posted 05-14-2012 10:10 AM
All: This will be a long post.
There may still be some confusion between a Peak of Tension Test and a Concealed Information Test. They are not interchangeable. I have taken the liberty of pasting the CIT module from the computer software into this post. In doing so some content like images and tables were either deleted or reformatted. However, the how-to portions seem to be intact. If you do not have access to the complete module on your computer polygraph, let me know.
Concealed Information Test
Summary of Practice
1. Thoroughly study the case, including photos and statements from witness and victims.
2. Select potential information that could be useful in the CIT. That information must be something that:
a. The guilty person would have noticed and would remember.
b. The examinee could not have learned from the investigators, witnesses, victims, news media or others.
c. Is not so obvious or likely that an innocent person could correctly guess.
3. Conduct an elicitation pretest interview and ask the examinee what he knows about the incident being investigated. Do not reveal what you know.
4. If the person acknowledges knowing a detail, have him explain why he knows it. Confirm independently whether this explanation is true.
5. If the person denies knowing certain details, consider using them for the CIT. Try to obtain as many of these as possible.
6. Construct one CIT for each “key,” and five unrelated items (See the CIT Worksheet at the end of this module). The unrelated items should sound plausible.
7. Review all of the test items with the examinee, though not in the order they will be used.
8. Randomize the items in your CIT lists.
9. Run only one test for each CIT, unless repeats are required by movements or instrument problems or by department policy or state law.
10. Each CIT begins with a preparatory statement such as: “If you are the person who did X, you know (what kind of gun was used, the location of the victim’s wound, the point of entry into that house, etc.). Repeat after me each (weapon, location, etc.).” Then read each CIT item to the examinee. Because only the electrodermal channel is important, the items can be paced as rapidly as every 12 – 15 seconds. If you also plan to use the information in the other data channels, the pace should be not less than 25 seconds per item.
11. Have the examinee repeat each of the items after you say them, to ensure he is paying attention.
12. The tests can be conducted back-to-back if desired, unless there is a cardiograph cuff that occludes circulation in the arm.
13. Score each test with the Lykken scoring system. Decisions of Recognition Indicated (RI), No Recognition Indicated (NRI) or No Opinion can be based on the Lykken scoring system. For precise probabilities, see the CIT probability table in this module.
Background
In 1959 Dr. David Lykken introduced a recognition test that he called the Guilty Knowledge Test (GKT; Lykken, 1959). His test was based on the premise that there would be an orienting response elicited from guilty examinees at the critical item that would, on average, be larger than other similar-but-unrelated items, while innocent examinees would respond to all items randomly. The method has more recently become known as the Concealed Information Technique (CIT), though some of the literature still uses the older term.
The CIT has some advantages over the POT methods. First and foremost, it is much better researched. Unlike the POT, the CIT has been conducted in scores of studies with many hundreds of subjects, resulting in a solid estimate of its accuracy. The National Research Council’s 2003 report summarized the CIT research as showing an accuracy of 88%. This accuracy is better than what is found in the albeit-incomplete research on the POT. There is a general trend in finding higher CIT accuracy than POT accuracy (Barland, 1984; Yokoi, Okazaki, Kiriu, Kuramochi, & Ohama, 2001).
Another important feature of the CIT is that error estimates can be calculated from the score. Unlike the POT which has no accepted numerical scoring system, and the 7-position scoring system of the CQTs, the scores from the CIT map nicely over a probability distribution. For example a score of 12 with a CIT using 7 different critical items would be expected from only 1 in 1,000 examinees who were truly naïve to the crime details. There is more about this later in the chapter. The capacity of CIT to produce sound probability-based estimates qualifies it as an excellent technique for evidentiary polygraphy.
Because the CIT relies on the orienting response, it may be used in situations where the CQT is not appropriate. Take for example where a woman is found killed in her home, and the police immediately suspect the husband who is now at his office. There may be a temptation for the police to pick up the husband and test him with the conventional CQT or even an RI to solve the crime or at least remove the husband as a suspect. However, the CQT and RI rely on emotional responses to test questions, and it seems logical that the husband, so soon upon learning of his wife’s death, would have strong emotions to any question suggesting he was involved in her killing. The underlying basis of the CIT is not so dependent on emotion, and could be given to the husband as soon as he is able to sit for the test.
There are specific principles that make the CIT effective, and these principles do not constrain the technique to a singular method but permit some variation. For the sake of simplicity, only a single method will be provided here, and the principles will be discussed so that users will understand the bounds of permissible variation.
It is recommended that when both the CIT and CQT are to be conducted on the same examinee that the CIT be administered first. Giving the CIT priority is one way of ensuring that the CQT process does not reveal the kinds of details that are used in the CIT. Also, the CQT can be emotionally fatiguing to an examinee due to the necessarily dynamic and interactive pretest interview process, and the development of probable-lie material. The CIT pretest can be very brief, and there are no probable lies to develop, resulting in fewer cognitive and emotional demands on the examinee. Finally, if the CIT indicates that the examinee is concealing knowledge about the crime, an interrogation can begin immediately, producing a savings of both time and effort because the lengthier CQT has been made moot. Research has found that the CIT does better with innocent examinees than guilty examinees, so one can have more confidence when the examinee fails the CIT than when he passes it. It is only when the CIT has not found evidence of concealed information would the CQT be necessary.
The CIT uses only two types of items: critical items and non-critical items. A CIT is constructed using one critical item and some number of non-critical items. The recommendation here is to use five non-critical items, not because science has necessarily established this or some other as the idea number, but rather in the quest for simplicity and elegance the use of five non-critical items entails mathematics that is easier and can often be calculated in one’s head.
As with all recognition tests, selection of a critical item demands care. Examiners must be reasonably certain that the examinee was paying attention to the detail on which the critical item was based, and that it is available for recall during the CIT. Examinees have an annoying habit of exercising their own preferences for what they pay attention to and remember about their crimes, and they may not be the same details that the examiner would think obvious. To help in determining what might be meaningful to guilty examinees, examiners in Japan often visit the crime scene to see firsthand what the guilty person saw, to notice the types and placement of things that catch the eye. This approach to polygraphy is not widely practiced in the US or most other countries, but it has proven to be immensely successful in Japan. It may be the best approach to overcoming the problems inherent to the development of critical items in the CIT. If this level of involvement is not possible, examiners should choose critical items that cover details that a guilty examinee would find patently memorable.
Non-critical items also require thoughtful consideration. These items must not be evocative in themselves, lest the guilty find that they compete with the critical item for salience and thereby affect the scoring of the responses. For the sake of the naïve examinee non-critical items should be roughly the same length, in the same category, and equally plausible as the critical item. The equivalent of a POT’s false key has not been reported in the literature. Because false positives do not appear to be a significant problem for the CIT, false keys may not be necessary.
To construct a test in the CIT, a non-critical item is always placed in the first position. The placement of the remaining items is by chance. Examiners can use a random numbers table (such as the one at the end of this module), roll a die, or use some other method to place the critical item in the sequence. This means that the critical item may be located in the second position, the last position, or anywhere in between. A ready made CIT worksheet follows this chapter.
During the pretest interview it is wise to review the test items with the examinee, though not in the order that they will be presented during testing. The review will ensure the examinee understands the meaning of the items, and that the examiner’s pronunciation is clear to the examinee. It also acts as a final check for the innocent examinee who may have inadvertently learned a detail of the crime, and thereby prevent an erroneous conclusion that might otherwise result from that test. Examiners must be mindful during the review of the items that one test does not reveal the critical item of another test. For example, if one test is about the location of a weapon found in a room (i.e., in a desk), and a subsequent test is to determine whether the examinee knew in which drawer of the desk the weapon was found, it would be necessary to test on the location of the weapon in the room first without reviewing the other tests. Under those circumstances the examiner may review and conduct each test in a sequential manner, not revealing the content of other tests until they are to be conducted.
Conducting the CIT is quite simple. Unlike any other technique in polygraphy, tests are given but a single time. The examinee repeats each item back to the examiner, rather than say yes or no. Simply answering no to each item, as is done in the POT, presents an opportunity for the guilty examinee to dissociate and reduce the arousal value of the critical item. Repeating the items is a way of ensuring that the examinee has paid attention to them.
Each test starts with a statement that establishes the context for the items to follow. For example: “If you are the person who did X, you know what the weapon was.” Next is the instruction. “Repeat after me each of these weapons.” Finally, each item is presented to the examinee only once, with a pacing of about 25 seconds. When this test is completed, it is set aside, and a new test is conducted with a different critical item and new non-critical items. It is recommended that a minimum of four different tests be run, and as many as 10, if possible. A test may only be repeated if there was some type of artifact that eliminated the diagnostic value of the test. When a test is re-run, the order of the items is changed according to the rules listed earlier.
To give a concrete example of how the CIT can be used, the following is a vignette on which a series of tests will be shown.
The body of 74-year-old Mary Smith was found in a recliner in her living room on the afternoon of July 14th. She had been dead about 4 hours. There was no sign of forced entry to the home, and all but one door had been found locked. Her adult children reported that Smith usually kept the back door unlocked during the day so that it was easier to let her cat in and out. Police found that the screen on the back door had been pushed out from the frame, probably by the culprit during his escape. Smith had been struck on the back of the head with a heavy stone bookend found at the scene, a blow that caused her death. There was no evidence of any other trauma to Smith’s body, and it appeared she had been struck while she sat in the chair in which her body was found. Missing from the home were her Sony color television set, $16 in cash from the top of the refrigerator, and a leather jacket belonging to her grandson. The police located one suspect who had been seen in the area earlier that day, 31-year-old parolee named Doug Williams. Williams denied any knowledge or involvement in the crime, and has agreed to take a polygraph examination.
The following tests will assume that no information about the crime had been released to the public.
Test 1: The weapon
Context statement: If you killed Mary Smith, you know the object that was used to strike her. Repeat after me these objects:
Non-Critical Item: Fireplace poker
Critical Item: Stone bookend
Non-Critical Item: Table lamp
Non-Critical Item: Steam iron
Non-Critical Item: Tire jack
Non-Critical Item: Wine bottle
Test 2: Location of the body
Context statement: If you killed Mary Smith, you know which room the murder took place. Repeat these rooms after me:
Non-Critical Item: Bathroom
Non-Critical Item: Garage
Non-Critical Item: Basement
Non-Critical Item: Kitchen
Non-Critical Item: Bedroom
Critical Item: Living room
Test 3: Theft
Context statement: If you killed Mary Smith, you also stole something from her home. Repeat after me these things:
Non-Critical Item: Kitchen radio
Non-Critical Item: CD player
Critical Item: Television set
Non-Critical Item: Antique clock
Non-Critical Item: Laptop computer
Non-Critical Item: Digital camera
Test 4: The escape
Context statement: If you killed Mary Smith, you damaged something when you left so we know the direction of escape. Repeat the following:
Non-Critical Item: Bathroom sliding window
Non-Critical Item: Double front doors
Non-Critical Item: Sealed bedroom window
Critical Item: Back screen door
Non-Critical Item: Wooden garage door
Non-Critical Item: Red basement door
Only four tests are covered here, but many more are possible: the location of the victim’s wound, that she had been sitting in a chair, the amount of cash stolen and where it had been located, the missing leather jacket, to name a few. The important point to remember is that the examinee must be able to recognize the critical item.
Something that may also increase the salience of the critical item to the guilty is to give feedback after each test. This option is available for when the charts are relatively flat, and it does not increase the prospect of an error as it might for CQTs. If the examinee learns that he has been responding to items that implicate him, it can have an additive effect on the signal value of subsequent critical items. Imagine for a moment that you are the culprit in the above Mary Smith murder, and after the first test the examiner comments, “Joe, on that test your largest reaction was to ‘stone bookend.’ Sit still and we’ll start the next test.” The examiner should not say that “stone bookend” is the critical item. Such comments may unnerve the innocent examinee, and perhaps cause the guilty examinee to quit the exam. It is better to simply tell the examinee of the reaction, and discourage any further conversation about it. If the examinee asks whether the reaction was on the one related to the murder, the examiner should advise the examinee that all will be covered at the end of the testing. Keep in mind that if the examinee is reacting to the critical items without giving the feedback, that there is no value in pointing this out between tests.
Analyzing the CIT
Scoring of the CIT does not use all of the data that is normally recorded with field polygraphs. The research has focused primarily on electrodermal responses (EDRs), with dozens of scientific papers supporting its use. Respiration line length (RLL) has been proposed as another diagnostic measure in the CIT (Timm, 1982, 1989; Elaad, Ginton & Jungman, 1992) and though encouraging, RLL has not moved beyond the experimental stage for the CIT. There is insufficient empirical support for scoring the cardiovascular channel. To remain close to what can be supported scientifically, only scoring of EDRs will be discussed. It is recommended that the pneumograph be recorded along with the electrodermal channel during the CIT so that examiners might detect respiratory activity that could generate EDRs, and avoid presenting items when EDRs might spring from breathing behaviors. There is no reason to record the cardiovascular activity during the CIT, except to satisfy legal, licensing, or policy requirements.
Before moving on to the mechanics of scoring, it is helpful to begin with an examination of the underlying premise of the CIT. Suppose that an examinee were being tested on the murder of Mary Smith described in the earlier vignette. Recall that there were six items presented in the first test. Because examinees are known to always react to the first item, any reaction to the first item is ignored. The examinee, then, may have produced his largest reaction to any one of the five remaining items, including the critical item. If the examinee did react greatest to the critical item, chance has not been excluded as the possible cause. Under normal conditions, examinees knowing nothing about the crime will react to the critical item by chance one in five times, or 20%. Suppose for the present purposes that the examinee did react to the critical item on the first test, “stone bookend.”
On the second test there are also six items, but ignoring the first item, there are five to which the examinee may react the largest, just as there was in the first test. Suppose the examinee reacted to the critical item on this test, “living room.” There is a one-in-five chance, or 20%, of this happening to an innocent person, too. However, the chance that an innocent person would react to both “stone bookend” and “living room” is a mere 4% (20% X 20% = 4%). A 4% chance is one out of 25 innocent examinees will react in this manner. Now consider the chance that the innocent person will react to “stone bookend” on the first test, “living room” on the second test, and the critical item on the third test, “television set.” This is 20% of 20% of 20%, or only 0.8%. There is roughly one innocent person in a hundred that would react to all three critical items. Add to this the fourth test where the person reacted to the critical item “back screen door” and the chances of the examinee being naïve to the crime details is very remote, rounding up to only 2 in 1000. Each additional test where the examinee reacts greatest to the critical item moves the examinee’s claims of ignorance further from the realm of credibility.
The examples above have assumed that there were five scoreable items on each test. This is not the only configuration for the CIT, but merely a convenience for instruction. Examiners might employ any number of items on a test, so long as there is only a single critical item. If in the present murder case there had been 10 scoreable items instead of 5 on a test, and the examinee reacted largest on the critical item, the likelihood of this happening by chance is 10% instead of 20%. Similarly, reacting the greatest to the critical items on two tests is 1% (10% of 10%), and 0.1% on three tests (10% of 10% of 10%.)
It is worth repeating that examinees who react to the critical items on the CIT are not necessarily guilty of the crime, though this would be the natural conclusion. These reactions only indicate the examinee has been exposed to the crime details that the guilty person would know. For this reason, the results of the CIT are Recognition Indicated (RI) and No Recognition Indicated (NRI). The use of RI and NRI make clear that these results do not speak to the examinee’s guilt, only that he knows information that he denies knowing.
Returning now to the principles of scoring, in his 1959 paper Lykken proposed a simple yet effective scoring strategy. First, the reaction to the first item is always ignored. If the largest EDR occurs at the critical item, the test is scored as a “2.” After that, if the second largest EDR occurs to the critical item, the test is scored as a “1.” Otherwise the score is “0.” Each test receives only one score, and the scores for all tests are totaled. Referring to Figure 1, if item B were the critical item in this test, the score would be “2” because it is the largest EDR. If the critical item were item C, the score would be “1”, being the second largest EDR. If the EDR to the critical item were ranked third or lower the score for that test would be “0.” Reminder: always ignore the reaction to the first item.
Figure 1. EDRs on a CIT test.
(did not copy to this bulletin board)
Lykken’s (1959) decision rules were also simple: if the total score was equal to or greater than the number of tests conducted, the examinee was classified as having concealed information. Lower scores indicated that the examinee did not have this information. There were no Inconclusives, though later writers have suggested that there must have been an EDR on one of the items to consider a test valid. A lack of EDRs to all items would therefore result in an Inconclusive, and this is the recommended practice. To obtain exact probabilities of scores with the CIT for up to eight tests, refer to Table 1.
Table 1. CIT probability table. (Table format was changed in this post, so it was deleted.
The endemic problem of all recognition tests, that guilty examinees do not always recognize the critical items selected by examiners, has a potential remedy. It is recommended that as many tests be conducted as is reasonable, thereby increasing the likelihood that the examiner will have covered the field properly. Having larger numbers of tests not only improves the prospect for correctly identifying those with concealed information, it incrementally and simultaneously reduces the chances of a false positive result. For evidentiary examinations, it is recommended that the number of tests be conducted until a score indicates there is less than 1% chance that the examinee is ignorant of the crime details, or 10 tests, whichever comes first. Utilization of the CIT in investigative examinations, where the tests are used for less than evidentiary purposes, may permit a more lenient threshold. Users of the CIT to advance an investigation or interrogation should have confidence in a CI decision when the score indicates a probability of less than 10%.
Challenges
One of the problems in the field of polygraphy is that many examiners have a nearly singular focus on deception testing, and have developed little or no appreciation of what recognition testing offers. Consequently, polygraphy has missed important opportunities to serve the courts, and standard deception testing has been force-fitted to situations where recognition testing would be better suited. The CIT is an under-taught approach, and rarely used even in circumstances where it would have been the better choice. It has much to contribute to the field practitioner’s effectiveness, especially those serving in law enforcement, and they should be part of each examiner’s toolkit. Once having grasped the principles, methods, limitations and potential of the CIT, examiners will be better able to resolve their cases than if they remain wedded to one or two of the most popular approaches.
References
Barland, G.H. (1984). Research in electrodermal biofeedback with stimulus tests. Final report to the NSA, (Contract MDA904-83-M-1150).
Elaad, E., Ginton, A., & Jungman, N. (1992). Detection measures in real-life criminal guilty knowledge tests. Journal of Applied Psychology, 77(5), 757-767.
Lykken, D.T. (1959). The GSR and the detection of guilt. Journal of Applied Psychology, 43(6), 385-388.
National Research Council (2003). The Polygraph and Lie Detection. National Academies Press: Washington, DC.
Timm, H.W. (1982). Effect of altered outcome expectancies stemming from placebo and feedback treatments on the validity of the guilty knowledge technique. Journal of Applied Psychology, 67(4), 391-400.
Timm, H.W. (1989). Methodological considerations affecting the utility of incorporating innocent subjects into the design of guilty knowledge polygraph experiments. Polygraph, 18(3), 143-157.
Yokoi, Y., Okazaki, Y., Kiriu, M., Kuramochi, T., & Ohama, T. (2001). The validity of the guilty knowledge test used in field cases. Japanese Journal of Criminal Psychology, 39(1), 15-27.
Recommended Readings
Ansley, N. (1992) The history and accuracy of guilty knowledge and peak of tension tests. Polygraph, 21(3), 174-247.
Polygraph Place Bulletin Board
posted 05-09-2012 08:00 AM
