Research into the black box of rehabilitation: the risks of a Type III error.

Type I and Type II errors in the interpretation of data from clinical trials concern statistical matters, and the probability of drawing erroneous conclusions from inadequate data. However in rehabilitation research a third possible error may arise. Successful rehabilitation depends upon the co-ordinated work of an expert multidisciplinary team, and can be considered as a network involving a whole system. Demonstrating that one part of that system looked at in isolation does not have the expected effect does not prove that the specific part is not necessary to the success of the whole system. The isolated intervention may still have an important effect when interacting with other variables or interventions. Failure to consider the interactive effects of an intervention might constitute a Type III interpretation error.