Prev: mscohere - window size
Next: Link axes with linkprop
From: Rick Tanney on 11 Aug 2010 13:09
I am planning my dissertation data analysis, and am looking for a visually effective method. The data is ordinal: the correct result would be a list of locations ordered from near to far, probably 20 or 30 data points. Respondent results will be the order as perceived by the survey subject. The mnrfit function gives a smooth curve when responses are correct, and shows increased crenelation as results differ from the predicted values. This is visually effective, but I do not know that it is a statistically sound approach. What is your advice?
From: Peter Perkins on 12 Aug 2010 16:12
On 8/11/2010 1:09 PM, Rick Tanney wrote:
> I am planning my dissertation data analysis, and am looking for a
> visually effective method. The data is ordinal: the correct result would
> be a list of locations ordered from near to far, probably 20 or 30 data
> points. Respondent results will be the order as perceived by the survey
> subject. The mnrfit function gives a smooth curve when responses are
> correct, and shows increased crenelation as results differ from the
> predicted values. This is visually effective, but I do not know that it
> is a statistically sound approach. What is your advice?

Rick, I'm not exactly following how you intend to use MNRFIT here. But
this sounds a little bit like a classic experiment that tested how well
non-experts perceived morse code. There's a demo of using
multidimensional scaling on those data here:

<http://www.mathworks.com/products/statistics/demos.html?file=/products/demos/shipping/stats/mdscaledemo.html>

No idea if this is helpful or not, hoping it is.
 | 
Pages: 1
Prev: mscohere - window size
Next: Link axes with linkprop