>Often the decision has nothing to do with, "What language produces systems
>that cost less or are more reliable?" (IMO Ada).  The language decision is
>Unfortunately, I have seen managers who ignore statistically valid data;
  All true.  But in addition to making managers pay attention, and
creating a large supply of expert Ada programmers, we should be
urging the advance of knowledge.

>would like to point out that data valid at the 50% confidence level
>is no better than a coin toss.
   I have spoken to the local zoo, and they agreed to toss a coin, and
if it comes up heads they will place a lion outside your door.
So the hypothesis "there's a lion outside your door" is true with
50% confidence.  Does that knowledge affect your actions?
  0% confidence means "I have no useful data".  50% means "I may
be wrong, but my data indicates ...".