Monday, August 15, 2011

Why collect data?

Can't we tell what works just by paying attention? Do we really have to go through the bother of collecting data or even designing experiments?

The problem with conducting experiments where people are involved is that people are hopeful. They want results to come out well. And, because of this hope, they are often biased in their observations, interpretations and actions. It is not that they are malicious or deceitful or intend to corrupt the results of experiments. To the contrary, it is because they are helpful and hopeful that they often unintentionally corrupt the results of experiments.

A question asked is better than a question not asked.

An experiment run is better than an experiment not run.

An outcome measured is better than an outcome not measured.

An experiment designed to eliminate bias is better than an experiment performed casually.

There are many ways to make experiments better. The greatest error, however, is to not experiment at all, to never risk performing better.

Posted at 12:00 AM (permalink) 2 Comments View/Leave Comment