Balanced Incomplete Blocks to Control Individual Differences

Balanced incomplete block designs are used when there are reasons that preclude the use of repeated measurements designs, particularly when there are at least two factors, each with several levels. However, these designs are usually not used when interaction effects are of interest, because they involve the confounding of the interactions with subject or group differences. This paper describes special computational procedures that can be employed to obtain partial, but unconfounded, information about interaction effects