Research, evaluation, analysis

A Good Shave

Ockham's (or Occam's) razor is a principle of analysis expounded by William of Ockham in the fourteenth century, and still guiding serious analysis today. Ockham proposed that explanatory entities not be multiplied beyond necessity; in modern terms, an explanation should use no more variables than are necessary. Such an explanation is said to be parsimonious.

A prominent contemporary example of the application of Ockham's razor is the concept of socioeconomic status, or SES. For example, we could predict children's success in school from knowledge of several social and economic characteristics of their families: income, parents' schooling, parents' occupation, and so on. However, many of these variables are correlated: the more schooling you have, the higher your income tends to be, and so on. You can simplify the explanation by combining all these variables into a single measure of socioeconomic status.

Not only does this make the explanation much easier to understand, but it also avoids technical problems of which William of Ockham could not be aware. These days we use multiple linear regression analysis to perform analyses like the analysis of socioeconomic variables and children's academic achievement. Scaling the socioeconomic variables keeps the multiple linear regression equation, which you use to predict achievement, stable. If you do not scale the variables, the equation will, for technical reasons, be less effective at predicting achievement.

Scaling variables is one of the most effective ways to clarify analysis. If you try to use ten independent variables to predict ten dependent variables, you have 10 X 10 = 100 relationships to explain. If you can combine each set of ten variables into two scales, you have only four relationships to explain.

A Good Shave © 2001, John FitzGerald
Home page | Decisionmakers' index | E-mail