(Note: I hope the following isn't considered cheating or spoiling - one could easily calculate my results without simulating the starting location; there are only 7 more or less common tiles that matter, so the results apply to most situations and not only gotm1)
I recreated the exact starting location on a duel map for testing purposes, and played the first 50 turns over and over again to get a definite answer on when to build the worker.
There are differences in the outcome, but not as big as you might suspect.
It's obvious that settler first is bad, since it takes 25 turns, and there's no way to get a total population of 7 in the following 25 turns, at which time your capital otherwise can pump out settlers at rate 8-9.
So I started with worker or warrior, built granary, then barracks, finally a settler, and always did one cop. I didn't change gov or religion to get comparable numbers of turns. The results are as follows:
Worker first: Everything finished after 53 turns. Total science value 1090, improvement value 33 turns (5 were lost due to movement and waiting for techs)
Worker at size 3: Also exactly 53 turns. Science 990, improvements 25. A small production bonus (obelisk prebuild).
Worker at size 3, but chop to rush granary before irrigating the corn (my worker even waited 2(road)+2 turns on the silk for bronze working): Took 1 turn more (54), but 1 additional warrior was built. Rest exactly as above, 990/25.
Worker at size 4: 54 turns, science 930, no additional warrior.
Conclusion:
It doesn't matter much if you build your worker first or at size 3. Building it first looks slightly better due to science (100 equals about 4-5 turns at this state of the game) and 2 more improvements, but waiting until size 3 might give you the extra early commerce to grab a religion - plus, you don't have to fear to lose your city immediately to barbarians

. It might even pay off to wait until size 4 if you deliberately focus on founding an early religion.