You Don’t Need to be a Statistician to Understand Research Methods
I will begin this blog with two confessions:
I have mild numbers dyslexia (and therefore someone always has to check even the simplest of my sums)
I became a Criminologist because I couldn’t pass the LSAT if my life depended upon it. I could not figure out – or give a rat’s ass about – problems involving a train leaving a station.
“WAIT!”, you say, “but now you are writing a blog on research methods. What gives?”
It’s simple: you don’t need to be a numbers guru or have advanced methodological training to intuitively understand the basics of research methods. In fact, much of advanced methodology will also be well within your grasp if you understand a little something that underlies all research: logic.
Yup, that’s what research (and stats) is really about: the application of logic and logical models to thinking through how you will better understand a particular phenomenon – from supernovas to the effectiveness of youth detention programs.
Let me give you an example. My friend Jerry Ratcliffe has done a lot of research on whether foot patrols are effective at reducing crime and disorder. The technical term for what he does is ‘randomized controlled trials’ and, when you speak to him, he’ll tell you that he had a lot of concerns about ‘treatment fidelity’ and ensuring the ‘reliability’ of his work because of potential ‘reliability issues’ with his ‘independent variable (IV)’.
Now, if Jerry was to say all of this to me over coffee using the language above, this is what I would hear:
Yes, I know what all of those terms mean independently, but when you string them together and throw in some ‘chi squareds’ and a few ‘confidence intervals’, I have now slipped into a coma.
Confession 3: Although I got an A+ in grad stats*, I only ever took one course, have zero training in stats, and don’t see methodological or other problems through that lens.
However, I don’t need to have that training to understand that Jerry’s study could have had a potentially BIG, BIG problem, and it’s the same one Jerry identifies when he talks about ‘treatment fidelity’ and ‘reliability’ and his ‘IV’. How is this possible? Because I understand logic (and policing)!!
Jerry’s foot patrol study requires the officers (the independent variable) to actually do what Jerry wants them to do in their assigned beats (treatment fidelity). If they don’t, we cannot say that any crime drops were actually caused by the foot patrol (reliability). Voila!
And guess what? I don’t know too many people in policing who wouldn’t have taken one look at that scenario and come to the same conclusion. "Oh, oh ... you've got to get them to do what you want on patrol. Hmm... (big pause for dramatic effect). That could be a problem".
*Yes, I HAD to throw that in. A+. Fortunately, we had computers back then, otherwise I'd still be there trying to figure out the math.
