A GANG RISK CLASSIFICATION SYSTEM FOR JAILS
by
George W. Knox, National Gang Crime Research Center
INTRODUCTION AND BACKGROUND
The Project GANGECON research task force was formed in 1994 and involved invitations to a number of gang researchers and experts to pledge their support and labor in a unique knowledge development project. The data used here come from the Project GANGECON study. The Project GANGECON report is available from the National Gang Crime Research Center.
One of the goals of the Project GANGECON research task force was to ensure that gang members were studied in a variety of social contexts. The universe became easy to define when we asked the general question: where could we most likely find some gang members today? The first answer was: custody, those in jail, etc. Thus, at an early stage in the research process various social contexts were identified, contacted, and persuaded to cooperate with our research mission.
The types of social contexts used in the Project GANGECON task force research therefore consists of the following:
(1) Juvenile correctional facilities in the midwest and west (California, Michigan, and Ohio),
(2) Jails in Illinois and Iowa,
and (3) and a private day jail program as well as a courtroom lockup setting.
The study reported here therefore includes N = 1,015 gang members surveyed in some form of custody.
A SECURITY THREAT GROUP ANALYSIS
The term "security threat group" is another modern day expression for what covers "criminal gangs". The concept of "security threat group" (STG) or "security risk groups" is one applied mostly to correctional systems. The value of carrying out such an analysis here is that the national gang sample in the present research does include a number of different correctional settings, juvenile and adult.
A new organizational position exists in most correctional facilities today,
and it is the role provided by the person designated as the STG-coordinator.
This role has different names
as well, but basically means the same, it includes
an intelligence function about gangs. This new role has arisen in corrections
because of the new problems that gangs create in correctional institutions.
Nationwide, this new occupation in criminal justice is just now getting organized
to be able to share and learn from their common experiences. One of the duties
of the STG-coordinator is to identify gang members --- or security threat group
members --- and to help keep the peace by limiting the damage that STG's can do
behind bars.
It is often necessary for the STG-coordinator to develop a classification system along dimensions of risk. In a correctional setting this is probably going to be mostly the risk or threat of violence to staff and inmates from gang or STG members. Admittedly, the very term "STG" is somewhat broader than the meaning and the definition of criminal gang, but in practice they amount to much of the same thing. The reason for a classification system for STG members is to be able to physically isolate them into manageable groups. That is, the very high security unit would be reserved for only the most "hard core" or those most prone or most predicted to be a threat of violence to staff or other inmates.
THE VALUE OF A CLASSIFICATION SYSTEM FOR MANAGING GANG PROBLEMS
In the correctional setting, once a person is identified as an STG member, restrictions will have to be placed on that person because they are a higher risk. We are suggesting that a way to establish this criteria for increased security levels is to use a classification system based on a Security Threat Group Analysis. This would be a way of validating the classification system used, to make sure it "works" on the target population. "Works" here means simply enough: is the type of threat analysis applied for classification generally accurate in differentiating various "threats".
Higher security means more staff resources in a correctional setting. Some of the types of restrictions that STG members face in correctional facilities could include: strip searches, frequent cell or living unit "shake downs" for contraband and weapons, cancellation of good time/work time credits, visiting restrictions (i.e., no contact visits), monitoring of mail and telephone communictions, monitoring of the inmate's commissary account, and placing STG members in a special form of segregation. This is necessary because of the large rise in serious security threats behind bars today: riots, killings, forcible gang recruitment, assaults on staff, and a host of other problems that arise when dealing with the gang as a group identity.
Some jurisdictions now use what is called a "performance based housing" approach for STG management. It begs for a classification system that can be validated and used in other jurisdictions. In the performance based housing approach, only the highest risk or highest threat level of gang offenders are targetted for isolation from the general population of inmates, they are also separated from the "non-problem" gang members. There are higher restrictions in the high risk gang living unit, but time spent there is dependent on the behavior of the individual gang member. In this type of administrative segregation the "hard core" gang member receives only what constitutes legally mandated activities, they receive no privileges, it is "hard time".
Most STG coordinators also maintain gang intelligence file systems. But most STG coordinators do not carry out a quantitative classification system or threat analysis of the type we will illustrate here. However, it is a direction that STG coordinators can be expected to take in the future.
THE SIMULATION SCENARIO
What if our current sample of N = 1,015 gang members was really our current inmate STG population and we wanted to be able to "triage" or establish a classification system that could effectively manage this population? In otherwords, from a practical point of view: could we develop a simple but effective classification system (i.e., a threat analysis) that effectively managed risk in terms of its ability to show consistent patterns of risk along the gang classification system. It would have to be able to differentiate between the "least risk" and the "highest risk" or the hard core gang members.
Some previous research along these lines has suggested the merits of a
"gang risk continuum", based on behavior and human life span development
milestones of gang risk behavior.
In fact, a logical extension of this is that
gang membership itself is something that can be effectively predicted.
The simulation scenario explored here is whether a simple classification system can be used to achieve remarkable results in differentiating various gang member behavior risks. Thus, we will explain the simple formula below, and then actually apply this "threat analysis" rating system or classification system to the entire sample, and then see how well we would be able to "manage" this population if they were our inmates.
THE SIMPLE PROCEDURE: AN ADDITIVE SCALE
The threat analysis scale developed here is a gang classification system that varies from a low of zero to a high of three. The higher the numerical score, the higher the gang risk. Everyone begins by having a score of zero. Then three questions are asked:
(1) Are you still active in the gang, if yes, add one point;
(2) Have you ever held rank or a leadership position in the gang; if yes, add one point;
(3) Have you been in the gang for 6 or more years; if yes, add one point.
These three simple additive equations were used to create the computer simulation of a gang classification system. The current procedure involved adding existing data responses in a computer calculation that created a new variable GANGRISK. But this could have just as easily been done by interviewing each member as they enterred the correctional facility, or assessing their records.
Thus, the lowest score one could get on this scale would be "zero" (lowest threat) and the highest score one could get would be "three" (highest threat). Applied to the current sample of N = 1,015 gang members, the results of this procedure were as follows in Figure 1.
FIGURE 1
What We Get From Applying The Three Questions
On A National Sample of N = 1,015 Gang Members
Gang Risk/Threat Analysis Classification
Level 0 Level 1 Level 2 Level 3
Lowest Highest
Threat Threat
Sample: N=143 N=350 N=365 N=157
Percent: 14.1% 34.5% 36.0% 15.5%
Joined Gang? Yes Yes Yes Yes
Current member? None 1/3 2/3 All
(0%) (30%) (70%) (100%)
Ever hold rank? None 1/4 3/4 All
(0%) (25.8%) (78%) (100%)
Gang Tenure: <=5 years Varies Varies >=6 years
Figure 1 therefore breaks down the sample into four groups, Level Zero through Level Three. The hypothesis we are making here is that the Level Zero gang members will be significantly differentiated along the gang risk/threat levels consistently such that for all types of "gang risk" the lowest risk or threat will be found for Level Zero gang members and the highest will be found for Level Three gang members.
We have used this more simple classification method because it is most easily adaptable to the real world of corrections. We do have more effective models that would be preferred strategies, however. The reader is reminded that we are simply trying to illustrate the value of such an approach for the purpose of gang management in a simulation procedure.
Thus, our hypothesis is that the Level Zero group will be an "easy to handle" type of group, while the Level Three group will be a "very difficult to handle" group. We can now turn to our analysis to see if this simple procedure yields an effective strategy for classifying or triaging gang members in terms of their risk.
VALIDATING THE GANG RISK/THREAT CLASSIFICATION SYSTEM
This is where analysis comes into play, as we must now examine the classification system in relationship to other known and measureable risks and threats. Table 1 presents the results of a number of such tests, applying "factors of risk" or "factors of threat" to the classification system.
Having developed the "classification system" with three simple questions; scored them in the simple procedure discussed (if yes, add one point); and then applied the overall additive scores (varying from zero to three); now we must see if this type of classification system would be able to consistently and significantly differentiate other known risks associated with gang life today. Table 1 presents the results of this analysis. Table 1 shows some 44 different factors or variables that are in fact significant and in addition 42 of these are also consistently differentiated by the classification system.
The term "significant" here means whether the Chi-square test statistic yields a probability level equal or less than .05; and the term "consistent" here means that the gradations of "risk" or "threat" all rise or fall in a monotone fashion up or down the scale as the case may be. In otherwords, if we had a variable measuring STAFFHIT ("have you ever actually assaulted a correctional employee?", Yes, No), then we would expect this to be the lowest for Level Zero (maybe %10), somewhat higher for Level One (maybe 20%), still increasing and somewhat higher for Level Two (maybe 35%), and consistently highest for Level Three (maybe 50%). The variable should not fluctuate along the classification risk dimension, but should be consistent in its risk or threat effects along the classification system, then it is consistent as a variable or factor that differentiates human behavior.
The reader is urged to examine Table 1 briefly or to refer back to it in the narrative discussion that now follows.
RESULTS
Table 1 shows 44 variables that were significant in relationship to the classification system and 42 of these were consistently differentiated by the simple gang risk/threat classification system we developed. The purpose of this section is to discuss the meaning of these variables in terms of validating such a scale or classification system. Generally, however, we find much evidence for the utility of such an approach.
Also, applying this to type of gang nation (People, Folks, Crips, Bloods, Other) yields no significant difference, thus it is not a classification system that is biased towards any particular gang. The test of the classification system by gang type was not significant (p = .44).
TABLE 1
Percentage Distribution of Various Risk/Threat Factors
By the Gang Classification System
Gang Risk or Threat Classification
Level 0 Level 1 Level 2 Level 3 Percent agree "you cannot
expect justice through the
legal system". 51.5% 67.5% 69.7% 80.5%
Chi-square = 28.1, p <.001
Percent Anger Control
Problems: Get angry enough
to want to hit or hurt
someone at least once a day. 30.8% 44.8% 51.8% 50.6%
Chi-square = 35.0, p = .009
(BUT NOT FULLY CONSISTENT)
Percent Who have ever
collected protection money
on behalf of the gang. 9.8% 13.5% 28.0% 32.6%
Chi-square = 45.2, p < .001
Percent who have ever had to
pay a fine to the gang. 8.4% 14.9% 17.3% 27.0%
Chi-square = 19.7, p < .001
Percent who have served time
in a juvenile correctional
institution. 45.0% 57.3% 61.9% 66.8%
Chi-square = 17.0, p = .001
Percent who have a
permanent tattoo. 39.1% 56.0% 64.6% 71.9%
Chi-square = 40.2, p < .001
Percent who have ever
assaulted a school teacher. 26.9% 30.8% 36.2% 40.1%
Chi-square = 8.12, p = .04
Percent with five or more
close friends and associates
who are gang members. 51.4% 73.6% 84.3% 89.7%
Chi-square = 106.4, p < .001
Percent who report that
their gang operates
legitimate businesses. 28.0% 36.6% 56.5% 64.9%
Chi-square = 63.1, p <.001
Percent who report that
their gang requires members
to pay regular dues. 31.9% 34.2% 35.7% 48.3%
Chi-square = 11.0, p = .01
Percent who have ever
worked for a politician. 17.5% 20.1% 22.2% 36.3%
Chi-square = 19.5, p < .001
Percent who have ever
worked in voter registration. 5.1% 12.9% 13.9% 23.0%
Chi-square = 20.0, p <.001
Percent who have ever worked
as a "runner" in drug sales. 25.7% 35.1% 50.0% 63.4%
Chi-square = 57.3, p < .001
Percent who report that their
gang provides money to needy
members in or out of jail. 57.7% 71.9% 86.7% 91.9%
Chi-square = 68.0, p < .001
Percent who have ever committed
a crime for financial gain
with their gang. 40.8% 57.4% 68.0% 70.1%
Chi-square = 35.9, p <.001
Percent who report having ever
been fired from any job. 24.8% 25.5% 27.8% 38.7%
Chi-square = 9.80, p = .02
Percent who report they
own a car. 31.0% 40.0% 52.4% 60.2%
Chi-square = 34.1, p <.001
Percent who report they
own a motorscooter. 9.3% 11.7% 19.0% 20.8%
Chi-square = 13.4, p = .004
Percent who report they
own a SEGA genesis program. 59.2% 65.2% 72.4% 74.4%
Chi-square = 11.5, p = .009
Percent who report their
gang provides money to its
needy members. 49.5% 73.0% 84.9% 91.2%
Chi-square = 78.5, p <.001
Percent who report they
have ever been a pimp. 22.2% 28.1% 34.8% 38.9%
Chi-square = 12.1, p = .007
Percent who report their
gang has a private attorney
used for defending members. 26.2% 29.1% 38.5% 54.2%
Chi-square = 31.4, p < .001
Percent who report their
gang keeps an account that
pays for legal defense. 27.9% 28.6% 32.9% 45.4%
Chi-square = 13.4, p = .004
Percent who report that their
gang holds regular meetings. 70.4% 78.9% 86.0% 90.0%
Chi-square = 21.2, p <.001
Percent female 25.8% 17.3% 15.0% 6.2%
Chi-square = 19.3, p < .001
Percent who report their
mother has arrest record. 13.1% 14.8% 21.2% 23.0%
Chi-square = 8.74, p = .03
Percent who report they have
a cousin with arrest record. 44.2% 54.1% 62.8% 64.3%
Chi-squre = 16.6, p = .001
Percent who report that they
have family members who are
gang members. 65.8% 73.1% 79.0% 87.0%
Chi-square = 19.3, p < .001
Percent willing to die for
their gang friends. 16.3% 40.0% 56.3% 55.0%
Chi-square = 77.4, p <.001
BUT NOT CONSISTENT EFFECT
Percent believing their gang
friends will die for them. 27.8% 42.6% 60.2% 66.1%
Chi-square = 63.1 p < .001
Percent who report they spend
extra money on night clubs. 14.6% 20.0% 22.3% 33.5%
Chi-square = 15.3, p = .002
Percent who report that their
gang has a treasury. 42.2% 49.4% 63.9% 77.6%
Chi-square = 43.8, p <.001
Percent who report that they
have ever committed a crime on
the property or or inside an
apartment of a public housing
project. 24.7% 38.7% 42.4% 50.3%
Chi-square = 18.8, p < .001
Percent who report that they
have ever been violated
(received a beating) by
their own gang. 23.2% 33.2% 38.8% 41.1%
Chi-square = 11.5, p = .009
Percent who report that they
have ever used a police radio
scanner while doing a crime. 32.2% 34.1% 50.9% 62.7%
Percent who report that they
have ever engaged in "shakedowns"
(i.e., offering protection)
of small businesses. 18.5% 24.9% 29.2% 36.3%
Chi-square = 11.0, p = .01
Percent who believe that if
the police were continually
stopping and searching gang
members in a certain area that
the gang would move its
operations elsewhere. 56.1% 47.7% 42.2% 30.5%
Chi-square = 18.3, p < .001
Percent who report that they
have ever used a motel or
hotel for drug sales. 28.0% 41.0% 47.3% 50.3%
Chi-square = 16.2, p = .001
Percent who report that they
would still carry a gun in
their car if they knew the
car would be confiscated. 23.8% 33.0% 40.4% 54.1%
Chi-square = 27.9, p <.001
Percent who report that the
fear of going to prison affects
whether or not they would
commit a serious crime. 58.2% 44.9% 38.2% 36.0%
Chi-square = 16.7, p = .001
Percent who report that the
gang they belong to engages
in legal business activities
such as auto repair, etc. 30.9% 39.3% 59.6% 67.1%
Chi-square = 51.8, p < .001
Percent who report that they
have brothers or sisters from
their family that belong to
the same gang. 22.5% 37.4% 45.0% 45.3%
Chi-square = 19.6, p < .001
Percent who report that the
gang they belong to spends
its treasury money on guns. 60.4% 65.5% 74.2% 75.5%
Chi-square = 11.1, p = .01
Percent who report that the
gang they belong to spends
its treasury money
on funerals. 41.2% 49.8% 60.9% 66.6%
Chi-square = 21.4, p < .001
It is also important to note that the Level Zero group showed that 80.5 percent had tried to quit the gang by a separate question to that effect, further validating this as an "inactive" group or one with a lower risk factor.
SUMMARY OF THREAT ANALYLSIS RESULTS
Table 1 shows a number of attitudinal factors that can be taken as measures of risk which are significantly and consistently differentiated along the gang risk/threat classification system. There are also a cluster of of experiential or background factors that define how "hardened" or experienced a gang member is. There are also specific factors of violence propensity. There are also social factors about gang life itself, in relationship to friends and family. Finally, there are also factors of "gang sophistication" that also emerge here. All of these sets of factors are discussed below in reference to Table 1.
Attitudinal Factors
The variable for lack of belief in the legal system shows its lowest risk (51.5%) with the Level Zero group and the highest with the Level Three group (80.5%). The anger control variable is significant across the classification system, but it is not consistent in its incremental effects along the classification dimension. Other factors include:
*** Percent not deterred by aggressive police measures (i.e., percent who believe that if the police were
continually stopping and searching gang members in
a certain area that this would not make the gang
move its operations elsewhere).
*** Percent not deterred by the fear of going to
prison (i.e., fear of prison does not affect
whether they would commit a serious crime).
Experiential or Background Factors
Table 1 shows the gang classification system is able to consistently differentiate a host of experiential or background factors often taken into account in security classification system. These include the following:
*** Having ever collected protection money for the gang.
*** Having ever paid a fine to the gang.
*** Having ever served time in a juvenile institution.
*** Percent having a permanent tattoo.
*** Percent who have ever worked for a politician.
*** Percent who have ever worked in voter registration.
*** Percent who have ever been fired from a job.
*** Percent who report they have ever been a pimp.
*** Percent who are female (inverse relationship).
*** Percent who have committed a crime in public
housing projects.
Specific Factors of Violence Propensity
Table 1 shows the gang classification system is able to significantly and consistently differentiate several variables that may be considered factors of violence propensity. These include the following:
*** Percent who have ever assaulted a school teacher.
*** Percent who believe their gang friends are willing
to die for them.
*** Percent who report they have ever engaged in
"shakedowns" (i.e., offering protection)
of small businesses.
*** Percent who would still carry a gun in their car even if they knew the car would be confiscated.
*** Percent who report that their gang spends its
treasury money to buy guns.
*** Percent who report that their gang spends its
treasury money to pay for funerals.
Factors Related to Gang Friends and Family
Table 1 shows the gang classification system is able to significantly and consistently differentiate sesveral variables relating to the social life of the individual, that is factors relating to gang friends, and the family. These include the following:
*** Percent who have five or more close friends and
associates who are gang members.
*** Percent who report their mother has previously
been arrested.
*** Percent who report they have a cousin who has
previously been arrested.
*** Percent who report that they have family members
who are also gang members.
*** Percent who report having been "violated" by
their own gang.
*** Percent who report they have brothers or sisters
from their own family who belong to the same gang.
Factors of Criminal Gang Sophistication
Table 1 shows that the gang classification system is also sensitive enough to be able to significantly and consistently differentiate differences in the organizational sophistication and organizational capabilities of the gang as an organization. These factors include the following:
*** Percent who report their gang operates legitimate
businesses.
*** Percent who report their gang requires members to
pay regular dues.
*** Percent who have ever worked as a runner in a
retail drug sales operation.
*** Percent who report their gangs provides money
to needy members in or out of jail/prison.
*** Percent who have ever committed a crime for
financial gain with their gang.
*** Percent who report their gang provides money to
its needy members.
*** Percent who report their gang has a private attorney it uses for defending members.
*** Percent who report their gang keeps an account
that pays for legal defense.
*** Percent that report their gang holds regular meetings.
*** Percent who report that their gang maintains
a treasury.
*** Percent who report having ever used a police
radio scanner while doing a crime.
*** Percent who have used a motel or hotel for
illegal drug sales operations.
*** Percent who report their gang engages in
legal business activities.
Other Factors: Economic.
*** Percent who report they own a car.
*** Percent who report they own a motorscooter.
*** Percent who report they own a SEGA genesis game.
*** Percent who report they spend their individual
extra income on night clubs.
DISCUSSION OF FINDINGS
The summary of results from testing the classification system in relationship to various "risk" or "threat" factors in this large national sample certainly deserves more discussion that we have provided here. However, our overall report is also directed at many other issues, this being only one of the concerns. We will, though, now come to some summation on the matter of security/threat group classification, and whether this particular simulation would have been a valuable way of "handling" or managing a gang population.
What this technique does is similar to the prediction methods of Burgess
and others in the Chicago school of sociology, where by examining everything from
marriage to recidivism it is possible to establish "base rates" for such risks,
and then examine in detail the factors that are highly associated with these
outcomes being predicted. In the format used here it is not pure prediction
research, it is more in line with an actuarial approach such as an insurance
company might use. The tests are therefore bivariate, not multivariate in
nature
.
The question, then, does such a simple classification system as we formulated for test purposes here, really have any valid application to gang management? We say yes. The issue is preventing violence, limiting risk, reducing risk, reducing morbidity, injuries, and mortality; and Table 1 shows how a wide number of other risks (42 risks in all) were significantly and consistently differentiated by the simple gang risk/threat classification used for test purposes. We want to reiterate we are not advocating such a simplistic classification system, but rather we used one that was very "elementary" for purposes of the scenario test made here. The fact is even an elementary system can work to achieve valid uses.
CONCLUSION
Since this early study, the National Gang Crime Research Center has worked with correctional staff in about 20 states to further develop a classification and risk/threat analysis system. It is possible that such a system can significantly reduce disturbances and assaultive behavior and other "risks" among jail inmates. This is, after all, a reasonable assumption given the fact that gang members are known to be a primary source of such behavior problems while in custody.
For further information about gang classification and risk/threat analysis, contact the NGCRC.
ENDNOTES: