In statistics, parsimony is best described as?

• A. A method to maximize the amount of data collected.
• B. The principle of keeping explanations simple; favoring the simplest model that fits the data.
• C. A criterion for selecting variables based on p-values.
• D. An approach to increasing the complexity of models.

Answer: (B)

Parsimony in statistics means preferring the simplest explanation or model that still fits the data adequately. It aligns with the idea of Occam’s razor: a model with fewer parameters that explains the data well is generally better because it’s less prone to overfitting and easier to interpret. In practice, we balance how well a model fits with how complex it is, sometimes using criteria like AIC or BIC that penalize extra predictors. The option that describes keeping explanations simple and choosing the simplest model that fits the data captures this idea. The other statements describe maximizing data collection, selecting variables by p-values, or increasing model complexity, which are not about favoring simplicity when a simpler model suffices.