>What, in your opinion, is the functional source of the phenomenon of program survival bias? Is it as simple as the lack of double blind controls?<
Program-survival bias comes from the fact that phase-2 trials which perform badly don’t progress to phase-3.
There are other reasons why phase-2 results may not be reproduced in phase-3 that pertain to such matters as blinding and patient selection; however, these reasons are distinct from program-survival bias and are easier to analyze on a case by case basis. Regards, Dew