Hi all!

I've been looking into the effects of pre-order promotion design on pre-order sales of videogames.

My IV"s are binary, 1 moderator (AAApublisher, referring to firm size) is binary, and my second moderator (Sentiment, low is negative, high is positive) is continuous. Sentiment has some missing values. My DV is number of pre-order sales, which I've had to log-transform due to its distribution.

I've attached my conceptual model.

When I make a simple linear regression model with only my IV's, the results suggest that they've got decent explanatory power of the DV, but when I add my moderators, these main effects disappear. As I understand, these moderators (and not my IV's) are actually the cause of the changes in my DV.

Looking forward to hearing your thoughts on these results. Thanks!