Part V: Analyzing Data with Excel
792
This procedure has two options: periodic and random. If you choose a periodic sample, Excel
selects every nth value from the Input range, where n equals the period that you specify. With a
random sample, you simply specify the size of the sample you want Excel to select, and every value
has an equal probability of being chosen.t-Test
Use the t-Test tool to determine whether a statistically significant difference exists between two
small samples. The Analysis ToolPak can perform three types of t-Tests:l (^) Paired two-sample for means: For paired samples in which you have two observations
on each subject (such as a pretest and a post-test). The samples must be the same size.
l (^) Two-sample assuming equal variances: For independent, rather than paired, samples.
Excel assumes equal variances for the two samples.
l (^) Two-sample assuming unequal variances: For independent, rather than paired, sam-
ples. Excel assumes unequal variances for the two samples.
Figure 38.10 shows output for the Paired Two Sample for Means t-Test. You specify the significance
level (alpha) and the hypothesized difference between the two means (that is, the null hypothesis).
FIGURE 38.10
Output from the paired t-Test dialog box.
z-Test (two-sample test for means)
The t-Test is used for small samples; the z-Test is used for larger samples or populations. You must
know the variances for both input ranges.