Which statement about Categorized Variables is correct?

Which statement about Categorized Variables is correct?

• A. They are never desirable.
• B. Dichotomization or other categorization of continuous or ordinal variables may sometimes be desirable.
• C. They must always be done after data lock.
• D. They only apply to qualitative endpoints.

Answer: (B)

Categorized variables are created when you convert a measurement that is continuous or ordinal into discrete groups. This can be desirable because it often makes results easier to interpret and can reflect clinically meaningful thresholds, such as turning age into age bands or defining blood pressure categories. It also helps align analyses with predefined endpoints or regulatory expectations. However, this approach has trade-offs. Grouping continuous data can throw away information and reduce statistical power, and the choice of cut-points can introduce bias if not pre-specified in a statistical analysis plan. When used thoughtfully, with pre-specified categories and justification, categorization can be a useful tool in data analysis. It is not restricted to qualitative endpoints and does not have to occur only after data lock.

Categorized variables are created when you convert a measurement that is continuous or ordinal into discrete groups. This can be desirable because it often makes results easier to interpret and can reflect clinically meaningful thresholds, such as turning age into age bands or defining blood pressure categories. It also helps align analyses with predefined endpoints or regulatory expectations.

However, this approach has trade-offs. Grouping continuous data can throw away information and reduce statistical power, and the choice of cut-points can introduce bias if not pre-specified in a statistical analysis plan. When used thoughtfully, with pre-specified categories and justification, categorization can be a useful tool in data analysis. It is not restricted to qualitative endpoints and does not have to occur only after data lock.