Are online chess Elo rating tests more accurate than OTB ratings?

There are many chess rating tests online, that measure chess strength, not as a comparative chess strength, which is the case with OTB Elo assessment; but as a positional/tactical ability assessment. I think this sort of chess test is a better indicator of true chess strength, than OTB ratings. 
The chief difference between Elo rating tests online, and OTB Elo tests is as follows. Elo rating tests measure the ability for a person to judge a chess position based on positional and tactical precepts. Is the pawn island defective? Is the king exposed? Is the pawn chain good or bad? Open files? Rooks? That sort of assessment varies in accuracy from chess strength to chess strength. The Grandmaster may opt for a different move in a chess puzzle than a club player, because he sees deeper into the position. Accordingly, computer chess analysis would accord a greater playing strength per position in a chess quiz, for a grandmaster than a club player. 
In an OTB chess game, the Elo strength is calculated as a comparative difference between two players. No doubt in the case of OTB chess Elo assessment, the positional/tactical ability of a chess player is roughly gauged based on the final Elo, but it doesn't say anything about how good at positional assessment a chess player is. And the ability to assess a chess position is what makes a player strong or weak. 
I conclude that Elo chess rating tests online, that offer a series of chess quizzes, with good moves, weak moves, and best moves, gauged by computer analysis, are the best way to really determine a chess player's rating. By contrast, in an OTB game the individual positional skill is not assessed. 

But Elo ratings and their descendants like Glicko and Glicko-2 are designed to be predictive of results, not a test of knowledge.

You don't name the tests you were using, so no one can really assess them. 

online: i refer to elometer. 

test of results vs test of knowledge, two different things. to truly assess positional chess skill, elometer is the only way 

Again, Elo was never designed to be a test of knowledge  It is a predictor of results.

not really lmao

One can have all the knowledge possible about chess but if they cant apply it in games of chess, it doesn't matter. Rating is not a measure of knowledge, but as @justbefair posted, as a reflection of past results in play against others.

Things like puzzles, where you know there is probably something there to find, is completely different than playing an opponent, with the clock ticking, and applying your knowledge to recognize opportunities. The rating you get from playing is in that process and the results gained by your performance.

Elo testing may be in the ballpark of your potential but it doesn't mean you will reach the results in practical play.

Online chess ratings tests are great for measuring your ability to take online ratings tests! OTB ratings reflect your ability to play chess! It's really that simple. 

You will find that waving your online test results in your opponent's face doesn't help much as he is checkmating you

Both are still more accurate than accuracy

according to the topic, online chess has many variants like bullet blitz rapid puzzles. all rating differs a lot from reality. 

sounds like 2200 puzzle player-facing 2200 otb rated player. do they both have even skills? not at all 

puzzles gauge the completeness of your thought process with regards to determining the proper move in a given position & it becomes an extremely complex task to continue solving them correctly as the puzzle ratings go up.

i'm an underrated A-player at only 1851 USCF because i haven't played many tournaments with people at my skill level, but i have a puzzles pb of 3600 here (currently #35 US) without intentionally memorizing any of them.

i've defeated several experts & masters in tournaments at classical time controls, usually winning by navigating tactical situations more exactly as they arise.

The problem is that puzzle proficiency doesn't automatically translate to playing ability.

Finding good moves in a ready-made position is one thing.

Being able to create such positions, move after move, in a game against an aggressive and resourceful opponent, is a different skill entirely ...

I've known plenty of players who seemed to have good chess knowledge but had poor results in actual games.

The reasons varied: Poor attention span. Poor situational awareness. Inattention. Impulsiveness. Bad nerves. Excessive emotional reactions to stress. Distaste for certain types of positions.

Regardless of their chess knowledge, none of these people should be considered as good chess players, since any reasonable definition of the term "chess player" includes an assessment of their PLAY.

the habit of always searching for the very best move in every position is what solving puzzles regularly creates in your play, which is why i feel that a high puzzle rating that doesn't come from simply memorizing the solutions (but rather from doggedly analyzing lines out to the end) could legitimately be considered a clear indicator of practical playing strength.

the harder puzzles are not based on simple tactical tricks, but rather feature all sorts of subtle defenses & surprising ideas that you must foresee in order to be successful, which (in a real game) translates to:

1. increased depth & accuracy with regards to calculating potential lines.

2. careful attention to detail when evaluating the implemention of a plan.

    No. The tests attempt to predict what your otb performance might be. The Elo ratings measure what you actually have done otb. A similar situation--most automobile races have qualifying sessions beforehand and set the starting lineup  according to those results, but they still run the race and the top qualifiers very frequently do NOT win. Speeding around an empty track is different from actually competing against a field of other drivers.