|
Elaboration:
This question was submitted by Timothy D. Ritchie, MA, from the department of psychology of Northern Illinois University (DeKalb, IL):
"My issue concerns the production of the exact p value, as shown by the code on p. 102 and within the output after running ex59.cmd and ex59.dat. I essentially used similar syntax but do not get any exact p values in my output. Why doesn't an exact p value appear in my output as it does in the example?"
Response:
ODA software computes exact one-tail p provided the following are true:
- The number of observations is 1000 or fewer;
- The class variable has two category levels;
- Priors is set off, or the class membership is balanced;
- The optimal value (number of correct classifications) is at least 75% of the total number of observations;
- No ties exist on the attribute; and
- The problem is not GEN or weighted. When exact p is not available, the Monte Carlo procedure may be used to estimate p to any desired confidence level.
|
 |
|
