval, specify the confidence level for the confidence interval, store the differences, or set the base for the random number generator.

From Type of interval, select the type of confidence interval you want Minitab to calculate. With a confidence level of 95%, you need at least 40 resamples to calculate a confidence interval, and at least 20 resamples to calculate only an upper or lower bound.

- Two-sided
- Use a two-sided confidence interval to determine a range of likely values for the difference in population means.
- Lower bound
- Use a lower bound to determine a value that the difference in population means is likely to be greater than.
- Upper bound
- Use an upper bound to determine a value that the difference in population means is likely to be less than.

From Confidence level, select the level of confidence for the confidence interval. Usually, a confidence level of 95% works well.

A bootstrapping distribution approximates the sampling distribution of the difference. Therefore, the middle 95% of values from the bootstrapping distribution provide a 95% confidence interval for the difference.

For a given set of data, a lower confidence level produces a narrower confidence interval, and a higher confidence level produces a wider confidence interval. The width of the interval also tends to decrease with larger sample sizes.

Select to store the differences from each resample in the worksheet. Minitab stores the values in the column Differences after the last column of data.

(Optional) In Base for random number generator, you can specify the starting point for the random selection of the bootstrapping sample by entering an integer that is greater than or equal to 1. When you use the same base number, you get the same sample.

For example, a professor generates 50 resamples of the original data for use in a classroom exercise. The professor and students each set the base to 1 to generate the same bootstrapping distribution and thus, the same analysis results.