Observational data and the modeling it generates are cold and static. And no statistical technique, regardless of its sophistication, can overcome the inherent limitations of observational data. In contrast, experimental data and the modeling it generates are alive and dynamic.
We will never know what messages, digital tactics or other campaign tools work or are a waste without experiments. As Alan Gerber, Donald Green and Edward Kaplan—two of whom are political scientists from Yale who brought experiments out of academia and into Democratic politics—conclude, "unless researchers have prior information about the biases associated with observational research, observational findings are accorded zero weight [in a test of a causal proposition] regardless of sample size, and researchers learn about causality exclusively through experimental results."
Big, integrated, and clean observational data are a necessity. But it isn't sufficient. Mathematician and physicist Henri Poincaré claimed, "experiment alone can teach us something new; it alone can give us certainty." I'd only caution that certainty is not something we can expect of this world. But experiments bring us as close to glimpsing it as we can hope.
- Read Susan Milligan: Obamacare Is Lowering Some Health Care Costs
- Read Ron Bonjean: Obama's Thomas Perez Trap Could Slow GOP Progress With Hispanics
- Check out U.S. News Weekly, now available on iPad.