Games without score were discarded from the chart. I don't know if that was a good way to do it because they were still voted on and that might possibly affect the points of other games. The point was to get a larger data set than our vote. I know the Codex is quite biased but same can be said about our vote as well, or any niche community for that matter. To get unbiased results, I would have to use data from a global vote.
The codex isn't 'quite biased', it's fanatical to the point extreme, it's the kind of site that actively encourages the concept of 'site consensus' as a means to validate itself and it's reasons to tar and feather people.
You say you wanted a larger data set, but the irony is that your primary source is actually a tiny data set anyway.
Take Lords of Xulima, for example. It's Critic score is based on just 9 individuals giving their opinion from a general gaming perspective. One of them out right says "Veteran players will be pleased, the others not so much." - well who's going to be interested in it if not veteran RPG players, who is the reviewer even speaking for when they say that?
Another writes "Lords of Xulima is strictly for the more masochistic of gamers." Which is absurd. Anyone who's even the slightest bit into RPGs will be able to pass through the game just fine, again, who are they even speaking to? Clearly not RPG specialists.
So these 9 people decide the 'longevity' score for something that isn't even their main bag? LOL, don't be absurd. LoX a 71? Absurd.
That data set is SIGNIFICANTLY smaller than the RPGWatch dataset.
The game has 33 User reviews - that's not a 'larger sample size'. As a result just 3 people who gave it a negative review have dumped it's user score down to 7.7. Well, that was extremely easy to bully out the way wasn't it.
Even combining the two, that's just 42 individuals providing your primary reference stat. And in that 42 you have people who don't even care for RPG specialisation.
The RPGWatch poll is actually a larger, and more relevant, sample size than the stat that Nereida is holding up as the DEFACTO reference to a game's UNDISPUTED quality.
So lets now look at steam reviews: 1,110 reviews, very positive, the same as Dov Os and PoE. This isn't to say "someone who likes PoE will like LoX and visa-versa, just that they are likely of similar quality to those who specialize in RPGs.
The RPGWatch 'sample size' was just fine. And provides a more relevant assessment of 'quality'.
And from the larger sample size you took from the codex, your red line of Critic Score versus Codex Score still shows a very similar arc to the RPGwatch one, that, yes, Score has a small effect on the likelihood of a game having a longevity of favour.
Their Site-Score versus User-Score is indeed more marked. And I must admit, I usually only look at user scores on Metacritic, due to the 'official reviewers' usually being just random individuals with no specific specialisation and who knows what axes to grind.