2011 Van der Graaf: No mutation = probable misdiagnosis.

Again, FCH is implicated in the failure to find mutations within an FH scoring system’s results. The problem is that these scoring systems are really just epistemological cousins of what in legal arenas serve as weaker forms of circumstantial evidence. Just because two suspects share the same characteristic does not mean that both or either are guilty.  This problem can also be understood through Kahneman and Tversky’s analysis of human bias and base rate (click here and here). If there is only one bank robber on video and he wore red shoes, then it would not be wise for me to declare that there were two robbers just because I found two people with red shoes. Finding shared characteristics with a target does not alter the actual population of that target. Having two suspects and only one bank robber, a process of exclusion is imperative, if I value few-but-accurate over many-but-inaccurate. Lowering the standard of accuracy increases the quantity of suspects to two, even though the video establishes the odds at 50%. If both FH and FCH have passing scores, have I really doubled FH prevalence?

2011 Van der Graaf: No mutation = probable misdiagnosis.