Player choice of unusual races within D&D classes

Poking about for new datasets I found a favoured class/race pick set and saw a weird pattern. It looks like certain class/race combos are so favoured it resembles the old 'race as a class' set up.

When you do a heatmap of top 30 class race combos they mostly cluster in the 'big races' with a few outliers. These are some of the stereotypes you may have encountered - half-orc barbarian, dragonborn paladin, firbolg druid. Note the numbers are not important, the distribution of the colours - clustered at the top - is the message here.

I got to wondering what the ratios for 'common-to-unusual' races were within each of the classes. Were there some classes that people cut loose and went with any race? Other classes with such an obvious optimal synergy in the common races that people never tried anything else?

Looking at the ranking of the classes - apparently Paladins are least likely to have an unusual race, barbarians most likely. There seems to be ~ 3 clusters - Paladins, Rogues, Clerics and Warlocks have an obvious 'right answer' that people stick with, Sorcerers, Wizards, Fighters, Bards and Rangers are in the middle and Artificers, Monks, Druids and Barbarians are much more likely to have people experiment with unusual races.

I pulled out a few other data sets to see if this was visible elsewhere - the 5-thirty-8 set, the OGANM app, and two more surveys including one from RPG-net in 2017 and Mia Gojaks D&D 5e facebook survey.

Looking at them all together two things jump out - first that the 'general players' on the App data run less unusual races in the first place (~ 1 in 8) compared to 'forum players' (~ 1 in 5). As a quick recap - we can consistently see on apps related to the game, e.g. D&D Beyond or other tools, that there is a consistent difference between these general gamers and those who respond to surveys on online forums, twitter, facebook, etc. General gamers like fighters and rogues, enthusiasts from the forums like spellcasters more.

Second the groupings broadly hold up - Rogue, Cleric, Warlock and Sorcerer seem to have low tendency to be an unusual race. Monk, Barbarian and Artificer seem to be high. The other classes vary between source but seem to be in the middle. Fighter and Druid stick out as being more likely to have an unusual race from forum players compared to general gamers.

In the 5-thirty-8 data, which is an extract from D&D Beyond, Barbarians being 'unusual' is driven by half-orcs and goliaths being two of the main choices and those races being outside what the 80% of players choose. From all the forum datasets we see the half-orc barbarian again and the tabaxi rogue standing out - otherwise it is a long tail of mixed choice.

What most surprises me is the absence of this long tail from the lower ratio classes; not just that they race picks are restricted to the top race choices but that they are very specific within those.
Warlock is the most severe - only humans, half-elves and tieflings need apply. Who knew the granters of eldritch pacts were so picky?
Clerics also somewhat surprising with human, elf, dwarf or aasimar being the strong consensus as the only way to go.
Rogues accept a few more options with halfling, human, elf, half-elf, tiefling.
Paladins can be human, half-elf, dragonborn, dwarf, aasimar.

If this was all driven by 'general players just want to stick with vanilla race/class combos' then fighters and rogues should look the same and wizards and barbarians should look the same as they are roughly equal as pairings in popularity among general gamers. Instead we see huge differences - rogues have a restricted race selection, fighters less so. Wizards are in the middle like fighters and barbarians are what you play if you want to play an unusual race. That rogues and fighters, wizards and barbarians are so different says that something else is going on beyond 'general players do not like unusual races'.

Assuming this is all mechanically driven - that those races relevant bonuses synergise so strongly with those classes - we have weirdly gone almost a full circle back towards the old 'race-as-class' from previous editions.

I am going to dig at this some more, see if I can find a better to highlight what is going on here.