110 BUSINESS APPLICATIONS OF MULTIPLE REGRESSION
Therefore,
H
0
: β
1
≤ 0
H
1
: β
1
> 0
Just by looking at the chart back in Figure 3.20, we know that this
assumption will lead us to conclude that the relationship is insignificant.
However, if we simply take the Student t- value for the β
1
coefficient of
0.0844 from the regression in Figure 3.21 and divide by two, we obtain
0.0422 and we therefore reject the null hypothesis and conclude the
model is significant. This time, we truly have a contradiction.
Briefly, the problem is that Excel computes the test statistic and then
finds the area on both sides. It takes the smaller of these two values and
doubles it to compute the two- tailed p- value. Simply dividing by two to
obtain the ...