posted 06-26-2007 08:20 PM            

Ben

posted 06-29-2007 09:09 AM            

some of the conclusions from Offe & Offe (2007)

quote:

---

The theoretically expected differences between RQ and CQ are found in the ratings of stress
of the different types of questions: RQare felt to be more stressing than CQ by guilty participants,
whereas innocent participants find the CQ more stressing.
A repeated measures ANOVA with question type (relevant or comparison) as withinsubject
variable and guilt as between-subject variable results in a main effect of question type
(F(1,63)=5.5896, p=.018, partial η2 =.086) indicating that participants experienced slightly
higher stress by the CQ (M=16.22) compared with the RQ (M=13,68). The interaction of
question type × guilt (F(1,63)=32.24, p=.000, partial η2 =.339) shows that, in agreement
with the theoretical assumptions, guilty participants felt higher stress with the RQ, innocent
participants felt higher stress with the CQ (Table 3).
To test the effects of the experimental conditions, univariate ANOVAs where conducted with
guilt, explanation and re-discussion as factors, stress by RQ or CQ as dependent variable and
Table 3 Means and standard
deviation for subjective stress of
relevant and comparison

stress by the other question type as covariate. Using stress due to CQ as dependent variable, no
significant main effects or interactions occurred. There is, however, a trend for an interaction
guilt×explanation (F(1,56)=2.86, p=.097, partial η2 =.049). While for innocent participants
there is no effect of explanation on stress due to CQ (M=18.93 and 18.46 with and without
explanation, respectively), for guilty participants there is a large difference: M=11.24 with
explanation and M=17.64 without explanation.
Using stress due to the RQ as dependant variable, there is a main effect of guilt
(F(1,56)=39.71, p=.000, partial η2 =.415). Guilty participants felt high stress due to RQ
(M=19.61) in contrast to the stress evoked by CQ (M=7.21).
This shows that the differential significance of questions is essentially achieved through the
different significance of RQ for guilty and innocent participants, but not through a difference
in significance of CQ. The experimental conditions (explanation and re-discussion) has no
significant effect on the stress of RQ or CQ. The weak interaction effect (guilt×explanation) on
stress by CQ, which miss significance, shows a tendency to effects only for guilty participants:
Explanation tends to reduce the stress due to CQ for these participants.

---

and some of their findings...

quote:

---

The identification rates depend on the manner of performing the examination. However, rediscussion
of CQ between tests has no effects on identification rates. Detailed explanation of
CQ, on the other hand, shows the expected effect: Identification rates for the entire group as well
as for the subgroup of guilty participants are significantly higher with such an explanation than
without. In addition, detailed explanation reduces the number of inconclusive cases. However,
if one looks at the subjective experience, it becomes apparent that detailed explanation of CQ
has the tendency to decrease the subjective stress of these questions for guilty participants–in
contrast to the increase in significance one would theoretically expect.
Results on subjective stress show, in agreement with the results of Honts (2003) andHorvath
(1988), the theoretically expected differential significance of RQ and CQ and thus confirm the
theoretical assumptions of the CQT. The fact that participants who could be correctly identified
based on their physiological reactions—in contrast to thosewho could not be correctly identified–
show a pattern of stress associated with RQ and CQ that is in agreement with theoretical
expectations, argues that the subjective stress is important for the test result. In addition, the
theoretically expected differences that were found in subjective stress for guilty and innocent
participants of the CQ and RQ suggest that subjective stress corresponds to the concept of
significance. After all, the substantial correlations between stress due to CQ and uncertainty
about answering them correctly agree with theoretical considerations about the concept of
significance. Based on these results, the attempt to measure important aspects of what is called
“significance” in the theoretical discussion by asking for subjective stress seems to have been
successful.

---

------------------
"Gentlemen, you can't fight in here. This is the war room."
--(Stanley Kubrick/Peter Sellers - Dr. Strangelove, 1964)

posted 06-29-2007 03:21 PM            

I didn't write that stuff. It is both dazzling and baffling, ain't it. Just ingore the jibberish and read the English parts and the point still comes through.

Its the point of the article.

Or more succinctly,

quote:

---

This shows that the differential significance of questions is essentially achieved through the different significance of RQ for guilty and innocent participants, but not through a difference in significance of CQ. The experimental conditions (explanation and re-discussion) has no
significant effect on the stress of RQ or CQ. The weak interaction effect (guilt×explanation) on
stress by CQ, which miss significance, shows a tendency to effects only for guilty participants:
Explanation tends to reduce the stress due to CQ for these participants.

---

Which also translates to:

• truthfull and deceptive people experience stress differently to the RQs, not so much to the CQs
  
• rediscussion of questions does not seem to produce any effect
  
• Explaining the CQs before the in-test produces a week trend for deceptive persons - it weakens their stress reactions to the CQs
  

The bottom line is the Offe & Offe seem to support the notion of differential reactivity to CQs and RQs for deceptive and non-deceptive polygraph subjects.

Its what we've been saying all along - we just use different words that sometimes place us out of step with other scientists.

Astute readers of Offe & Offe will note that they don't seem to subscribe at all to the idea that polygraph subjects react either to the RQs or to the CQs. Instead they illustrate that subjects react to both, though differently. It is the mathematical significance of those measurable differences that allow us to make classifications or decisions about truthfulness and deception.

Also, nowhere in Offe & Offe will you read the words "psychological set" and they do not attempt to embody the concept of differential reactivity in overly circumscribed concepts like fear, threat to survival, or general well-being. They correctly understand the construct of differential reactivity in very general terms such as "stress" and "reactions," and refrain from making unsupported psychologized explanations about the mechanisms or constructs underlying those internal experiences or observed reactions.

Friday,

r

------------------
"Gentlemen, you can't fight in here. This is the war room."
--(Stanley Kubrick/Peter Sellers - Dr. Strangelove, 1964)

posted 06-29-2007 11:49 PM            

I wish I could read the whole study, as this portion does not explain how it is Offe & Offe came to label the response as stress. Depending on how the battery of self reported stress was carried out, it could be a lot like the teacher who announces that there is has been a case of head lice in the school, what the symptoms are, and then ask the children if any of their heads itch.

The salience of the question has been a theme in polygraph for at least a few years now, at least at the last few APA seminars. Interestingly enough, this is much like the orienting response in the CIT, as salience of the correct item creates a larger orienting response for the knowledgeable subject when they recognize it out of the other not so salient incorrect items.

Again, great find!!

posted 06-30-2007 11:38 PM            

I have read some of the study and I find it to be an interesting study. I do like the effect they created with subjects choosing their condition, as they said this would be more realistic to real word conditions (e.g. people in the real world chose to commit a crime or not).

I am still reading over the study but there are some potential flaws that might have lead to the high rate of accuracy, if not other things as well. If I am reading the study correctly, one that I noticed was scorers reviewed the charts independent of each other but conferred when there were discrepancies to come to a decision. This could lead to a high inter-rater reliability and higher accuracy, depending on how often this occurred and what discrepancies were remedied.

I still am not convinced that this study proves anything by way of increase stress. In a study on CVSA and LVA done by the University of Florida individuals were tested for hormonal changes by taking saliva samples for increase in cortisol. The measurement of this hormone did not produce a useful result in that study but four were used: galvanic skin response, pulse rate, the emotion checklist, and the modified Hamilton scale. I am not saying that these four measurements would necessarily prove stress but at least it took into consideration multiple indicators that may more accurately correlate to the condition. To look at the amplitude of the responses and compared them for significance could be a better way of comparing at least significance of the questions, but I think they used an analog instrument and, if so, this could have made for a lot of manual measuring. Maybe, as suggested by Offe & Offe, it is relative situational significance. An entrance (prior to taking the exam) and exit (following the exam) survey might have also been useful to find if there were any changes.

Anyway, I am still looking at it and trying to get through it and digest the study.