posted 02-07-2008 01:13 PM
quote:
DYKFS the identity of those subjects that took the money out of the register?Did you have prior knowledge that money would have been taken out of the register on that date?
I'd begin with a clear understanding of how you want to score the test, not just the test questions.
I also favor a single issue diagnostic test whenever possible in these cases, as these exams offer the best balance of sensitivity to deception and specificity to truthfulness - as long as you use Senter (two-stage) rules. It is my understanding that all variants of the Zone Comparison Test were initially intended to be single issue exams. Their application to mixed issues and multi-facet exam situations in field practice requires a departure from the scoring procedures for the single issue zone exams. Such departures from standard Zone procedures essentially regard the Zone technique according to MGQT principles. Existing studies (just read the journal for the cites) seem to suggest that if you use traditional Zone rules with the spot scoring rule, then you probably have a test with good sensitivity, but weak specificity. Senter rules are much better.
It is my understanding that all variants of the MGQT technique, and the GQT from which those were originally derived were initially intended to be mixed issue or multi-facet exams, though MGQT variants are often effectively used in single issue testing situations. The published studies seem to indicate the MGQT may offer good sensitivity to deception, though with rather punishing and disappointing specificity to truthfulness. Some data would suggest the MGQT can provide a more effective balance of sensitivity and specificity simply by using Senter Rules instead of MGQT/spot rules.
Multi-facet exams are always tempting from an investigative perspective, because we want to attach a laundry-list of various issues of concern (like getting free floor-mats when you buy a used car). Unfortunately there is still a lot we don't know about sensitivity and specificity with multi-facet exams. We think that attacking a known event or known allegation from various semantic angles will improve sensitivity. However, we haven't really shown that to be reliable. Theory tells us the test should become more sensitive, with some corresponding increase in FPs (=decrease in specificity to truthfulness). Polygraph, as a profession and science, has to grown beyond test design supported by mere theory (read: untested hypothesis). One thing we are concerned about is that screening exams such as LEPET, PCSOT, and security (which are mixed issue, not multi-facet), may not be as sensitive as we might be inclined to assume.
The difference between a mixed issue exam and multi-facet exam is simple: multi-facet exams require a known incident or known allegation. By definition: mixed issues exams are those for which it is conceivable that a test subject could lie to one or more questions while being truthful to other questions. With that in mind, in multi-facet examinations of known incident/allegation, it is inconceivable that a test subject could lie to one or more test questions while being truthful regarding his or her involvement in the issue/allegation.
Mixed issues exams are of unknown incident and without known allegations. They are therefore screening exams, for which we want to optimize sensitivity, with the goal of not overlooking any potential problem. Because they are screening exam, and because adverse action should not directly follow the results of screening exams, we are prepared in advance to tolerate some overprediction of problems (because it improves sensitivity). Errors will be remedied upon further investigation, and the narrowing of unresolved issues can contribute to an effective process of isolating and diagnosing problems.
The single issue exam is free (free-er) of the unquantified complications associated with multi-facet and mixed issues exam, and provides a more robust mathematical framework within which to resolve known incidents or known allegations.
So, what makes a single issue exam? Answer: when it is inconceivable that a subject could lie to one question while being truthful to another. That's enough for me.
So, it could be argued that your questions (repeated below) represent a single issue exam.
quote:
DYKFS the identity of those subjects that took the money out of the register?Did you have prior knowledge that money would have been taken out of the register on that date?
The only problem I have with these questions is the chunkiness of the language and use of semi-legalistic jargon (“identity” “prior knowledge”). Only polygraph examiners and police investigators talk like that.
Good (great) questions are any version of “did you do it.”
So, perhaps:
Multi-facet (MGQT)
- Did you plan with those people to have that money taken from the register on (date)?
- Did you ever talk to those people before they took the money from the register on (date)?
- Do you know the people who took the money from your cash register on (date)?
or Zone
- Did you plan with those people to take any money from your cash register?
- Did you plan with those people to have that money taken from the regiter on (date)?
- DYKFS those people had planned to take that money from your cash register?
Its still not great.
.02
r
------------------
"Gentlemen, you can't fight in here. This is the war room."
--(Stanley Kubrick/Peter Sellers - Dr. Strangelove, 1964)