Featured Post

Homeland Security: The Sworn Duty of Public Officials

Homeland Security: The Sworn Duty of Public Officials     The United States has a unique position amongst the countries of the world;...

Wednesday, January 7, 2015

DUF/ADAM Substance Use and Crime Correlation

In 2004, Martin, Maxwell, White, & Zhang investigated the correlation between substance abuse and crime, specifically, cocaine and alcohol abuse.

Martin et al studied the correlation by taking data from the Drug Use Forecasting (DUF)/Arrestee Drug and Alcohol Monitoring (ADAM) program in the years 1989 through 1998 and data from the Uniform UCR Offenses Known and Cleared by Arrest databases. The researchers matched this data by linking the UCR data to the DUF/ADAM catchment areas. Due to issues matching some areas, the researchers were left with 22 sites of study. Outliers were accounted for by a process of imputing other sources of data ( such as the FBI's Supplementary Homicide Reports ). Differences in socio-economic conditions (SES) between sites were controlled by creating a construct, the quantification of the “Urban Underclass”. This was intended to “to partially control for between-site differences that may account for observed violence and property crime rates thereby providing a more precise estimate of change over time within each site” (Martin, Maxwell, White, & Zhang, 2004, para. 16). One issue arising from this is that “The DUF/ADAM data are not representative of arrestees as a whole because the data come only from large metropolitan areas and the sample is weighted toward selecting the more serious offenders.” (Martin et al, para.7) In addition, the study excluded data associated with heroin use due to a previously known correlation between heroin and crime, in particular, property crime. The data was then analyzed by site and within each site by year ( referred to as an “individual”. Martin et al “examined the bivariate correlations among DUF/ADAM positive rates for alcohol (self-report), cocaine (urinanalysis), and UCR property and violent crime rates.” (Martin et al, para. 17) They then used the SPSS (a predictive analytics software package) Multilevel Mixed Model to account for violent crime and property crime respectively. This approach allowed them to combine data as they could separate data and run regression analysis by site and year; “regression analysis helps one understand how the typical value of the dependent variable (or 'criterion variable') changes when any one of the independent variables is varied, while the other independent variables are held fixed.” (Wikipedia, 2014, para. 1). To summarize, this allowed the researchers to estimate the correlation between “annual crime rate and the aggregated cocaine and alcohol annual use rates” (Martin et al, para. 17)

Martin, Maxwell, White, & Zhang found that in 20 of 22 of the sites there was a correlation between alcohol use and violent crime, that there was a weaker but still positive correlation between cocaine use and violent crime, that there was a weak but still positive relationship between alcohol use and property crime, and finally there was a positive correlation between cocaine use and property crime. In each case (substance use to type of crime) there was a positive correlation between substance use and crime, however weak.

The research of Martin et al does not surprise me. I don't feel that substance use in an indicator of criminal proclivity in of itself, but that the inherent distortion of perception and the emotional changes, especially in inhibition, brought on by use make it more likely that a person commits a crime under the influence. Increased use leads to a greater probability of crime. Furthermore, in my opinion, crimes are more likely to be committed by people with low impulse control; the abuse of alcohol and cocaine is also more likely in a person with low self-control, exacerbating that problem. I certainly disagree with the idea of alcohol controls on the general public as a remedy for these individuals. As an aside, what is it with academics always wanting to raises taxes?

I think that to replicate this study in my city that I would scrap the use of the DUF/ADAM data completely. If substance use has a positive correlation with crime, then that condition should apply to society as a whole, not simply the “Urban Underclass”, and should be tested for. I could research the per capita use of alcohol by pulling tax receipts from the area's liquor sales; I was unable to find a good method for determining a measure of cocaine use for the area, but I will follow up if I find a better answer then measuring arrests for possession (again, considering that the “underclass” accounts for more arrests due to usage patterns). Then once I had that data I would proceed to compare those numbers versus the UCR data over time.

Martin, S., Maxwell, C., White, H., & Zhang, Y. (2004). Trends in alcohol use, cocaine use, and crime: 1989–1998. Journal of Drug Issues, 34(2), 333–359. Retrieved July 16, 2014 from http://search.proquest.com.southuniversity.libproxy.edmc.edu/docview/208856473

Regression analysis. (2014) Wikipedia. Retrieved July 16, 2014 from http://en.wikipedia.org/wiki/Regression_analysis


You did a good job clarifying that alcohol is linked to more crime then drugs are;  I'm sure that for people that fully understand statistics the researchers intent was clear, but it's nice to see that pointed out in plain language.

victims perceived the offenders to be under the influence of alcohol compared to to those under the effects of drugs at almost a 3 to 1 ratio, with 13.8% of respondents reporting their perception that he offender was under the influence of alcohol, and 5.1% reporting that the offender was under the effect of drugs.

Greenfield, L.1998 Alcohol and Crime Bureau of Justice Statistics Retrieved January 17, 2014 from HTTP://www.bjs.gov/content/pub/pdf/ac.pdf
No problem.

My opinion is that most crime is committed by people with low self-control;

"I see it, I want it, I take it"  or "You made me mad, so I hit you".

Low self-control may be a result of poor socialization, but I do see middle-class people do dumb spur-of-the-moment things all the time (they have a little more leeway financially).  If I knew rich people, I would expect to see a certain percentage of them behave completely irresponsible.

Another way of saying low self-control would be the lack of ability to calculate the pain/pleasure matrix of consequences.

This ties into substance abuse/use because it is a bad decision to drink or snort so much you lose control; maybe somebody has low self control, but enough to not commit crime when sober, unfortunately, the self-control doesnt prevent him from getting wasted, and escalates from there.

Let me know if I'm still not being clear!

I think a person with low self-control is more likely to commit a crime (whether he is sober or not) then the "normal" person.

I think that a person with low self-control is even more likely to commit a crime when under the influence of a substance then he is when sober.

Finally, I think that a person with low self-control is more likely to be under the influence of a substance then a "normal" person; leading back to that second condition (the last sentence)

responding to a question regarding identification of motive based on lack of self-control
Not in the UCR which,  according to the NIJ, "Gives a tally of the incidents. Does not contain information on each reported incident."

The NIBRS may make it easier to determine motive, but doesn't specifically collect the motive as data; what it collects that may have an effect on motive are"
Characteristics of victim(s) and offender(s).
Relationship between the victim and offender."

Even that would involve guesswork;  I think you would have to collect data on motive specifically, and that would be difficult in cases where the arrestee is proclaiming innocence.

"Sources of Crime Data: Uniform Crime Reports and the National Incident-Based Reporting System"
I don't want to get into a trap about one single cause as a source for crime;  I think there are many causes, but as far as my opinion goes I say that most crimes are committed by people with low self-control.

To test for this you would have to identify other indicators of low self-control;  speeding tickets, at-fault traffic accidents, failing grades in school, children out of wedlock ( or having kids without the personal ability to feed them), substance abuse, multiple job firings, etc.

You would have to be clear as to HOW these indicators would mark low self-control; for example, some businesses hesitate to hire people who move jobs every three years or so, but that isn't something I would consider as an indicator.

You would then set up some sort of scoring measure for the value of "self-control"; or perhaps you could measure for each indicator of low control and aggregate them.

You would then take random samples of the population by using a survey that measured the self-control indicators;  you could either use a self-report on the same survey to measure crimes committed, or you could gain permission to check criminal records.  You would then check the proportion of crimes commited by those with low self-control versus those that did not.

I don't think you would want to sample arrestees because you are drawing from a biased sample.

I did want to reiterate that is what I think causes most crime, but not all crime!

To address both your points I think we need to look at the started purpose of the study again:
“This study explores the relationship between alcohol and cocaine use and crime from 1989-1998, based on findings from the Drug Use Forecasting/ Arrestee Drug and Alcohol Monitoring Program and the Uniform Crime Reports Program “ (Martin et al, para.1)

The first clause is simple enough, however it does lead to some ambiguity when modified by the second clause.  Is the study only supposed to apply do the population of substance use arrestees, or to the general population at large?

Martin et al clarify their intent with further remarks linking alcohol use and crime.  “This lack of policy attention to the alcohol-crime link beyond DWI is perplexing since research findings that stretch back nearly 50 years to Wolfgang's classic study of homicide have consistently documented an association between alcohol consumption and violent crime “ (Martin et al, para.4)  They further state that studies have "indicated that alcohol outlets raised the rate of violent crime within the immediate neighborhood context “ (Martin et al, para.4)

Their policy solutions also suggest a causation between alcohol use and crime; "The findings suggest that to reduce violent crime rates, policy makers need to focus on addressing the contribution of alcohol” (Martin et al, para.1)  Their solutions of taxes and other restrictions upon the freedom of the general populace are based upon a relation between alcohol use and crime in general, not upon a correlation between alcohol use and criminal offenders (DUF/ADAM subjects).  An example of a solution based on the second relationship would be additional incapacitation for offenders under the influence.

As to the use of an illicit substance being a crime in of itself, it is hard to address without coming close to circular logic, but I'll try to return to my interpretation of the purpose of the study:  Why is the use of some drugs illegal if that use does not cause crime to begin with?   The researchers controlled against heroin users because of an known correlation.  Martin et al did note the legal distinction, and their focus seemed to be on the alcohol/crime link.  They also contrasted cocaine use against alcohol use; “Cocaine use, in contrast, is not closely associated with either property or violent crime rates in the multivariate analyses “(Martin et al, para.1)  And their proposed solutions were also focused on alcohol.

Both points return to the purpose of the study and it's practical significance.

Good point about the reliability of the UCR data.  I wonder what effect it would have on the correlation factors if the hierarchy rule were removed, and had to account for additional crimes that the rule basically "hides" from the Martin group's research


No comments:

Post a Comment