On the overall picture, that is what data shows, sure that there are cases that do, but that is unavoidable.
This could serve to help track down merit abusers, but it is not my primal option, since the objective was a different one and thus the focus. For example, injective merit abuse (unidirectional from A->B) would have a 0% Reciprocal ratios and so the user would seem like an Ok user in this analysis.
To help track merit abusers there are other kinds of analysis that can be performed (I did one way back with a link to the file that could be used (see Additional sMerit Analysis (smerit.txt)). Its two months old now and was rather complicated to follow, so I didn’t insist on the line of work. 

Yes, well pointed out. There’s a 7 point average difference between reciprocal network size and awarded merit to the reciprocal network. That gap is even greater when we look at the Rank breakdown of the data, and in my particular case it adds up to a 20 point difference!
That mean, as you say, that we do send more sMerit to reciprocal users than to non-reciprocal users. Nothing wrong with that really, since we favour the lecture of their topics and encounter them often on the posts we read.

That kind of sums-up nicely the non-numerical view, and gives us most of the practical reasons behind our choice of merit awardable posts.

I’ve modified the OP to break-down the aggregate data by groups of sMerit Awarding Totals, which allows us to see the view you mention. The average for those groups don’t differ too much from the global averages, although the [40..49] sMerits Sent group does stand out a bit more.
We can also see the extremes differ too for obvious reasons: The [1] sMerit Sent group is the lowest reciprocal group of all, followed by the [500+] sMerits Sent group.

Many of the examples you provide on the list have no reciprocal sMerit involved, and are thus with 0% ratios. This is rather much like what I stated above when answering to mdayonliner (Injective only sMerit (abusers) show up on this analysis with 0% Reciprocal Ratios).

I think the additional data break-down I added to the OP lets us see the information for groups of sMerit Senders that do not have Merit Sources. Comments replied to LoyceV’s post apply, so I think we can see the distribution pretty well without deteting the Merit Sources from the analysis (All 80 could be deleted, but I only know a bunch of them really from their explicit comments on the board).

I’ve added a full list of uses with their ratios here: Bitcointalk Reciprocal aggregates by user 20180531. I don't normally use Google Sheets, since for large data it's a bit slow, but I cloned my Excel sheet there. I did the exercise both from the Sender’s point of view an the Receivers, so the list has all the data to both sides of the equation (but the OP concentrates on the Senders).


Any percentage on the ratios can be good or bad depending how it is eventually played-out. For example, we can have 0% ratios but the users behind being injective (unidirectional) merit abusers. We could also have 100% Ratios but with only 1 sMerit involved.
The other factor that helps is to contrast with the amount of sMerit sent involved (which I have added to the OP).

I figure that 90% sMerit sent to Reciprocal sMerit with 2 basic receiving users may indicate something weird (above a certain Total sMerit threshold), while the same happening with a stronghold of 30 users is not the least bit suspicious.
I’ve added a full list of users with their ratios here: Bitcointalk Reciprocal aggregates by user 20180531

Yes, quite a silly claim based on 3 sMerits given vs 1 sMerit “return”. I contested the post ( Re: [New format] Merit Abuser Gang [Now 18 Cases and 14th yet to be completed]).
I even think the user is creating some decent post now, but I don’t like being dragged into vendettas based on thin-air and have thus refrained from enhancing the size of my sMerit Network in this case. The accusing post was actually posted in various sections, but I only answered in one. Anyway …