|Beautiful patterns tell us how...not why|
"When imagination sleeps, words are emptied of their meaning." Albert Camus - 1960
Last blog discussed a study by US Forest Service researchers showing how tree canopy in Baltimore cut urban crime by 12%. I just read a 2010 study commissioned by the Center for Disease Control showing how business improvement districts cut violent crime by the same amount in Los Angeles.
So we cut crime 24% by planting tree canopies in a business improvement district (BID)?
Sounds silly. That's because both studies are correlational - studies that show a pattern between two things, not a cause. We don't know why canopies or BIDs work, only that they seem to have impact.
Dozens of correlational studies appear each year. Like research showing underarm deodorant causes cancer. Or toothpaste. Or cell phones. Or smoking which, it turns out, is true. What to do?
There is a mantra in science: correlation-is-not-causation. Because there is a relationship doesn't mean you can infer cause. Tree canopy and BIDs may coincide with crime declines. That doesn't mean they cause them.
Unfortunately neither can we dismiss correlation studies. Doing so may result in dismissing a good cause assertion. The International Agency for Research on Cancer lists criteria for testing correlational studies - a plausible mechanism between cause and effect, a singular relationship between before/after effects, a reasonable time between the cause/effect, and so on.
Satisfying these criteria helps mitigate the correlation-is-not-causation dilemma.
CORRELATION IS NOT CAUSE
I think what correlational studies need is simple - a plausible theory explaining why. That way we'll know if a study shows a correlational relationship we are seeing a shuffle not a full step.
Sometimes I wonder if theory-building is a dead language. It shouldn't be! That's where research creativity and imagination really show up (not in clever data manipulation or new statistical methods).
There are some great publications that provide imaginative theory, for example Gilligen's Preventing Violence and Kennedy's Don't Shoot".
It would be wonderful to produce a 24% cut in crime by planting tall trees with great canopy in a business improvement district. It would be a shame if they don't and we have no idea why.