Economic theory states that the value of a security will drop by the dividend amount at the same instant that the dividend is paid (or committed, or allocated, or whichever equivalent verb you want to use). But I would contend that the theory here is outweighed by a lot of other variables that impact the price of that security. If you assume that all the other stuff is unpredictable, I'd prefer to capture the dividend rather than sell the security immediately before it's paid - if you sell on day 29 of a 30 day period you're forgoing the return for that month that you've already held the asset. Now you can argue that the market prices this in, but I'll contend that IF it does, you aren't losing anything by waiting a day... and if it does NOT you might be gaining something.
And I can't imagine any case where you would not want to get the dividend, regardless of having to pay taxes on it. Paying 40% of a dollar is a lot better than not getting a dollar at all, right?
An interesting way to back-test this would be to go find something that pays a dividend on a very regular basis, then compare the closing price on the day before the dividend and the day after. Again in theory the difference should be exactly the value of the dividend itself... but I'm willing to bet at least a nice dinner that it's not. ;-)