We agree with Professor
Paykel that the recall and reporting of life events is more complex than is generally recognized. Use of an interview has the potential of providing better-quality data than an unsupervised paper-and-pencil method in that the interviewer can better focus the subject on the defined time period and apply a consistent set of definitions of events and their thresholds. There is a danger, however, that this reduction in error variance might be partially replaced by systematic bias unless the interviewers are kept "blind" to the clinical status of participants in a retrospective study and unless all probing and defining is done in the same way for all subjects. This can only be achieved if interviewers adhere to a standard protocol for conducting their inquiry.
The cited 9% difference in the frequency of events recalled (at a single interview) for two consecutive, sixmonth periods is not notably different