Confound Variable

#1
From my understanding, a confounding variable is a variable that can influence the IV and DV as well as the outcome. However, I am still trying to figure out the many confounding variables in the following situation. I am hoping someone can help me identify them and explain why so I can understand confounding variables.

"I am examining the risk of recidivism of youth offenders who attend a treatment program compared to the youths who did not attend the treatment program. The independent variable is the treatment program and the dependent variable is the risk of recidivism. What are some confounding variables that influence the results? "

I first thought that race could be a confounding variable but my professor said I am wrong...

Thank you for any help I get,
JJ
 

Karabiner

TS Contributor
#2
Race would be a confounder if it affected the probability of attending a program, as well as recidivism risk.

The dependent variable is a bit surprising. It is not recidivism yes/no, but recidivism risk? Maybe it is
an index of several risk factors, or some rating scale used by the program directors? In the latter
case, race could well bias raters' perceptions and conclusions.

Or maybe there's evidence that race does not predict neither program attendence nor recidivism/
recidivism risk, in those who already are offenders?

A confounder would be something like the determination of a subject not to relapse. Those subjects
probably have a lower recidivism risk, and they choose programs supporting their determination.
So even if such a program did not have an effect, its trainees would have lower relapse rates.

With kind regards

Karabiner
 

hlsmith

Less is more. Stay pure. Stay poor.
#3
So if the treatment was not randomized, the background characteristics for those that opted to participate or not may differ. The issue is that some of these characteristics can impact the outcome as well. Social network could influence someone's ability to get a ride to the intervention and impact recidivism. Also, recidivism seems like an incomplete outcome, since I could engage in a crime and not get caught. So the outcome is a subgroup of committing a crime and getting caught, which may not fully evaluate the impact of the intervention.