Exercise: Statins, pneumonia and confounding, 4

Question 4: After using splitting into fifths to check the propensity scores, the authors say that they entered the score into their regression models as a continuous variable. What are the advantages and disadvantages of using score as a continuous variable rather than as a categorical variable with five categories?

Suggested answer

The advantages of using propensity score as a continuous variable are:

  1. we use all the information available in the continuous propensity score,
  2. we need estimate only one coefficient if we assume the relationship is linear, although we would need more if it were not.

The disadvantages of using propensity score as a continuous variable are:

  1. we need to assume that the relationship between outcome and score has a particular form, such as linear on the logistic scale,
  2. the coefficents are difficult to interpret.

The advantages of using score as a categorical variable with five categories are:

  1. we do not need to assume the relationship is linear,
  2. we estimate coefficients with a fairly simple interpretation, as log odds ratios comparing each fifth to a reference category.

The disadvantages of using score as a categorical variable with five categories are:

  1. we do not use all the information in the propensity score, though with five categories we do not lose a great deal,
  2. we must estimate four coefficients rather than one.


Back to Exercise: Statins, pneumonia and confounding.

To Applied Biostatistics index.

To Martin Bland's M.Sc. index.

To Martin Bland's home page.

This page maintained by Martin Bland.
Last updated: 12 December, 2006.

Back to top.