The Interval Problem, and Why No Controlled Test Exists

The methodological objection is not that the investigators were careless, and Stevenson's critics generally concede that his documentation was unusually careful for the subject. It is structural. By the time an investigator hears about a case, the child has been speaking for months, the family has interpreted the statements within a framework that expects them, a candidate previous personality has often already been identified by relatives or neighbours, and the two families may have met and talked. Everything the investigator records is therefore downstream of a process the investigator did not observe.

The interviews are frequently conducted through interpreters, which adds a further layer between the child's words and the written record. And there is no controlled version of the study available: one cannot randomise children to cultures, or blind a family to its own conversations. Stevenson knew all of this and argued that the strongest cases involved verified details that could not plausibly have travelled between the families, which is the right defence to attempt and one that turns on the particulars of each file. The collection continues at the University of Virginia under Jim Tucker, and the standoff between the case-by-case defence and the structural objection is exactly where it was thirty years ago.