-----Original Message-----
From: [EMAIL PROTECTED] [mailto:[EMAIL PROTECTED]]On
Behalf Of David Emery
Sent: Wednesday, May 01, 2002 1:40 PM
To: [EMAIL PROTECTED]
Subject: Multiple regression and levels of interest
I'm a student trying to see how mulitple regression might be used to
predict levels of interest in interactive video presentations based on
pre-picked attributes of the content in the video. My question is
whether anyone has seen anything done like this before, academic or
proprietary?? Has anyone seen this done using a different method of
analysis?
Since there are potentially millions of attributes that could affect
interest (or some large unknown number), and only 4 are really
examined, the expected levels of significance are very small for each
attribute.
--------------------------------------------------------------------
Another approach would be through the use of Structural Equation Modeling.
Here the approach would be to postulate some key hidden latent variables
that have a direct causal relationship to the levels of interest for a given
content. There may be different groups of (like) attributes that represent
independent "measures" of each latent variable. In this way you can deal
with very large numbers of attributes, and end up with an equation that has
a small number of latent variables (that make sense, but can't actually be
directly measured). Each latent variable may show strong correlations to the
level of interest for a given content. This however requires a lot of
research into past psychological studies, to find out possible empirical
constructs to test.
The method also allows use of categorical and dichotomous data, Likert
scaled survey data, peer assessments, personality factors, etc. that are
more ordinal than interval. Use of regression requires that the numbers
going in represent interval or ratio levels of measurement at the very
minimum. If the data you have is less than interval, you cannot rely on
linear regression tests of significance to determine key attributes for a
given content.
DAHeiser
.
.
=================================================================
Instructions for joining and leaving this list, remarks about the
problem of INAPPROPRIATE MESSAGES, and archives are available at:
. http://jse.stat.ncsu.edu/ .
=================================================================
