harcourt_against_prediction_powerpoint

Professor Ernest Burgess, University of Chicago
Ernest Burgess: Concentric Zones
Psychiatric Personality Type in Relation to Parole Violations
Violation Rate by Institutions
Personality Type
Pontiac
Menard
Joliet
All persons.........................
22.1%
26.5%
Egocentric..........................
24.3
25.5
38.0
Socially Inadequate............
20.0
24.7
22.6
Emotionally Unstable........
8.9
*
16.6
28.4%
* Number of cases insufficient for calculating percentage.
Social Type in Relation to Parole Violation
Violation Rate by Institutions
Social Type.
Joliet
Pontiac
Menard
All persons.........................
22.1%
26.5%
28.4%
Hobos.................................
Ne’er-do-well.....................
Mean Citizen......................
Drunkard............................
Gangster.............................
Recent Immigrant..............
Farm Boy...........................
Drug Addict......................
14.3
32.8
____
37.5
22.7
36.8
11.0
4.3
46.8
25.6
30.0
38.9
23.2
16.7
10.2
66.7
70.5
63.0
9.5
22.7
24.1
4.0
16.7
83.3
States Using Parole Prediction Instruments
30
Number of States Using Prediction Tool
25
20
15
10
5
0
1900 1905 1910 1915 1920 1925 1930 1935 1940 1945 1950 1955 1960 1965 1970 1975 1980 1985 1990 1995 2000 2004
Year
19
79
19
81
19
83
19
85
19
87
19
89
19
91
19
93
19
95
19
97
19
99
20
01
20
03
Proportion of states using
risk assessments of states
with parole
Use of Risk Assessments Among States with Parole
0.8
0.7
0.6
0.5
0.4
0.3
0.2
0.1
0
Year
Rapid Risk Assessment for Sexual
Offender Recidivism (RASOR)
1. Prior Sex Offenses (not including index offense)
None
1 conviction or 1-2 charges
2-3 convictions or 3-5 charges
4+ convictions or 6+ charges
Score
0
1
2
3
2. Age at Release (current age)
more than 25
less than 25
0
1
3. Victim Gender
only females
any males
0
1
4. Relationship to Victim
only related
any non-related
0
1
GRAPH I.A.1: An Economic Model of Racial Profiling
African-American
Motorists
Time 3: Racist
Policing
10
Y-Axis: Offending Rate
(Percent of Racial Group Offending
9
8
7
Time 1: Color-Blind
Policing
6
5
4
3
2
Time 2: Efficient Policing
White Motorists
1
2.5
5
7.5
10
12.5
15
17.5
20
22.5
25
X-Axis: Internal Group Rate of Searches
(Percent of Searches Conducted Within Racial Group)
GRAPH: Racial Profiling with Different Elasticities
Time 2: Efficient Policing
10
Y-Axis: Offending Rate
(Percent of Racial Group Offending
9
8
Time 2: Societal
Average Offending
Rate of 7%
7%
6.4%
Time 1: Societal
Average Offending
Rate of 6.4%
6
5
Minorities
Majorities
4
3
Time 1: Color Blind Policing
2
1
0.25
0.50
0.75
1
1.25
1.5
1.75
2
2.25
X-Axis: Internal Group Rate of Searches
(Percent of Searches Conducted Within Racial Group)
TIME 1: ORIGINAL ASSUMPTIONS
Total
Minority
Majority
City population
1,000,000
200,000 (20%)
800,000 (80%)
Police searches
10,000
2,000 (20%)
8,000 (80%)
Searches as %
of relevant
population
1%
1%
1%
Offending rate
6.4%
8%
6%
Successful
searches
640
160 (8% of 2,000)
480 (6% of 8,000)
Number of
offenders
64,000
16,000 (8% of 200,000)
48,000 (6% of 800,000)
TIME 2: ASSUMPTIONS
Total
Minority
Majority
City population
1,000,000
200,000 (20%)
800,000 (80%)
Police searches
10,000
4,000 (40%)
6,000 (60%)
Searches as % of
relevant
population
1%
2%
0.75%
Offending rate
7%
7%
7%
Successful searches
700
280 (7% of 4,000)
420 (7% of 6,000)
Number of offenders
70,000
14,000 (7% of
200,000)
56,000 (7% of
800,000)
Figure 4.3: Actuarial Methods as Applied to Criminal Sentencing
40
Time 1: No
Profiling
36
32
Y-axis: Group Offending Rate
28
Recidivists
24
20
Total Social
Offending Rate at
Time 2
Time 2:
Profiling
16
Total Social
Offending Rate at
Time 1
12
8
4
Ordinary Citizens
-100% -75% -50% -25%
0
+25% +50% +75% +100%
X-axis: Change in Expected Length of Sentence
Figure 5.1: Basic Model
Higher-Offending
Minority Group
10
9
Y-Axis: Offending Rate of Group
8
7
6
5
4
Time 1: Carceral Population with No Profiling:
Carceral Population at Time 1:
3
2
Minority,
25%
Lower-Offending
Majority Group
Minority,
75%
1
10
20
30
40
50
60
70
80
90
X-Axis: Distribution of Searches Between Groups
100
Figure 5.2: Criminal Profiling at Time 2
Higher-Offending
Minority Group
10
9
Y-Axis: Offending Rate of Group
8
7
6
5
4
Time 2: Carceral Population with 40/60 Profiling:
3
2
Carceral Population at 40/60:
Lower-Offending
Majority Group
Majority,
53%
1
10
20
30
40
50
60
70
Minority,
47%
80
90
X-Axis: Distribution of Searches Between Groups
100
1997
1994
1990
1986
1982
1978
1974
1970
1966
1962
1958
1954
1950
1946
1942
1938
1934
1930
1926
Percent African-American
FIGURE 1
Racial Distribution of Admissions to State Prisons, 1926–1997
60
50
40
30
20
10
0
FIGURE 2
Percent of US Adult Population in State or Federal Prisons or in
Local Jails, by Race and Gender, 1985–1997
8%
Black males: 6 .8 4 %
6%
4%
2%
0%
1985
White males: 0 .9 9 %
Black females: 0 .4 9 %
White females: 0 .0 8 %
1987
1989
1991
1993
1995
1997
Profiling in Terrorism
Context?
Racial Profiling as DCTM
“Just smart law enforcement” or “just crazy”?
Proponents are right that it will immediately
increase likelihood of detection
(though low base-rate event)
Opponents are right that it will likely produce
substitution effects over time
Studies of DCTM
Landes 1978: metal detectors reduce airplane hijacking
Cauley and Im 1988: metal detectors increase other types of
terrorist attacks
Enders and Sandler 1993: metal detectors increase
assassinations and hostage-taking events; retaliatory
strikes like Libya 1986 cause intertemporal substitition;
fortification of embassies produces substitution toward
assassinations.
Dugan, LAFree and Piquero 2005: not looking at substitution,
but finds RAT; contagion; but no effect on terrorismrelated hijacking attempts.
U.S. Aircraft Hijackings, 1961-1976
80
60%
70
1973: Installation of Metal Detectors
in United States Airports
Number of Skyjackings
60
40%
50
40
30%
30
20%
20
10%
10
0
0%
1968
1969
1970
1971
1972
1973
Number of Skyjackings
1974
1975
1976
1977
Skyjackings as % of All Terrorist Acts
1978
1979
1980
Skyjackings as % of Total Terrorist Acts
50%
Key Issue: Substitution Effects

Substitution of non-profiled offenders for
profiled offenders
This raises issue of comparative elasticities.

Substitution of type of attacks toward those
that are less susceptible to RP detection