Minitab v13 -
Hypothesis Testing: Continuous Variables (2 Sample)

The Minitab Project is available. It contains the following data:

Type t	Data Format	Column	N	Comments
Between or Indep- endent Groups	unstacked	Written	9	scores on the written papers
		Typed	10	scores on the written papers
	stacked	Code	19	dummy codes for groups (1=written, 2=typed)
		Scores	19	all of the "Written" & "Typed" data
Paired or Dependent Groups		Red	10	reaction times to red lights
		Greed	10	reaction times to green lights

t test - Between Groups

Use the "2-Sample t..." command off of the "Stat, Basic Statistics" menu. That is:

That will take you to the following dialog box:

For the first part of this example choose "Samples in different columns" and type the relevant variable(s) in the box (or double click them from the left hand menu). Make sure the "Assume equal variances" box is checked. You might check the "Options..." button to verify that alpha is set to .05 with a nondirectional test. Since we are dealing with only two means, we won't bother with a graph. The output will look like this:

Two-Sample T-Test and CI: Written, Typed Two-sample T for Written vs Typed              N    Mean    StDev    SE Mean Written      9   82.00     4.03        1.3 Typed       10   85.60     3.03       0.96 Difference = mu Written - mu Typed Estimate for difference: -3.60 95% CI for difference: (-7.03, -0.17) T-Test of difference = 0 (vs not =):           T-Value = -2.22 P-Value = 0.041 DF = 17 Both use Pooled StDev = 3.53

We could have also run this test with stacked data (where each subject would represent a row in the worksheet) rather than unstacked where the two groups are in separate columns. Stacked data is actually a more realistic approach, since we typically record multiple DV's as well as demographic and other information for each subject. Having each subject represent one row of data is convenient and efficient.

For stacked data we would chose "Samples in one column" in the dialog box and typed the dependent variable in the "Samples:" box and the codes with the levels of the IV in the "Subscripts:" box.

Again it is important to make sure the "Assume equal variances" box is checked. In this case, the output would look like this:

Two-Sample T-Test and CI: Scores, Code Two-sample T for Written2 vs Typed Code         N    Mean    StDev    SE Mean 1            9   82.00     4.03        1.3 2           10   85.60     3.03       0.96 Difference = mu (1) - mu (2) Estimate for difference: -3.60 95% CI for difference: (-7.03, -0.17) T-Test of difference = 0 (vs not =):           T-Value = -2.22 P-Value = 0.041 DF = 17 Both use Pooled StDev = 3.53

t test - Dependent Groups

Use the "Paired t..." command off of the "Stat, Basic Statistics" menu. That is:

That will take you to the following dialog box:

Type the relevant variable(s) for the "First sample:" and "Second sample:" as Minitab calls these multiple measurements (or double click them from the left hand menu). You might check the "Options..." button to verify that alpha set to .05 with a nondirectional test. Since we are dealing with only two means, we won't bother with a graph. The output will look like this:

Paired T-Test and CI: Red, Green Paired T for Red - Green             N    Mean   StDev   SE Mean Red        10   25.30    5.66      1.79 Green      10   27.90    4.75      1.50 Difference 10   -2.60    3.27      1.03 95% CI for mean difference: (-4.94, -0.26) T-Test of mean difference = 0 (vs not = 0):           T-Value = -2.51 P-Value = 0.033