Is Randomization Foolproof?

Despite all efforts at a priori criteria for inclusion and exclusion and case definition and random assignment to groups, sometimes the "luck of the draw" can't be trusted because there are some especially critical concerns that must be controlled by the investigator.

For example, in order to insure that randomization of patients to multiple treating dentists did not somehow assign cases to dentists in unequal numbers, a stratified randomization schedule was used, which built into the randomization procedure the assurance that all the TMD care providers saw about the same number of cases for each intervention.

Question 5.2

What do you think is the best choice if randomization fails and the groups turn out not to be comparable on a critical baseline parameter?

Drop the variable from the analysis

Use appropriate post-hoc statistical methods to control for non-comparability at baseline

Combine groups on the relevant variable(s)

Re-calculate sample size estimates and increase sample size

Page 23 of 81
Previous Page	Next Page

Next Section