Database Reference
In-Depth Information
FIGURE 11.7
A repeat of
Figure 11.5
; SPSS with illustrative example.
It can be requested in SPSS by going back to
Figure 11.5
, repeated here as
Figure 11.7
, and clicking on “Options” (see arrow in
Figure 11.7
).
We see the “Logistic Regression: Options” dialog box in
Figure 11.8
.
We clicked the box for the “Hosmer-Lemeshow goodness-of-it” (see horizontal
arrow in
Figure 11.8
) and then clicked “Continue” to return us to
Figure 11.7
(see
vertical arrow in
Figure 11.8
). Now, in addition to the output in
Figure 11.6
(that we
discussed), we also obtain additional output that includes what is shown in
Figure 11.9
.
So, we earlier noted the twist: usually, the focus in hypothesis testing is to see if
we obtain a low
p
-value (less than 0.05); here, the key is to note if we obtain a high
p
-value. Obtaining a “high”
p
-value means
we do not have evidence to reject that the
logistic regression model its the data well
. Here, we have a
p
-value of 0.388; that is
nowhere near 0.05, and would be generally considered a “high”
p
-value. Thus, we
can conclude that this model is a good it.
11.4.2.2 Finding the predicted probability of a “1” for each data point
Quite often, we wish to ind the predicted probability of getting a “1” (here, com-
pleting the task successfully) for several of the X values. In the previous section, we
showed how to compute these predicted values. While it is not really dificult, it can
Search WWH ::
Custom Search