Nathan J Gordon, William L Fleisher
Page 26
T/13
T/15
T/18
T/17
2 6
T/15
T/26
T/25
T/22
T/30
2 7
T/13
T/25
T/23
T/21
T/25
3 1
Theft
T/14
T/18
T/24
T/18
T/14
3 2
T/7
T/6
T/15
T/9
T/11
3 3
T/3
T/9
T/20
T/12
T/12
3 4
T/2
T/15
T/20
T/15
T/9
3 5
D/ 6
D/6
D/6
D/ 3
D/0
3 6
T/12
T/20
T/25
T/29
T/7
3 7
T/4
T/16
T/18
T/22
T/13
3 8
T/6
T/13
T/17
T/13
T/12
3 9
T/9
T/23
T/24
T/24
T/21
3 10
T/0
T/7
T/12
T/5
T/19
3 11
T/ 5
T/9
T/7
T/7
T/1
Continued
11. THE VALIDATION OF THE FORENSIC ASSESSMENT INTERVIEW (FAINT)
181
TABLE 11.11 Scores and Assessments Made by Evaluators Using the FAINT Weighted
Scoring System Cont’d
Case
Type
Ground zero
Evaluator 1
Evaluator 2
Evaluator 3
Evaluator 4
3 12
T/13
T/20
T/7
T/17
T/10
3 13
T/7
T/27
T/25
T/29
T/19
4 1
Rape
D
D/ 5
D/ 6
D/ 1
D/0
5 1
Aggravated assault
D
D/ 6
D/ 9
D/ 9
D/ 9
6 1
Molestation
D
D/ 5
D/ 4
D/ 7
D/ 8
7 1
Molestation
D
?/5
D/ 3
D/ 4
D/ 5
8 1
Theft
D
D/8
D/8
D/11
D/9
8 2
T
T/16
T/17
T/14
T/17
9 1
Theft
D
T/15
T/13
T/10
T/12
9 2
T
D/1
D/6
D/3
D/1
10 1
Theft
D
D/ 4
D/3
D/0
D/0
11 1
Theft
D
D/7
D/15
D/9
D/1
11 2
T
T/19
T/30
T/22
T/16
11 3
T
T/21
T/18
T/27
T/14
12 1
Theft
T
T/17
T/15
T/10
T/1
12 2
D
D/ 5
D/ 5
D/ 7
D/ 4
12 3
T
T/34
T/34
T/35
T/29
13 1
Robbery
D
D/ 9
D/ 1
D/ 10
D/ 9
14 1
Sexual assault
D
D/2
D/ 1
D/ 4
D/ 4
15 1
Rape
D
D/ 2
D/1
D/0
D/ 1
16 1
Theft
T
T/12
T/17
T/14
T/14
17 1
Sexual touching
D
D/ 4
D/ 3
T/7
D/ 3
18 1
Rape
D
D/0
D/ 9
D/ 3
D/ 3
19 1
Theft
D
D/ 1
D/ 3
D/ 0
D/ 3
20 1
Sexual assault
D
D/ 3
D/ 2
D/2
D/ 3
21 1
Rape
T
T/12
T/8
T/13
T/8
22 1
Arson
T
T/21
T/25
T/21
T/21
22 2
T
T/14
T/13
T/10
T/10
22 3
T
T/13
T/19
T/12
T/12
22 4
D
D/ 1
D/1
D/ 1
D/ 4
Errors are reflected by the bold numbers.
204 total assessments: evaluator 1: 94% accuracy (48 correct/2 errors/1 inconclusive); evaluator 2: 94% accuracy (48 correct/3
errors); evaluator 3: 94% accuracy (48 correct/3 errors); evaluator 4: 96% accuracy (49 correct/2 errors).
182
11. THE VALIDATION OF THE FORENSIC ASSESSMENT INTERVIEW (FAINT)
TABLE 11.12 Evaluators’ Distribution of Scores Using Weighted Scores
Prediction of accuracy
Score
Evaluators’ determination
100
35
T
100
34
TT
100
30
TTT
100
29
TTT
100
28
T
100
27
TT
100
26
TT
100
25
TTTTTTT
100
24
TTTT
100
23
TTT
100
22
TTTT
100
21
TTTTTT
100
20
TTTTTT
100
19
TTTTT
100
18
TTTTT
100
17
TTTTTTTTT
100
16
TTTTT
&nb
sp; 97
15
TTTTTTDD
98
14
TTTTTTTT
94
13
TTTTTTTD
96
12
TTTTTTTTTD
95
11
TD
94
10
TTTTD
93
9
TTTTDD
92
8
TTDD
91
7
TTTTTTDD
89
6
TTDD
89
5
TTD
91
4
T
91
3
TDDD
95
2
DD
Continued
11. THE VALIDATION OF THE FORENSIC ASSESSMENT INTERVIEW (FAINT)
183
TABLE 11.12 Evaluators’ Distribution of Scores Using Weighted Scores Cont’d
Prediction of accuracy
Score
Evaluators’ determination
95
1
TTTDDDD
100
0
DDDDDDDDD
100
1
DDDDDDD
100
2
DD
100
3
DDDDDDDDD
100
4
DDDDDDDD
100
5
DDDDD
100
6
DD
100
7
DD
100
8
D
100
9
DDDDDD
100
10
D
100
14
D
100
18
D
100
19
D
TABLE 11.13 Single-Suspect Case Accuracy with the FAINT Weighted Scoring System
Ground zero
Truthful assessments
Deceptive assessments
Inconclusives
20 truthful
20 (100% accuracy)
0 (0% F/P)
0
52 deceptive
1 (2% F/N)
50 (96% accuracy)
1 (2%)
12 of 13 to be deceptive and 1 to be inconclusive, and one evaluator assessed 12 of 13 to be
deceptive and 1 to be truthful (Table 11.13).
From an analysis of all individual assessments, it appeared that an assessment score of a
þ8 or higher would be 100% accurate for determinations of truthfulness, and an assessment
score of a þ7 or lower would be 100% accurate for determinations of deception. Table 11.14
is a distribution of the evaluators’ numerical scores for the 72 assessments they made of the
15 single-issue suspect cases.
In reviewing the research in the UNISA study, the initial panel, using a 3-point scoring
scale, was able to accurately reproduce the results of the original FAINT interviewer. Their
blind evaluations resulted in a 92% accuracy of truthful suspects, 83% accuracy of the
deceptive suspects, and an overall accuracy of 88%, with only 5 of the 204 cases determined
to be inconclusive (2.5%).
184
11. THE VALIDATION OF THE FORENSIC ASSESSMENT INTERVIEW (FAINT)
TABLE 11.14 Evaluators’ Distribution of Scores and Predictability Using the Forensic Assessment Interview
Technique (FAINT) Weighted Scores
Prediction of accuracy
Score
Evaluators’ determination
100
25
T
100
21
TTT
100
19
T
100
17
T
100
14
TTT
100
13
TTT
100
12
TTTT
100
10
TT
Truthful
þ8
100
8
TT
Deceptive
þ7
100
7
D
100
5
D
100
3
D
100
2
DD
100
1
DD
100
0
DDDDDD
100
1
DDDDDDD
100
2
DD
100
3
DDDDDDDDD
100
4
DDDDDDD
100
5
DDD
100
6
DD
100
7
D
100
8
D
100
9
DDDDDD
100
10
D
Although this accuracy does not meet the 95% accuracy required for scientific accep-
tance, it greatly exceeds that of chance expectation, and when compared to the more tradi-
tional Behavioral Assessment Interview (BAI) designed by John Reid, the study clearly
shows that by using a system of numerical evaluation for quantifying observations, and
integrating other innovations in detecting deception into the interview process, inconclu-
sive results are greatly diminished (BAI had a 34% inconclusive rate).
11. THE VALIDATION OF THE FORENSIC ASSESSMENT INTERVIEW (FAINT)
185
When employee theft investigations were eliminated from the cases being evaluated,
accuracy improved to 100% when inconclusive results were not considered and 96% when
they were counted as errors. The FAINT interview process met the standards of scientific
acceptability in either case.
The research indicated that a numerical evaluation of a 0 or higher would be 88% accu-
rate in determinations of truthfulness, and a score of 5 or lower would be 89% accurate in
determinations of deception. These numbers changed when employee theft cases were
removed from the group to 100% accuracy for decisions of truth where a score of a þ3 or
higher was rendered, and 96% accuracy for determination of deception when a score of a
þ2 or lower was rendered.
By analyzing the predictability of a suspect’s answers with the known outcome of the
case, a weighted scoring system was devised and investigated. The derived weighted
criteria utilized by the second panel of blind evaluators resulted in 97% accuracy of truthful
suspects, and 90% accuracy of deceptive suspects, with only one inconclusive. Overall
accuracy for the 204 decisions made was 94%, with the inconclusive counted as an incorrect
decision.
The net effect of using the new “weighted” scoring system was a 5% increase in truthful
determinations, a 7% increase in deceptive determinations, and an overall increase in accu-
racy of 6%, when compared to the original 3-point scoring system previously being used.
This relatively small percentage improvement represents a considerable improvement in
statistical reliability and must be seen in this latter context.
Based on their weighted evaluations of the suspect’s behavior, predictability
scores were
established indicating that a total score of þ7 or higher would be 91% accurate for determi-
nations of truthfulness, and a score of þ4 or lower would be 91% accurate for determina-
tions of deception. In single-issue cases, scores could be adjusted, with a þ8 or higher
resulting in 100% accuracy for determinations of truthful, and a þ7 or lower resulting in
100% accuracy for determinations of deception.
The UNISA validity study of the FAINT interview determined that it was a valid proce-
dure for forensic interviewers to use in their search for truth. Accuracy is determined by the
validity and reliability of a procedure. Validity examines whether one is measuring what
they purport to be measuring, in this case, truth from deception. Reliability examines
whether the results are repeatable. This research clearly demonstrates that the Forensic
Assessment Interview Technique (FAINT) is an accurate method in detecting deception.
Notwithstanding this, the following caveat must be considered: the weighted method,
although increasing accuracy, resulted in much more positive scores, even for the “deceptive”
suspects. This presents no problem when the suspect pool is known to contain the “decep-
tive” suspect(s). For example, in employee theft case number 11, there were three suspects.
All of the blind evaluators properly identified the guilty suspect; however, they did this by
selecting the suspect with the lowest score in the group. One evaluator identified this person
as being deceptive with a score of þ15. Although this appears to be an extreme case, it is
important that the interviewer in such a case not rule out the possibility that all of the sus-
pects are innocent. In this situation, the FAINT system cannot be the be-all and end-all.
The person with the lowest weighted score should be reassessed using the nonweighted
3-point scale. The interviewer must also look “within” the case information to determine
whether a suspect with a very positive score deserves additional investigative focus. The
186
11. THE VALIDATION OF THE FORENSIC ASSESSMENT INTERVIEW (FAINT)
original FAINT interviewer had 100% accuracy in his assessments of these fifty-one suspects.
In fact, it was his ability to look “within” the case information, compare the suspects to each
other, identify the suspicions and comments of the suspects, and use this as additional assess-
ment criteria that allowed for the increase in his assessment accuracy.
The research from the UNISA study reconfirms that there are demonstrable differences
in the way truthful and deceptive suspects use nonverbal, verbal, and written behavior
when being interviewed by a structured technique.
Further, this study clearly indicated these differences are observable and quantifiable,
and that the weighted FAINT interview is one of the most statistically reliable tools that
can be effectively used to evaluate the culpability of an individual suspect (Figures 11.1
and 11.2). When presented with a case involving multiple suspects, FAINT will allow the
investigator to focus the investigation on the most likely suspect to have committed the
crime.
The following are actual Forensic Assessment Interviews to which you can apply the
weighted scoring system. After you have scored each interview, you can compare it to
our actual assessment, which will follow.
Case Study 1 deals with child molestation, Case Study 2 involves the theft of a diamond
ring, Case Study 3 involves the theft of a tote of controlled substances from a warehouse,
Case Study 4 is a man accused of having sexual intercourse with a minor female, Case
Study 5 is a grandfather convicted of masturbating in front of his minor granddaughter