There is a crucial point for us to consider: Users aren't behaving the same way here. If one would've asked me to divide the users of chemistry.SE I would divided them into groups:
Teachers (in the literal sense):
The fundamental answerers of our community, these guys are glad to help others with their chemistry problems. They're either devoted to this community, or just visit occasionally or rarely to provide exceptionally awesome answers to Qs we have. These guys don't ask questions to even worry about this issue. All hail them! :-)
People like me ask questions in here. Their point of view is really important as an answer to this question. Personally I always let answers remain for a couple of days or so, so the community will be there to decide which are helpful and which aren't. Nice answers get the votes they deserve (hopefully) and incomplete or wrong answers are given a chance to be identified and corrected subsequently.
The Devoted (askers that answer Qs as well):
This post concerns them as well. Usually with relatively close amount of answers to questions, they also build up what's the gist of this community.
Newbie's in a pursue of what they've posted:
New users are a main concern of this post. There are new users who do want to find out what SE is about. They ask a question, and always check for the answers provided, simply because they're curious enough to do so.
Mainly concerning newbies, these are people who ask something and then instead of being patient for a nice response I looking for quick posts even if they're wrong. "PLEASE ACCEPT THE ANSWER" is for these people, not users who find themselves responsible for expanding our community.
We should think of different best approaches for choosing the accepted answer for each of the types of askers:
- In my opinion, persistent askers (learners) are best to choose an answer when they're sure no other response will be provided and in an overall sense, the views of the question won't arise but for time-to-time views that are random.
- The devoted are better people than me, and I can say for sure they're very familiar with what this is all about. They can provide their responses as great answers to this post. I think it's possible they have a better notion about dealing with the time of acceptance.
- Newbie's that are wary of the response to their post usually get to do the best thing they can, but sometimes they accept answers too soon. I think if they will come back after a couple of days, it's best to choose the accepted answer by then.
- Freddy's right about some people not even being aware that such a feature named accepting an answer exists. Well, I have no idea about how to inform them about doing the job.
As a summary, I would suggest that the later an answer is accepted, the better. But, there are two issues that have to be taken into account: First, when the question lost the connotation of being a "new" question, it mostly will get "random" views. If you accept an answer very late, you might leave the negative impression on a viewer that the post isn't "status-complete". Second, if it'll be hard for you to do future visits, it's better to accept the answer right away than leaving it for eternity.
This is only my insight to the problem. Please feel free to add yours. Plus, I can't think of a way of informing people about the issue and about when the community sees the best time to pick an accepted answer.