Q. Why did you include the "silence" condition? My initial guess is that you included it as a "control" condition to which you will compare sound1 and sound2. If so, I cannot imagine why you would want to exclude it from your model.

What are the sounds for the sound1 and sound2 conditions, btw?

Hi Bruce,

Thanks so much for your reply!

Sound 1 is consonance (harmony), sound 2 is dissonance (tension), which are expected to influence how one perceives viewed faces' intention to socially engage. I'm employing adults.

Indeed, I introduced silence as comparison and an answer to a question: what judgements are made when there is no music at all?

The rationale for splitting sound and silence analyses is that I'm predominantly interested in consonance and dissonance effects. Also, when I am running separate analyses, the 'narrative' is neater - I am comparing these results to an infant study I run earlier, which did not have a silence condition (as it was a replication of a specific design that did not introduce silence).

Is it justifiable to split the analyses for sound and silence conditions, or would it be a fallacy?

Many thanks!

Tatiana

Hello Tatiana Sobolewska In this matter that you mentioned, I like the model including the interaction factor more. Of course, this is my opinion.

In the following model, it means gamma

yijk=μ+αi+βj+γij+εijk (i=1,2,….,a j=1,2,….,b k=1,2,….,n)

Of course, maybe using a covariance model (instead of the variance you said) is better. By using the covariance model, you can control some confounding factors. This means that the model below may be better than the model above.

yijk=μ+B(xijk-x ̅ )+αi+βj+γij+εijk