Monday, January 22, 2007

Trusting the Random

A common source of confusion for new students studying research methodology is lack of faith in the random. If you're trying to determine whether or not, for example, classical music improves memory retention, the students will want to break down the control and experimental groups based on everyone's -existing- memory skills to make sure it all breaks down evenly.

You can do this, but it has its drawbacks: you run serious risk of thinking your research is more significant than it is by breaking everyone down along arbitrary lines in a fashion which doesn't have a factual basis. You also take on unconscious experimenter bias when you make intentional efforts to break down the groups in an equitable fashion.

If, however, you break them down along totally random lines, you place yourself in a better situation by removing yourself from the equation of choosing who goes into which group. You also run a risk of these random factors influencing your experimental results, but this gets taken care of through replication and reassessment. The random elements involved become statistically insignificant should you have multiple experiments that suggest the same results.

No comments: