SPSS/Stata Repeated Measures - Contrast question

Question

I have data from 8 subjects collected at 5 time points. I would like to see if there is a difference between time points (1 + 2) and (3 + 4 + 5) using either SPSS or Stata.

When in SPSS I use 'Analyze > GLM > Repeated Measures' I am able to specify the 5 time points, but I can not create contrasts. I can only select the predefined contrasts.

I guess I could just average time point (1 + 2) and (3 + 4 + 5) and compare the means in a paired-sample t-test.

Q1) Are there any disadvantages with averaging the time points and using a paired t-test (e.g. loss of power?)

Q2) How could I model these contrasts in SPSS or Stata?

Answer 1

In SPSS, you have to use the MMATRIX subcommand of the GLM procedure. In your case it would be (assuming you want to test that time 3 - 5 are higher on average than 1- 2):

/MATRIX = "last three higher than first two" ALL -1/2 -1/2 1/3 1/3 1/3

In STATA this requires the manovatest postestimation command of the MANOVA procedure:

gen groups = 1 //since all participant belong to a single between-subject group

MANOVA time1 time2 time3 time4 time5: groups, noconstant

Matrix M = (-1/2, -1/2, 1/3, 1/3, 1/3) // the desired contrast weights

manovatest groups, ytransform(M)

All this and much more (covering between-, within- and mixed designs, as well as contrast specific effect sizes) are covered in: http://pareonline.net/getvn.asp?v=23&n=9