Running Deviation Contrasts in SPSS

[kyril]

---

Members do not see this ad.

I know that I am likely missing something big, but i just handed in a homkework assignment in which we were supposed to run a bunch of analyses (ANCOVA mainly) on a standard set of data in SPSS. One thing we had to do was run a deviation contrast twice-first and last omitted. What is the benifit of running a deviation contrast? I read the bits on deviation contrasts in the Andy Field and the parts on post hoc tests and planned contrasts, but I still don't really know how interpret my findings, and it just makes me think that either I interpreted them wrong or it's wierd that we were told to run them. My IV was composed of 6 independent groups, IV sig related to the DV both w and w/o controlling for the covariate. When omitting the last group (thus using it as the ref group) I got sig results on all other groups but when omitting the first group I only had a few sig results, and at a lower level of sig. Now the last group had been found in an earlier linear regression to be sig related to the DV, whereas the 1st group wasn't--it seems like these diffs could have lead to these differences, but maybe I'm totally off base? Ultimately i still don't really get why i'd run a deviation contrast as opposed to a simple one, or devising my own.

Any insight would be much appreciated--this is totally going to bug me til I understand it!