What does it mean "with control for differences in factors" in the sentence :

"Several studies used cross-sectional analyses to detect a positive relationship between the prevalence of gun ownership at the neighbourhood, county, regional or state level and homicide rates, with control for differences in factors associated with homicide (e.g., urbanization, race/ethnicity, unemployment, poverty, crime, and alcohol use)."

What did they do to those factors when they "control" them ?

The full report can be found here, if that helps for context: https://ajph.aphapublications.org/doi/full/10.2105/AJPH.2013.301409

Does that mean that they took in consideration all those factors (urbanization, race/ethnicity, poverty, etc.) in the method to minimize the impact those factors have in the results of the factor that matters the most to homicide rates in this study, in this case gun ownership? If so, how can it be done? An illustrative example is enough.

  • $\begingroup$ This might be more appropriate on Cross Validated SE (I can migrate it there). At a glance, it does not seem directly related to psychology or neuroscience. Good question though! +1 $\endgroup$
    – Steven Jeuris
    Apr 18, 2018 at 9:27
  • $\begingroup$ Please, move my question to where is more appropriate. I was in doubt if this is the correct board but couldn't find any other that related closer to this subject than this one. $\endgroup$
    – Bert
    Apr 18, 2018 at 18:50

1 Answer 1


There are various ways of controlling for other other factors.

Probably the most common is to include these control variable as additional variables in whatever statistical model you are running.

For example, you could run a multiple regression predicting homicide rate from gun ownership on a set of neighbourhoods.

You could then run another regression model that included all the control variables.

In both models you would get a coefficient for the effect of gun ownership on homicide rate. However, in the one with control variables, it would be statistically adjusted for the effect of these other variables.

There are lots of debates about how best to control for other factors and which factors to include. But that's the general idea.

  • $\begingroup$ So in the model that includes all the control variables you would get a less accurate coefficient of statistical correlation between gun ownership and homicide rate? or is it the other way around ? I'm asking this because the study wants to highlight the importance of gun ownership to the homicide rate statistic, but if the model used takes many other additional variables the final coefficient might be less true to the gun ownership reality, and the final result biased. $\endgroup$
    – Bert
    Apr 20, 2018 at 13:49

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

Not the answer you're looking for? Browse other questions tagged or ask your own question.