Data Analysis with Microsoft Excel: Updated for Office 2007

Chapter 6 Statistical Inference 263

Under the pooled test the p value changes to 0.048, leading us to reject the

null hypothesis and accept the alternative: Nonrural homes generate more

revenue than rural homes. The 95% confi dence interval for the difference in

revenue ranges from $29 to $7,958. Notice that the t test and the confi dence

interval agree, as they always do: the test rejects 0 as the population mean

and the interval does not include 0.

We have a problem. Depending on which test we use, we reach a different

conclusion. The difference between the two tests depends on the assump-

tion we make about the population standard deviation. Should we use the

pooled or the unpooled results? To get a clearer picture, it would be a good

idea to create a chart of the two distributions with a boxplot.

To create a boxplot of the two samples:

1 Click Single Variable Charts from the StatPlus menu and then click

Boxplots.

2 Verify that the Use a column of category levels option button is

selected.

3 Click the Data Values button and select Revenue from the list of

range names.

4 Click the Categories button and select the range name Location.

5 Click the Output button and send the chart output to the chart sheet

Revenue Boxplots.

6 Click the OK button twice to generate the boxplots. See Figure 6-22.

Figure 6-21

Results of the

two-sample

t test analysis
with pooled
variance

conditions of the hypothesis test

assume unpooled variance