Well, guys, here are some stats I ran last Feb for Sep2002 through Jan2003 (for a former employer). "Big" bet is defined as having a LAY(risk) > $500.
Source Size Hold%
------ ---- -----
Net Big -0.4 (yep, book lost these)
Net Small +6.2
Phone Big +3.1
Phone Small +9.7
In both cases, phone and internet, the difference between big and small was 6.6%. You can look at it as 3.1% to 9.7% is a 6.6% difference or as winning three times as much.
Revere, despite the size of the big bets, the book actually won more from the smaller wagers. (This does not take bonuses and stuff into account. To be honest, it was so long ago, I don't remember if these number included parlays or were just straight bets. It was a preliminary experiment for a debate on wager limits that never got further analyzed.)
Now, the book did have a tendency to allow in sharp players so they could lay twice as much at the same number at other books, before the sharp players caused DonBest to move. So, the book was willing to take a loss on those bets, in order to make even larger bets themselves with the sharp information. As a result, the hold % for large bets may be slightly lower than a normal book might accept.
My personal opinion is to keep the limits low and stick with DonBest. For a really sharp player with good information, it isn't worth the trouble to bet a place with $500 and $1000 limits. A book can always make a case-by-case exception for known squares with deep pockets, but this should be a rarity. So, by keeping the limits low, you keep out the really top players, that are going to beat you in the long run.
My other opinion on this is: NO BONUSES FOR THE BIG WISEGUY PLAYERS. That may sound harsh, but these guys want to play because they have better handicapping, inside info, or need an out. Why give them a bonus for the privilege of coming in to take advantage of you?