Here's another thought I just had as I continue to analyze this and other data. Let's assume that all the viewing and previewing activity is being captured in my report, even from activity and communication streams.
Could it be that using Jive 5 is more efficient than using Jive 4.5, and therefore people do less clicking around, therefore recording lower levels of "activity"? They can skim content now, with no need to even expand or preview much of it if it's short or if they see from the bit displayed in their activity stream that they aren't interested in it. So they can consume more content even while registering less activity in the tracking reports.
I truly am not trying to force an explanation. I just want to find one. If the upgrade has had the positive effect of stimulating more contributing activity, it would seem likely that people are spending more time in the site, too -- they are having a more positive experience. Yet overall activity levels decreased slightly. The question is, why?
Interesting observations. Just to clarify something that I think I have been inconsistent about in this space, here is the deal with Views and the activity stream
(1) Just having something show up in a person's stream does NOT mean that a View gets registered for them
(2) If the user click "Full Preview" to open up the full content, then a View DOES get registered
So what you propose is certainly plausible: that a more stream-centric interface might lead to fewer Views since the content does not need to be Viewed to browse it.