The crisis apparition case rests entirely on personal testimony gathered after the outcome, the death or crisis, was already known, which is precisely the evidentiary structure that makes such claims resistant to rigorous evaluation: a respondent's memory of exactly when their vision occurred, and how precisely it matched the real event, is reconstructed retrospectively and is vulnerable to the same distortions that affect all retrospective recall, including a tendency to remember hits more vividly than the far larger number of ordinary anxious thoughts about absent loved ones that turned out to mean nothing.
The SPR investigators themselves were aware of this weakness and tried to address it through methodological care, corroborating witnesses, written statements requested as soon as possible after a case came to light, comparison against actuarial death rates, but even sympathetic later historians of the Society acknowledge the Census of Hallucinations could not rule out selective reporting, in which people who experienced a coincidental vision were simply more likely to remember and report it than the much larger population whose ordinary worried thoughts about someone came to nothing. No controlled, prospective test of crisis apparitions, one that could establish in advance who was likely to experience a vision and verify it against an independently timed real event, has ever been conducted or is straightforwardly possible given the spontaneous, unpredictable nature of the claimed phenomenon, which is why the case for crisis apparitions remains exactly where Gurney, Myers, and Podmore left it in 1886: a large, carefully collected body of testimony, not a controlled experimental result.