On Thu, Nov 18, 2010 at 2:48 AM, Stephen Smith <[log in to unmask]> wrote:

HI - this is indeed the generally recommended approach - however, IF the trend is truly linear then you will get a slightly more sensitive analysis having only one EV (or one for mean and one for slope). Your residuals will go up (unless things are perfectly linear), but so will your degrees of freedom. I'd be surprised though if this makes things better.

Cheers

On 17 Nov 2010, at 21:04, HC wrote:

Hello experts!

I have a design in which I have 4 levels of a treatment: 1, 2, 3, and 4. I predict that signal should be a monotonically increasing function with 1 having the lowest signal and 4 having the greatest. What is the best way to set up my EV files to test this?

In the past, I have tried 1 EV per treatment level, and then used a contrast of :

level 1   level 2   level 3   level 4
  -3          -1          1            3

But this method seems to not be quite sensitive enough to detect the effect. Any other suggestions?

Thank you in advance for your help!

---------------------------------------------------------------------------
Stephen M. Smith, Professor of Biomedical Engineering
Associate Director, Oxford University FMRIB Centre

FMRIB, JR Hospital, Headington, Oxford OX3 9DU, UK
+44 (0) 1865 222726 (fax 222717)
[log in to unmask]    http://www.fmrib.ox.ac.uk/~steve
---------------------------------------------------------------------------