Suppose they say that there is no difference, on average, in the time parents (Group A) and non-parents (Group B) spend watching TV on a typical day.

You believe that, on average, there is a difference.

You take a sample from parents and non-parents and record the minutes spent watching TV on a typical day.

Assume both groups have equal variances.

Group A has a sample sum of squares of 123

nA = 8

Group B has a sample sum of squares of 134

nB = 6

The observed sample mean difference is -7.3

If we used a score in Group B to estimate the mean of the population of Group B, on average, how far off would we be?

How far do we expect our sample mean difference to be from the expected?

On average, how far are the scores in Group A from the sample mean of Group A?

What can you conclude?