This tool executes a two-sample student's t-Test on data sets from two independent populations with unequal variances.  This test can be either two-tailed or one-tailed contingent upon if we are testing that the two population means are different or if one is greater than the other.

The example below gives the Dividend Yields for the top ten NYSE and NASDAW stocks.  Use the t-test tool to determine whether there is any indication of a difference between the means of the two different populations.

To run the t-test:

1. On the XLMiner Analysis ToolPak pane, click t-Test:  Two-Sample Assuming Unequal Variances.
2. Enter B2:B11 for Variable 1 Range.  This is our first set of values, the dividend yields for the NYSE stocks.
3. Enter E2:E11 for Variable 2 Range.  This is our second set of values, the dividend yields for the NASDAQ stocks.
4. Enter "0" for Hypothesized Mean Difference.  This means that we are testing that the means between the two samples are equal.
5. Uncheck Labels since we did not include the column headings in our Variable 1 and 2 Ranges.
6. Keep the Alpha  = 0.05.
7. Enter G1 for the Output Range.
8. Click OK.

The results are below.

• Cells H4 and I4 contain the mean of each sample, Variable 1 = NYSE and Variable 2 = NASDAQ.
• Cells H5 and I5 contain the variance of each sample.
• Cells H6 and I6 contain the number of observations in each sample.
• Cell H7 contains our entry for the Hypothesized Mean Difference.
• Cells H8 contains the degrees of freedom.
• Cell H9 contains the result of the actual t-test.  We will compare this value to the t-Critical one-tail statistic.  Note:  Use a one-tail test if you have a direction in your hypothesis, i.e. if testing that a value is above or below some level.
• In this example P(T <= t) two tail (0.1477) gives us the probability that a value of the t-Statistic (1.58) would be observed that is larger in absolute value than t Critical two-tail (2.26).  Since the p-value is larger than our Alpha (0.05), we cannot reject the null hypothesis that there is no significant difference in the means of each sample.