Suppose a credit card company wants to examine the difference between credit card spending on groceries and leisure. To do so, it generates a paired sample of 7 credit card customers' spending in each category. Assume spending in each area is normally distributed. Data are in thousands.

Groceries	Leisure
6.8	5.4
5.9	3.6
7.5	6.0
8.7	2.5
10.5	3.2
5.2	3.9
6.8	5.4

1. If the credit card company believes the population difference in spending is $4,500 (4.5), test whether the mean spending difference in the sample is different from the population mean. Use a .10 level of significance.
2. What is the 90% confidence interval for the difference in spending?
3. Why is a paired sample better than an unpaired sample when testing hypotheses concerning the difference between two population means?