• If there was no specific nonverbal behavior each question was assessed based on the
verbal answer alone and given a score of þ1 if perceived truthful, and a –1 if perceived
deceptive.
• The same score was given if the nonverbal behavior was in agreement with the verbal
behavior.
• If there was a disagreement between nonverbal and verbal behavior, the question was
given a score of 0.
*Ground zero truth for interviewees determined to be deceptive was confirmed by confession, and for those
determined to be truthful by the confessions of deceptive interviewees in the same cases.
Effective Interviewing and Interrogation Techniques
167
# 2011, Elsevier Ltd.
168
11. THE VALIDATION OF THE FORENSIC ASSESSMENT INTERVIEW (FAINT)
TABLE 11.1 Breakdown of Suspects and Categories of Crimes Used in the Study
Suspects (51)
Crimes (22)
36
Theft (10)
4
Sexual assault/rape (6)
5
Molestation (3)
4
Arson (1)
1
Aggravated assault (1)
1
Robbery (1)
The four evaluators correctly solved 86% of the twenty-two cases they reviewed, exclud-
ing “inconclusives” (which are defined as suspects whom they were unable to classify as
truthful or deceptive, because of “insufficient” data), and 83.5% where “inconclusives”
were included and viewed as errors (Table 11.2).
The data involved 31 truthful and 20 deceptive suspects, requiring 204 assessments to be
made (124 truthful assessments and 80 deceptive assessments). Of the 124 assessments of
truthful suspects, 3 were assessed to be inconclusive, and 111 were correctly assessed as
truthful (Table 11.3). Thus, there was a 92% accuracy rate in the assessment of truthful suspects. Of the 80 assessments of deceptive suspects, 2 were assessed to be inconclusive, and
65 were correctly assessed to be deceptive. Thus, there was an 83% accuracy rate in the
assessment of deceptive suspects. Overall accuracy for the 199 assessments made was
88%, with only 5 of the 204 assessments deemed inconclusive (2.5%).
Utilizing the traditional FAINT 3-point scoring scale, the evaluators had an 8% false pos-
itive rate and a 17% false negative rate. Qualitatively, the errors made were more likely to
be false negatives: deceptive suspects assessed to be truthful. This is the same result as in
the earlier cited Behavioral Analysis Interview study [2] and is exactly the opposite of most polygraph errors, which are more likely to be false positives, truthful suspects assessed to
be deceptive. Given that the error rates between these two processes (FAINT and
TABLE 11.2 Cases Solved by Evaluators Using FAINT 3-Point Scoring System
CASES SOLVED EXCLUDING INCONCLUSIVES: 86%
Evaluator:
1
2
3
4
88%
82%
82%
92%
CASES SOLVED INCLUDING INCONCLUSIVES: 83.5%
Evaluator:
1
2
3
4
87%
74%
83%
91%
11. THE VALIDATION OF THE FORENSIC ASSESSMENT INTERVIEW (FAINT)
169
TABLE 11.3 Accuracy of Evaluators Using FAINT 3-Point Scoring System
Ground zero
Truthful determinations
Deceptive determinations
Inconclusives
124 truthful
111 (92% accuracy)
10 (8% F/P)
3
80 deceptive
13 (17% F/N)
65 (83% accuracy)
2
Polygraph) are in opposite directions, it may further support the use of the FAINT inter-
view assessment as the pre-test interview tool for polygraph examinations to give a more
accurate overall assessment of a suspect’s truthfulness.
Table 11.4 shows the actual scores and assessments of the four evaluators of each sus-
pect, as well as each evaluator’s accuracy.
Analysis of all of these individual assessment scores appears to indicate that an assess-
ment score of 0 or higher would result in 88% accuracy for determinations of truthfulness,
and an assessment score of –5 or lower would result in 89% accuracy for determinations of
deception.
Table 11.5 shows the distribution of the evaluators’ numerical scores in the 51 cases they assessed. Of these 22 cases, there were 10 involving employee thefts ranging from a single
suspect, to 1 case of employee theft with 13 suspects. Multisuspect employee theft cases
TABLE 11.4 Scores and Assessments Made by Evaluators Using FAINT 3-Point Scale
Case
Type
Ground zero
Evaluator 1
Evaluator 2
Evaluator 3
Evaluator 4
1 1
Theft
D
D/ 7
T/2**
D/ 6
D/ 7
1 2
T
T/11
?/0
T/8
T/12
1 3
T
T/14
T/5
T/4
T/3
2 1
Theft
T
T/19
T/10
T/16
T/7
2 2
T
T/16
T/2
T/14
D/ 6
2 3
T
T/16
T/5
?/5
T/ 2
2 4
D
T/11
D/ 6
T/11
T/ 4
2 5
T
D/ 2
T/9
?/5
T/5
2 6
T
T/15
T/13
T/13
T/7
2 7
T
T/15
T/17
T/11
T/6
3 1
Theft
T
T/18
T/ 8
T/5
T/ 2
3 2
T
T/7
T/ 4
T/5
T/8
3 3
T
T/7
T/8
T/4
T/ 2
Continued
170
11. THE VALIDATION OF THE FORENSIC ASSESSMENT INTERVIEW (FAINT)
TABLE 11.4 Scores and Assessments Made by Evaluators Using FAINT 3-Point Scale Cont’d
Case
Type
Ground zero
Evaluator 1
Evaluator 2
Evaluator 3
Evaluator 4
3 4
T
T/21
T/22
T/10
T/10
3 5
D
T/9
T/5
T/4
D/ 8
3 6
T
T/15
T/10
T/18
T/12
3 7
T
D/ 4
T/ 2
T/ 3
T/ 6
3 8
T
T/12
T/1
> T/6
T/15
3 9
T
T/32
T/19
T/17
T/20
3 10
T
T/2
D/ 10
T/2
T/1
3 11
T
T/18
T/ 4
D/ 5
T/5
3 12
T
T/19
T/17
T/9
T/13
3 13
T
T/21
T/6
T/14
T/13
4 1
Rape
D
D/ 12
D/ 11
D/ 8
D/ 5
5 1
Aggravated
D
D/ 5
D/ 5
D/ 5
D/ 7
assault
6 1
Molestation
D
D/ 9
D/ 8
D/ 8
D/ 11
7 1
Molestation
D
D/ 9
D/ 10
D/ 16
D/ 11
8 1
Theft
D
D/ 1
T/9
D/ 1
D/ 9
8 2
T
T/9
D/0
T/2
T/7
9 1
Theft
D
T/8
T/11
T/7
T/3
9 2
T
D/5
D/0
D/ 5
D/ 4
10 1
Theft
D
D/ 3
D/ 3
D/ 4
D/ 4
11 1
Theft
D
D/ 1
T/11
D/3
D/1
11 2
T
T/12
T/16
T/13
T/13
11 3
T
T/16
T/4
T/15
T/4
12 1
Theft
T
T/13
T/1
T/ 4
T/ 3
12 2
D
D/ 20
D/ 18
D/ 18
D/ 20
12 3
T
T/13
T/19
T/22
T/14
13 1
Robbery
D
D/ 6
D/ 16
D/ 14
D/ 18
14 1
Sexual
D
D/ 5
D/ 8
D/ 3
D/ 8
assault
15 1
Rape
D
D/ 4
D/ 5
?/2
D/1
16 1
Theft
T
T/13
T/13
T/14
T/3
Continued
11. THE VALIDATION OF THE FORENSIC ASSESSMENT INTERVIEW (FAINT)
171
TABLE 11.4 Scores and Assessments Made by Evaluators Using FAINT 3-Point Scale Cont’d
Case
Type
Ground zero
Evaluator 1
Evaluator 2
Evaluator 3
Evaluator 4
17 1
Sexual
D
D/ 6
D/ 4
?/1
D/ 6
touching
18 1
Rape
D
D/ 6
D/ 8
D/ 4
D/ 5
19 1
Theft
D
D/ 9
D/ 8
D/ 12
D/ 5
20 1
Sexual
D
D/ 1
D/ 7
D/ 6
D/ 6
assault
21 1
Rape
T
T/7
T/13
T/9
T/10
22 1
Arson
T
T/18
T/15
T/19
T/22
22 2
T
T/9
T/ 2
T/10
T/3
22 3
T
T/3
T/ 1
T/10
T/18
22 4
D
D/ 11
D/ 8
D/ 2
D/0
Figures appearing in red indicate errors in assessments.
204 assessments: evaluator 1: 88% accuracy (45 correct/6 errors); evaluator 2: 88% accuracy (44 correct/6 errors/1 inconclusive); evaluator 3: 89% accuracy (42 correct/5 errors/4 inconclusive); evaluator 4: 92% accuracy (47 correct/4 errors).
TABLE 11.5 Evaluators’ Distribution of Scores and Predictability of Determinations
Predictability of a
Evaluators’
Ground zero truth
Predictability of a
truthful determination
scores
outcome
deceptive determination
100
32
T
0
100
22
TTT
0
100
21
TT
0
100
20
T
0
100
19
TTTTT
0
100
18
TTTTT
0
100
17
TTT
0
100
16
TTTTT
0
100
15
TTTTTT
0
100
14
TTTTT
0
100
13
TTTTTTTTTTT
0
Continued
172
11. THE VALIDATION OF THE FORENSIC ASSESSMENT INTERVIEW (FAINT)
TABLE 11.5 Evaluators’ Distribution of Scores and Predictability of Determinations Cont’d
Predictability of a
Evaluators’
Ground zero truth
Predictability of a
truthful determination
scores
outcome
deceptive determination
100
12
TTTT
0
93
11
TTDDDD
7
94
10
TTTTTTT
6
92
9
TTTTTDD
8
91
8
TTTD
9
90
7
TTTTTTD
10
90
6
TTT
10
90
5
TTTTTTTTTD10
90
4
TTTTD10
89
3
TTTTDD11
88
2
TTTTDD
12
88
1
TTTDDD
12
88
0
TTTD
12
85
1
DDDDT15
25
2
DTTTTTT75
19
3
DDDTT
81
17
4
DDDDDDTTTTT83
11
5
DDDDDDDDTT
89
9
6
DDDDDDDDTT91
6
7
DDDD
94
7
8
DDDDDDDDDT
93
5
9
DDDD
95
6
10
DT
94
0
11
DDDD
100
0
12
DD
100
0
14
D
100
0
16
DD
100
0
18
DDD
100
0
20
DD
100
11. THE VALIDATION OF THE FORENSIC ASSESSMENT INTERVIEW (FAINT)
173
present variables not present in other types of criminal investigations. In multisuspect
employee thefts there may be many employees who have committed previous thefts from
their employer, but are innocent of the crime that is being investigated. These prior thefts
often result in attitudinal changes that cause innocent employees to emit nonverbal, verbal,
and written behavior generally associated with deception. This situation, although interest-
ing, remains an area for future research.
When these multisuspect employee thefts are eliminated from the study, there remain 15
cases involving 18 suspects. Among these were 6 sexual assault/rape cases, 1 aggravated
assault case, 3 sexual molestation cases, 3 theft cases, 1 robbery case, and 1 arson case that
had 4 suspects. Among this cohort, 13 of the suspects were guilty of the crimes being inves-
tigated and 5 were innocent.
Accuracy of the four evaluators for this subset of FAINT assessments was 100% when
“inconclusives” were eliminated and 96% when they were included and counted as errors
(Table 11.6). There were no false positives or false negatives. Although the truthful suspect group may be too small for statistical conclusions to be drawn, it is important to note that
all 5 were assessed truthful by all evaluators. As to the larger, more statistically significant
deceptive group, three evaluators assessed all 13 to be deceptive, and one evaluator
assessed 11 to be deceptive and 2 to be inconclusive.
An analysis of these individual assessments of these single-issue suspect cases indicated
that an assessment score of a þ3, or higher, would be 100% accurate for determinations of
truthfulness, and an assessment score of a þ2, or lower, would be 96% accurate for determi-
nations of deception.
Table 11.7 presents a distribution of the evaluator’s numerical scores for the 72 assessments made of the 15 single-issue suspect cases.
A meta-analysis was performed on the responses of both truthful and deceptive suspects
to determine whether certain questions elicited answers that needed to be weighted, given
that they might be more consistently predictive of truth or deception.
This proved a fruitful line to pursue. Some questions indeed showed lesser value in pre-
dicting truth or deception, whereas others appeared to have greater value. Based on that, an
analysis was performed that confirmed that certain questions did have a higher correlation
in predicting outcomes.
For example, when asked, “Who would you suspect?” 20 of the 41 suspects asked the
Nathan J Gordon, William L Fleisher Page 24