Across all game sizes, each additional player adds an average 9.7 days to the game length. I've had some terrible experiences with 4ers, which skew this trend line. The sample sizes are low as well. These numbers reflect only completed games, and not games where a player disappeared.
I wanted to shout out the players with whom I've had my top 25% fastest 4ers:
Here are the players in my 10% longest 4ers:
There are two accounts on both lists, which shows how this analysis really doesn't work to find slow players, since its often that one player holds up a game while everyone else suffers. As well, a "slow" player in a game may just have a poor timezone or schedule to play soon after the last player. Turn order matters too. Consider players in North America and Hong Kong (16 hour difference) who both check their tables before going to bed (2200 local time) and again in the morning (800 local time). If the North American player followings the Hong Kong player in turn order, he will likely let 8 hours lapse on his turn. If instead he staffs up a military unit and causes the Hong Kong player to follow him, then the lapse occurs during the Hong Kong player's turn. If the other players at the table are in Asia this may reduce the total table wait. If they are in North America it will probably increase the total wait.
Obviously the solution is to check frequently throughout the day so the problem of sleep is the longest delay a table faces.