
The statistical methods in MIL-HDBK-1823A require that min(POD)=0, and max(POD)=1. These conditions are easily met by most, BUT NOT ALL, POD data.
If your min(POD)>0, due to background noise, for example, or your max(POD)<1, caused perhaps by difficulties in accessing the inspection site, then the POD vs. size model will not agree with your POD data.
Whenever ANY model disagrees with the data being modeled, the result will be WRONG. You may be able to coax the software into producing a POD(a) curve, but it will be WRONG, sometimes with a90/95 values where none exist.
If your data suggest that their POD minimum value never reaches zero, or their maximum doesn't approach one, please keep me in mind - I can help you correctly determine your system's POD performance.
Figure 1 - POD(a) Model with "floor" (
)
and "ceiling" (
)