I think your anomaly is actually an artifact of a poorly designed survey. You asked participants if they support bands through “buying their music AND merchandise.” You did not parce out if they were buying both or one or the other. Therefore, participants can download 100% of their music through torrents or other means and still truthfully select this option, because they buy MERCHANDISE.
I can support this theory with anecdotal and experiential evidence. I know I and most people I know behave in this way. I know many bands and artists who give away their music freely, and make their living on merch sales (with a small number of purchased downloads and CDs by people who elect to buy those things because they want the hard copy). I participate in many online fandoms and communities surrounding artists, and know this is not only how consumers of their work behave, but also how many artists have come to EXPECT their audience to behave, adapting their business model as such.
Your not going to be able to make relevant inferences without separating these two behaviors: purchasing music and purchasing merch.
You also found that one reason some people illegally download music is because they don’t believe the artists need their support because they’re already famous. But “artists” were treated as a monolithic block in the survey, whereas there may be significant behavioral differences according to different kinds of artists. I suspect, again from experience, that you will find higher levels of music purchasing for artists that are independent, less mainstream, or just starting out. Someone can behave in different ways towards different artists, but the survey doesn’t allow for that to show.