I'm interested in compiling a compendium of truly dysfunctional experimental
designs with the goal of making the point of how important it is to involve
a statistician from the very initial phases of research. Examples would
maximize the (time wasted):(degrees of freedom) ratio. A concrete example
would be my 1st undergraduate research project in which I spent an entire
summer doing behavioral observations on paperwasps, only to find at the end
that I had but 7 dF for my contrasts because I had spent far too much time
on each nest, while observing far too few _different_ nests -- which,
combined with the variability of data on behavioral observations, no doubt
left the power of my comparisons groping asymptotically toward nil. The
only positive result of the study was that the following semester I __very
eagerly__ started taking statistics courses!!
I'll compile the list of horrors and post it here in a couple of weeks so
all can use the completed document.
Thx in advance -- Eric
=================================================================
Instructions for joining and leaving this list and remarks about
the problem of INAPPROPRIATE MESSAGES are available at
http://jse.stat.ncsu.edu/
=================================================================
