Please note, Chess.com policy is that we don't make comparisons with our competitors in our own forums, so this conversation needs to stay at a general level without making such comparisons.
As to the rating differentials, it's not an absolute measure of your actual playing strength, but a relative number indicating where you stand within that particular group of people. So it depends on who you're playing with.
Not sure about the ratings. My rapid rating is higher on a certain other prominent site than my daily rating is here.
Ratings here begin at 1000, which could be a factor when comparing with a site where players begin at 1500. A site with a high turnover can also have high variance in its ratings. Let's say a strong player comes here and is given a 1000 rating and then wins a few games against highly-rated players, taking rating away from them, and then stops playing here. That's going to bring some numbers down.
Because not everyone plays on all sites.
Imagine a club where only GMs are allowed to play and everyone starts at 1200. After a year, some may be under 1000, the top guy might be over 1500.
Conversely, in a club full of beginners, some beginners will be rated 1500.
---
So that other site might start players with a really high rating, like 1800, or chess.com might have better players on average. Maybe a little bit of both.
when I tell people my online rating I don't use my chess.com rating
Experienced people know there are differences though. If you tell them your _____ (webiste) rating is 1800, and they don't know that site, they might ask you for a different rating.
If they do know the site, and also know chess.com then they'll know you're rated a lot lower here without even asking 
Well, it depends how many games you played on each, how good you played on each and how much you won or lost on each. On one you might have 17 wins 10 loses and 0 draws and the other 24 wins 6 loses and a draw then they would be different. And depending what the opponents rating is, it changes what points you get or lose when you win or lose.
Yeah, my examples were just on the extremes.
Intuitively, I would expect the average skill for large sites to be roughly equal. If beginners are started around the same rating then it's certainly unexpected. (Although chess.com uses the Glicko formula which makes fast adjustments to new players so not sure how much new player's ratings matter).
Another thought is it can be a marketing decision. If you make the ratings on your site higher than average, you might expect to attract more players (?) Such adjustments are possible to make from outside the rating formula i.e. a site can instruct programmers to artificially inflate the ratings.
By the way, if artificially inflated ratings attract more beginner level players, that might be a feedback loop because having more lower rated players also inflates the ratings.
It could also work the other way. Strong players may want what they see as the more difficult rating, and to play on a site with less beginners, which in turn makes the high ratings on the site even harder to get.
Another thing that could raise ratings is a high turnover rate.
Lets say site X has much better than average advertising. They attract many new players. Lets say 100 new accounts a day (just making up a number).
Lets further assume 80% of those noobs quit after losing 5 games (because chess is way harder than they thought and they don't want to play 1000000 games just to get better). Losing games then quitting injects points into the rating pool.
Another factor is certain sites restore ratings if you lose to someone playing with assistance. Certain others sites (cough cough) do not. That alone has to be at least part of the difference. Think about it. Someone makes an account here and cheats. They take X points out of the player pool, get banned, then those points are gone. That same player then makes a new account here and repeats the process. Over time that amounts to a lot of points disappearing from the pool. If that same player does the same thing on a site that refunds points, the points don't leave the pool when the cheater is banned.
Another factor is certain sites restore ratings if you lose to someone playing with assistance. Certain others sites (cough cough) do not. That alone has to be at least part of the difference. Think about it. Someone makes an account here and cheats. They take X points out of the player pool, get banned, then those points are gone. That same player then makes a new account here and repeats the process. Over time that amounts to a lot of points disappearing from the pool. If that same player does the same thing on a site that refunds points, the points don't leave the pool when the cheater is banned.
Good point.
Although chess.com blitz is pretty close to OTB... at least if a person is 900 blitz here like the OP it's pretty much guaranteed they're not 1400 strength OTB.
So I'd tend to think of other sites as inflated and chess.com as close to "correct" (whatever correct means, in my case I'm thinking OTB though).
Legit, I didn't expect such a big difference here. Not that I have a problem with it, but when I tell people my online rating I don't use my chess.com rating because it sounds pretty bad.
I'm actually 1400 on another site and 900 here, not sure if this is how it should be but goddamn. Why is the difference so huge?
That other site has super-inflated ratings, so inflated that they don't even mean anything. They also consider 8/0 Classical, so...
Well yeah, that does make sense.
On another site, I am +2200 in blitz while on a third site, which is a pay site, I rarely break 2000.
Ratings reflect the population. Chess.com is beginning to attract more serious players, which is great for me, but my rating will reflect the tougher competition.
One thing the site could do better regarding ratings is to have a "tournament" rating and a non-tournament rating. The difference is that people who never play any tournaments can pick and choose their opponents. They can, if their opponent is willing, play the same person over and over again, gaining rating points as they beat up on key opponents and stop playing tough opponents.
In tournaments, you don't have any control over who your opponent is, and you have to play against tough opponents. People who play exclusively in tournaments here will generally have lower ratings than those who never play in tournaments. Another site recognizes this and has pool ratings for those who agree to play random opponents.
I don't usually tell people my rating when they ask, because I have like 30 different ratings and they all mean different things. I just tell them I have a lot of different ratings and I don't know them all and don't care what they are anyway.
A friend of mine would, when he was at a tournament and was asked what his rating was, would always stick out his hand and say "John, what's yours?"
Legit, I didn't expect such a big difference here. Not that I have a problem with it, but when I tell people my online rating I don't use my chess.com rating because it sounds pretty bad.
I'm actually 1400 on another site and 900 here, not sure if this is how it should be but goddamn. Why is the difference so huge?