FiveThirtyEight, a website better known for predicting election results with statistical methods, a few weeks ago posted a statistical analysis of 5th edition Dungeons & Dragons characters regarding their choice of class and race combination. The result is boringly predictable: People overwhelmingly choose a race that gives at least a +1, if not a +2 bonus to their main class stat. As the race system has been designed to favor combinations that appear in the cast of Lord of the Rings, you get a lot more elven rangers than dwarven rangers, etc. However that is much less an attempt to emulate the classics but rather simple stat minmaxing.

I don't know how these statistics look for other games or previous editions of Dungeons & Dragons. But I suspect that the "bounded accuracy" math of 5th edition D&D makes a +1 or +2 bonus a lot more important in this edition, because there aren't so many other bonuses around. Unless you "roll lucky" (which is my way of saying cheat with your stats), an elf paladin is simply too bad stat-wise to be playable. That from a role-playing point of view a haughty elf might make a brilliant paladin if played well is of no importance. You choose your class, then take one of the few races that are good at that class, or default to humans who are good at everything.

As a role-playing challenge I would love to try a gnome barbarian. But as a 5th edition character with standard array or 27-point buy stats I'd end up with something not very fun to play in combat. The World of Warcraft system where your choice of race has only a very minor influence on your combat performance is more to my liking.