Latin Square Design
Why is it that if there's a main effect of significance, it means that there's something weird, and when there's a main effect of order, it means that time has come to play? Shouldn't it be the other way round? What's the rational behind that?
posted by Karlin @ 9:51 PM
1 Comments:
At 10:00 PM,
Mari said…
I'm not certain what you mean by "a main effect of significance," but I am guessing that you mean "a main effect of sequence." Remember that sequences are the rows in a latin square and orders are the columns. Remember also that I told you that I think that these are badly named.
If the orders differ from one another, that means that there are differences between order 1 (at which point participants had received only 1 treatment), order 2 (when they had received 2), ... and order n (when they had received n treatments). Because all treatments are included in all orders, it is not a treatment effect, but rather a difference across amount of treatment/time.
Sequence, in contrast, is the sequence in which treatments were applied. If there is a difference by sequence, it isn't a treatment effect (because all sequences get all treatments) but rather tells you the the sequence in which treatments were applied is important. Such might be the case if you were looking at a dismantling study in which relaxation training, imagery, and exposure were your three treatments. That is, the outcome might be best when these treatments were administered in this sequence (as opposed to, say, exposure, imagery, relaxation training).
Post a Comment
<< Home