<html><head></head><body>Anthony Simmons wrote:<br>
<blockquote type="cite" cite="mid:3.0.5.16.20010522151449.0877df0a@krl.org"><pre wrap="">I have a couple of questions:<br><br>How will you deal with a situation in which there are, say,<br>ten variables, all correlated, but none colinear?</pre></blockquote>
<blockquote type="cite" cite="mid:3.0.5.16.20010522151449.0877df0a@krl.org"><pre wrap="">How will you deal with the fact that different voters will be<br>positioning candidates in different spaces, with different<br>scales and different dimensions?</pre></blockquote>
<blockquote type="cite" cite="mid:3.0.5.16.20010522151449.0877df0a@krl.org"><pre wrap="">I imagine a good way to deal with these issues would be to<br>ignore them.</pre></blockquote>
Exactly my plan.<br>
<blockquote type="cite" cite="mid:3.0.5.16.20010522151449.0877df0a@krl.org"><pre wrap=""> After all, if the model were to take these into<br>account fully, it would probably be necessary to construct a<br>smaller, simpler model to describe it.</pre>
</blockquote>
Of course, it's not really possible for a sim to take into account the<br>
infinite variations that are possible. I will just try to model things<br>
to what I believe to be a sufficient level of accuracy and complexity.<br>
Any conclusions that can be drawn must be drawn with an understanding<br>
of the simulation's limitations.<br>
<br>
Richard<br>
<br>
</body></html>