List,
I have a simple design where I compare a single patient's SPECT brain image
to a group of normal controls. For this example, the patient's age is 45.
I have a group of 55 normals, 33 female, 22 male. The age range is 18-55,
mean±SD of 34±11.
Which of the following would be the most valid and accepted way to do the
analysis?
1.) pt vs 55 controls
2.) pt vs 55 controls, with age and sex entered as covariates of confound
3.) pt vs same sex controls
4.) pt vs same sex controls, with age entered as covariate of confound
5.) pt vs same age range ± 10 years
6.) pt vs same age range ± 10 years, with sex as confound
7.) pt vs same age range ± 10 years, with age and sex as confounds
8.) pt vs same sex and age range ± 10 years
9.) pt vs same sex and age range ± 10 years, with age as confound.
The problem is of course that for 8.) and 9.), we have an n of about 6 or 7
controls, which greatly lowers the df.
Thanks,
James
=========
=James C. Patterson II, MD/PhD
=Department of Psychiatry and Behavioral Sciences
=Clinical Research Center
=University of Texas Medical Branch
=Galveston, TX 77555-0431
=Email:[log in to unmask]
=http://marlin.utmb.edu/~japatter/patterso.htm
=========
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
|