|
| |
Demystifying Statistical Sampling:
Tests of Controls
(a.k.a., Attribute, Hypotheses, Acceptance Testing)
In the good old days of audit, we performed
a LOT of tests of controls (not “reviews” of controls … TESTS). And we utilized
statistical sampling so that we could, scientifically, project our sample
results to the population. Remember, if you do not take statistical samples, you
cannot project your findings to the population (and thus you are cheating
yourself of your fine work!)
In the 80’s and 90’s auditors moved “away”
from controls testing. One audit partner said to Dr. Dan, “We don’t do ANY test
of controls.” (Note: Dr. Dan trained all 8 of the old Big-8.) Oh well. Enter
SOX!
So, statistical sampling has made a come
back (it never went away in the governmental sector) as a premium test of
controls. There are multiple methods of attribute testing: 2-step, discovery and
parameters unknown. We need to learn all of these methods, to apply the proper
method given our audit objectives AND the characteristics of the population.
What’s the problem with learning stat
sampling? PAIN! No, no, that’s the old way, with those awful tables. It IS
easier. We’ll use a great … and proven, freeware package to learn 1) the
multiple methods of attribute testing, 2) multiple methods of selecting a
sample, and 3) sample projection. Lots of FAQs will be addressed along with
work-paper design. We will also review the audit literature which supports stat
sampling and discover why “LARGE SAMPLES AREN’T NECESSARILY GOOD SAMPLES.”
Finally, we’ll consider some urban legends like “always take 33,” and find out
where the heck THAT came from!!!
We will discuss actual business cases with
huge recoveries.
THIS CLASS IS 100%
HANDS-ON, WITH SEVERAL COMPUTERIZED CASES, USING A GREAT FREEWARE PACKAGE
DEVELOPED BY A FEDERAL AUDIT AGENCY (and windows driven).
|
