Mszlazak wrote:
>To empirically test pre-release MOZ is a good idea. "Bias of ascertainment" is
>a DUMB ASS excuse. You can always balance out test groups in different ways to
>control enough for bias. Don't go to a stats book to learn experimental design,
>talk to a good experimental scientist that does some related emperical
>work...probably in a psychology department.
>
hehehe, does a PhD in Genetic Analysis qualify me as an experimental
scientist? Last time I checked it did. ;-), but that is not the point.
Bias of ascertainment is not the same thing as just 'bias". It is quite
a complex concept to deal with.. As a general rule it should be avoided
... lets keep it parsimonious and apply Occam's Razor to the criteria
selection.
BTW, I visit these newsgroups for a particular reason, I am not a
programmer at all but I am also teaching to my graduate students in
Genetics a course on Process Analysis, scientific facilitation and group
processes and by reading these groups I can see the practical
applications for group theory as applied to scientific teams.
