Department of Defense Polygraph Institute Racial Bias Study · 2019. 6. 1. · A few words are in...
Transcript of Department of Defense Polygraph Institute Racial Bias Study · 2019. 6. 1. · A few words are in...
1
AntiPolygraph.orgE-mail: [email protected] • Telephone: 1-202-810-2105
Racial Bias in Polygraphy and Possible Cover-up: Cause for Concern?!The attached 22-page document was presented by Dr. Gordon H. Barland, then Director of Research, Department of Defense Polygraph Institute (DoDPI), to members of the federal polygraph research community at a group meeting in 1990. Shortly thereafter, the DoDPI director, who attended that presentation, requested that the documents be returned or that the portion which referred to racial bias studies conducted by DoDPI (the last nine pages) be destroyed. All of the above may be of great concern and interest because:
• Although not enough information was presented at the aforementioned meeting to determine any true significance (statistical or otherwise) of the results, these studies at face value suggest that innocent black polygraph examinees are more likely to be found deceptive on polygraph examinations (a false positive result) than are innocent white examinees;
• The polygraph formats examined in these studies are two of the most common “control” question polygraph formats used in the United States: the Zone Comparison Test (ZCT) and the Modified General Question Test (MGQT);
• The material was presented in a serious manner as a serious research effort by one of the leading polygraph researchers at the time (who is often now offered by the polygraph community as an expert in discussing that community’s views about polygraph countermeasures) to leading members of the federal agency community;
• These studies involved a large number of “test” subjects—approximately 1,100;• There appears to be little reason for DoDPI to have withdrawn this study, as it did at the
time, and for it never to have been published at all. Even if successive studies were properly done and resulted in differing outcomes and conclusions than those contained in this study, it was altogether improper for DoDPI to have simply simply “buried” this study. And if this one study (with clear and obvious negative implications for the polygraph community) was ignored, how can we be certain that other studies have not been similarly treated?
Contained with this document are the following troubling results:• Less than 60% of these tests resulted in a correct decision having been rendered for
all 1,141 subjects involved, black and white, guilty and innocent (p.14; p. 16 of the PDF file);
• Only 23.5% of innocent blacks were correctly classified as being non-deceptive, which was considerably less than the 36.9% of whites correctly classified (p.16; p. 18 of the PDF file). Applicants for federal agency positions should be particularly concerned with this: it suggests that if they tell the truth on a polygraph exam, they would have a roughly 63% chance (if white) and a 77% chance (if black) of either being found deceptive or having an inconclusive result. Either outcome would likely eliminate them from further consideration for federal employment;
• Only 14.6% of blacks were correctly classified as non-deceptive through use of the MGQT polygraph format, compared with 33.3% of white examinees (p.19; p. 21 of the PDF file);
• Nearly twice as many innocent blacks were found to be deceptive as were innocent whites through the ZCT polygraph format (51.9% vs. 28.6%).
A few words are in order regarding why this study should be considered carefully and why certain possible “straw-man” arguments that might be raised to discredit it should be carefully questioned and likely dismissed. First, in order to have any validity, a polygraph bias study (racial or otherwise) would have to be conducted such that examiners had no idea that a bias study was being conducted. Otherwise, in the case of a racial bias study, examiners would simply try to balance the number of blacks and whites who were found to be deceptive. Even if there existed some substantial number of false positives, they would be equally balanced, and there would appear to be no racial bias.It has been suggested informally by the polygraph community that these large numbers of exams were conducted by federal polygraph examiner trainees (students) during their course of basic instruction at DoDPI and that this is a weakness and perhaps a reason for discounting these results. In fact, quite the opposite is true: because the exams were training exams and not conducted for purposes of detecting possible bias, they are far less susceptible to being manipulated to disguise any bias that may exist. This is precisely how such a study should be conducted in this regard. The polygraph community has suggested that because these were trainees with limited experience, these results should be discounted. Nonsense! Any bias that might be exhibited by these or other individuals has little to do with the trade school instruction of a few weeks of polygraph training, but rather with the lifetime of impressions and influences that 30- to 50-year-old law enforcement/intelligence community officers and agents (those who constitute the polygraph classes) have amassed. Bias is likely to be reflected not in the technical operation of the polygraph instrument (calibration, etc.), but rather in the pre-test interview (question formulation, etc.) and the in-test phase question presentation to the examinee.In summary, does this study conclusively prove racial bias in polygraphy? No, it does not. But it raises some very troubling concerns that need to be thoroughly investigated.
###
II j
Analysis of DPI Studies
DoD Polygraph Institute 10 August 1990
Gordon H . . Barland, PhD
-11
II
I I I
List of DPI Studies
Mock crime
1. Multi- v. Single-Issue tests N = 100 2. Time Lag & Question Specificity study
(criminal half) N = 50 3. ZCT study N = 40 4. MGQT study N = 88 5. Motivation study N = 60
Mock Screening
1. Security Screening validation study N = 207 2. Time lag & question specificity study
(screening half) N = 50 3. CSP question study (old v new) N = 151 4. CNP-l (Scenarios) N = 86 5. CNP-2 (CQs) N = 120 6·. CNP-3 (Pretest) N = 183
All DPI Studies, Guilty & Inno Accy of decisions' (excluding incls)
100~------------------------------------------,
%
A c c y
60
60
40
20
o
% Cor Gu _ 92 64
% Cor In .~ 79 90 74 90
76 62
95
79
34
97
78
90
58
89
63
58
56
90
72
64
KEY
TITLE Gl.1
% Incl n ~ ~
1. Barland & Raskin (1975) 36 28 2. Podlesny & Raskin (1978) 20 15 3. Raskin & Hare (1978) 24 12 4. Rovner et al (1979) 24 12 5. Dawson (1980) 12 6. Hammond (1980) 32 25 7. Bradley & Janisse (19Bl) . 96 26 8. Szucko & Kleinmuntz(1981) 15 9. Ginton et al (1982) 2
10. Bradley & Ainsworth (1984) 16 11. Gatchel et al (1984) 14 43 12. Honts et al (1985) 31 19
/13. Honts et al (1986) 20 35 '/14. Foreman & McCauley(1986) 22 14 ./ 15. Honts et al (1987) 10 20 ;1'-16. Driscoll, Honts, & Jones 1987 20 10
17. Kircher & Raskin (1988) 50 6 /18. Patrick & Iacono (1989) 24 4 /19. Horowitz (1989) 15 20 --------20. Multi-Issue 77 19 21. 1/2 Time lag 40 8 22. ZCT 20 5 23. MGQT 44 5 24. Mot-l 30 27
Study 1 2 3 4 5 6 7 B 9
10 11 12 13 14 15 16 17 18 19
20 21 22 23 24
--
o
Decision accy, Criminal, Guilty Ss Others and DPI
82
, , , ,
82 , 73
,
85
, ,
64
74
r -40 80
Percent accy of dec
88
66
66
89
87
92
02
90
-
I j
96 i
!
06
i
i IH
i
I I
951
100
100
100
100
100
100
- i -------100 120
Study 1 2 3 4 5 6 7 a 9
10 11 12 13 14 15 16 17 16
Decision accy, Criminal, Innocent Ss Others and DPI
71 !
I 95 1 91 I
91
90
87 . Ii
66 ·47· I
i 65
66 100
70
83
lcse 78 I
1
100
93 , 54
J7 i i
I I I )79
I I 00
78 , ea
I 79 I I I I
o 20 40 eo 80 100 120
Percent accy of dec
Key Findings
1. DPI mock crime accuracy:
a. 21 % inconclusive b. 84% accuracy of decisions (excl incls)
2. DPI is just as accurate as other labs.
a. 19% inconclusive b. 86% accy of decisions
Accuracy of Mock Crime studies DPI (N = 338) v. Other Labs (N = 961)
'erect lOOr-----------------------------------~
80
60
40
20
o Other Labs . DoDPI
_ Correct ~ Incorrect
locI: Other = lU". DPI = 21';
Ii
i \
Accuracy of Examiner Decisions Criminal (N = 338) v. screenini (N = 793)
,....,~ . 100r------------------------------------
84
Criminal Screenini
_ Correct ~ Incorrect
Incl: Crlmlnlll = 21 ". Screeoin( = 23"
JUL 3 1 1990
Key Findi~gs
DPI mock crimes compared to non-DPI mock crime studies:.
3. ·Guilty Ss are just as easily detected.
4. Innocent Ss are just as easily cleared.
[But ... DPI has more inconclusives on innocent Ss (33%) than do other labs (21%)]
Key Finding
Mock Screening Studies
5. The. polygraph is less accurate in mock screening studies than in mock crime studies.
6. Most errors in mock screening are FNs.
Oerr.nQ~Df~ .. ~ ... -.... _- .-............. -.. -., .. -- ".' o· ••• -
NOV - 3 19m
l\ e C ll.l' a C)t () f ~'1: (~{l ~r I) -'!-l . stt J ( I e f'l ts (~: f ~-l ( ~ l J I t ,v ~llock~ crilnt~S at. .Dl'l
J <.10
~H) ....... '" ..... .... ... .... .......... ....... .. . .............. .
'''(J' I~
0-Ht Iucl WI'
, ." ... ~.,.~ .. "_"~'.N"""'" ....•.. ....... N .•. "_ ......... _.. .. . . .• 4.·._ .. _ ..... _ ............ _ ............ •... ,.,
Fu (~~ .. i 1 tJ"
~blde.nt.s
Gt1
(i(l
In 17
2'l la
I·:t ~\' I'
~ .• , I • l ",;, 5111
• • J • d .. · .. 5.l'l . . IU
I I
.~ .... _ .... ~ ... ~_s·-~-~t--·I· ........ _ .. _ .......... -._ ... -.... -............ . , . . 1 !
I
1\ C ~.! lJ r H. C)r ()f
i bt{oek
'~'l (; (~rr "\,.!! (;~, nil t.y criHH'~S ;··d. J) PI
I
I :::.u,,_ --j"~" :----- " . - .',"
l,iO . . .. ' ' .. , , .1. . . . .. .. .. .. .... ! .,
i I
'1(l .................. j ...................... .
i ! i ! .. ; ..... _. __ ......... -_ ..... ···· .. ·· .. ··_-_·· .. ·····_··1
14'ucu) ~.y ~J I , I f~~.'~:~\.~J~ t~:~ ~' .... : , i
,
I
lU
St.d.;.;! N~'1-'15 ~:; .. :~r.n (; i";:W! N:-:-:4 .. i (~ .. ,t-{. I
i . "'--'''''' -.. 'j ... .• ........... . ......... •... .. - ..... . .................. - .... _ ... " ... -'. - ................ .
.1 ~·~!l;·~·:I. III NIH
10
: 1
NOV - 3 1990 .
..-,.-
: ·1 .\
Ul
~Hl
:J:I
! I , , i i
_Qeart ... ' .. Of2-ru.~-... ~y- .. --.-. -·-·-·---·-~··----··-·····t····~ .....
l\CCll "ae~T' ()f ~"I.(~c~r.r. ,'vl Ill[l.()(~(;I} t 1\.1oclr c:.rirnes at DPJ
'{-'<:I .-..•.. _ ................................................ _ ................ -._ .... -.... .
(.;<::, " .......................................... .
r.<J ..................... ...................................... .... . ..... .
··10
:~o
10
F: <.:uU,,· . ., S~ \1'1 ·"l)(·~ •. 1.. <., ..••..
NIH Incl Dl
·1·1 ·:·t 1 .., .
Std.,." "'44ii G. :1511 I; ~'ac: Nr44 G. 44 I
.-... ... ... . . ....... --. .._... ..• - .. ," ... _. . ......... .
!
.. 1 .
NIH
n;J 1)·5·
... j ~ . ", f),5.
1.11
~IU
·Hi
NOV - ~ 1990
',0'
....
..
...
Race for MGQ-r andZCT
Percent Accuracy 100····· 90-ao-70· 60· 50·' 40· 30· 20··
10' o ... ::~
Caucasion Black
.... ~ ._ .. __ •.. ____ ..... __ ~. __ . ___ .. _. __ .. _ ... _0. __ .. "._., .. ___ . __ .•. _. .... -........... - ........ _., ... ,
·······--··T .. · .. · ...... .
Crt Inc Wrg Crt ... -----.---... -~-- " .............. _" ... _ ...... , .... _._-,-- -- ... -...... ---•... ~ .. -.. " .... - ,.," ..
60.4 26.8 12.8 82.5
56.4 30 13.6 80.5 •. _ ..... ______ ........ _ .••. _....... ........ . .......... ,.. .. _ ••.. _ ••• ' .•.•. __ ..... • ....... ...... • .......... _____ .... ,," ............ " ............... " .......... , • '·".M ................. .
Decision
_ Caucasion [:"L:::j Black
N: W-921 B-220
Wrg
17.5 19.5
1 .. 1 .'
Race for MGQT aWld zc-r Guil ty Examinee~i
Percent Accuracy . _____ ~._ ... *_~ .. __ .w._.~_~ __ •. ·_ .. ___ ·_¥_··· __ ·_··_· ... ···_· ___ .. ·-.. --.--..... , ... -.. " .... ,....... ..
100··' . 90-80-70 60·· SO-40·· 30 20'·'
1 ~ : ..... :.~~~~~_ ... •• .......... --.... 1" .. .
NOI INC 01 .. _.----_ ............ _--_._.... . ........ -._._ .. __ .. __ ... . .•• _-----_. __ .... -._ ••.. -+ -.-..... -.............. - .. - ... -.. . ........ _. • ....... ..
Caucasion Black
N: W-S01 6-139
7.3 4.3
19.8 20.1
72.9
75.5
Decision
_ Caucasi on [,~::~:J Blac k
. ... ::"~··1"·1 ~:;:; .......... :: ..
NDI 01
9.1 5.4
~~O.~~
94.6
IOV 131900
Race for MGQT and ZCT Innocent Exami nees
Percent Accuracy
70
60
50
40
30··
20
10
... - •.. _---_ ... _-_ .. _-_ .. _---_ .. _._ ........... --_ ..... __ ... _ ... "'''''''''-'' ................ -. ...... . .... -....... _........... ..... . ... .
o NOI INC 01 NOI 01
_ ... _ ..... _-. ·_-_·_--···· .. 1 -------. --.-.~--- .. -··--·-·-··I··· .. -· .... _ .... --- .- ..... -..... ..... .. .......... -_._ ... -.. ..... .... ...................... .
Caucasion .. 36.9 Black 23.5
40 46.9
23.1
29.6
61.5
44.2 ...... ___ .... _. __ ..... _ .•... l .... _ .• _ ................... __ ...... _ ............. _ ....... __ .... _-..................... .
Decision
_ Caucasion [~::: .. ] Black.
N: W-320 B-81
38.5
!55.8
'IOV i 3 f99C
Race for MGQT
Percent Accuracy 100 .
A--·---··-----·-··-·-----·--·------··---·--------···-· -.... --------... . . . ..... .
90 80 70--
eo 50 40 30 20 10 0-·
Crt Inc Wrg Crt ,.-----.. -----------.-t-.. ------------ --.------.... -- -.----J ----.-----.- .... -.--.- ..... . --. ..-.. ... .. .-.... -.. - ...... .
Caucasion 59.7 Black 53
N: W-514 6-115
25.7 35.7
14.6
11.3
Decision
80.4 82.4
_ Caucasian [~~:-:~-:l Black
Wrg
·19.6
17.6
'v i :3 ·1990
100 90 80 70 60 50 40 30 20 10 o
Race for MGQT Guil ty Examinees
Percent Accuracy ~.r-.. -"--------.-'.----'.-"-"--'.--" ..... _ ..... _ .... -_._ .. - ................. - .... __ .. _ ........... - ...
=-__ ;.A;_ • ___ ·:1::·=·~-·:::-·-·::-~:i~~:~~1J.. .. NOI 01 NOI . INC 01 ,._--_ ... _.- ... _---_._ .. _--+_ .. _-_.".--. __ .... ------_ ... _ ... - ........... --... ...... .. ........... _ .......... _._.'..... " .... ,_ ............ _...... . ... ..
Caucasion Black
7.1 17.9 75.1 4.1 21.6 74.3
l._. __ , ____ ,., •• __ • __ ........... .1 , ...... _._, __ .• _ .... __ .. _ .. _ - ... _-_ ••• _ .••. __ .. -1 _ •. _ ••.•. --._ ...... -, ........... - ... , ..
Decision
_ Caucasion L~-:::::] Black
N: W-325 e-74
8.6 5.2
91.4
94.8
70
60 .
50
40
30··'
20 -.
10
o
Race for MGQT Innocent Exami nees
Percent Accuracy __ '~_M _______ "_"_' ___ '_ •• ___ ... ______ .... __ ...... _M ••• _ •• • ......... ~ •• ,,-~, ___ ~ •••• _,~ •••••••• _ •• __ ................. " ................ _...... • ..... -.... '". -.--..... - •• -- ••••
NOI INC 01 NDI 01 .----.--... -.-.-.... - ... -t-----------~-------+----· Caucasion
Black
33.3 14.6
____ . _________ .. ___ .•.... L. _________ ..•. __
N: W-189 B-41
39.2 61
27.5 54.8
24.4 37.5
45.2
62.5 ... ___ ... __ ._ ....... , • ~ .. __ .• _ •.. , ....... ".0 ..... .. .• _ ..•• _ ..• _ ....... ' ~ .•....•.. ", ..••..• __ ••.•..•• -.•.. -...• - •. -••
Decision
_ Caucasi on [::~~~] B I ac 1\
i 3 1930
100 90 80 70 60 50 40 30 20 10 o
Race for. zc·r Percent Accuracy ... - ... __ ._-------_._ .. _ .. _--_._ .. _ .. _. _ .... __ .... _._ .. _ ..• _ ........... " .......................... _ ...... " .......... " .... "
J-. _ .... -._.-.. _ ..... ".- ..•..• - .... -_ ... , ..... " .. " .. -. ! :
Crt Inc . Wrg Crt Wrg ·--------f .-.---.--+-------f-.-----;.---.-. -.,,-... - ........... - ............. _ .. --.......................... _ .. _ ........ _ ... .
Caucasion 61.2 28.3 10.6 85.3 14.7 Black 60 23.8 16.2 7~.8 21.3
.. _._ .... _ .. _.1.- ." ...•. _ ... ___ .• _ .. _. _.. ... . ...
Decision
_ Caucas i on L;:':; .. ::l B I ac k
N: W-407 e-105
· ... v : 3'19'30
100 90 80 70 60 . 50·· 40 30 20 10-o
Caucasian Black
Race for zc"r Guilty Examinees
Percent Accuracy ..r--------.---.--.-.-----.-----.-.. -.. ---............... _.--.. _ ....... -.................... -...... -.......... -.
NOI
7.6 4.6
INC
22.1 18.5
01
. . __ = :,. __ :.::: ~:.:.i~V· . NOI 01 ... --_. __ .- .•..... _......... _ .... _. __ ... _ ........ - ... -........................... . 9.8 90.2 5.7 94.3
l. _____ . ___ L-..._. __ •.. ___ ..L ___ ._ .• L _______ L. ___ •....•.•.••..••...••......... _ .....••.•.• __ ••...•...•.•.••....•......•.
Decision
_ Caucasian t~~~] Black
N: W-276 a-65
'IV i 31930
•.•• ;0
NOI
Caucasion 42
Black 32.5
N: W .. 131 D-40
INC
41.2
32.5
Race for. ZCT Innocent Exarninees
01
16.8
35
_ .. -..... _ ... _ .. ,
Decision
NDI
71.4
48.2.
.. Caucasion [.'·:.:1 BlacJe.
...... --'
DI
28.6
51.9
"i3!9S0