With in Google Analytics you can use Google Experiments to set up a test. (It is located under the Behavior list option on the left hand side.) The interface will walk you thorough setting up your test. It will ask you the % of your traffic you want to include in your test, what goal you are measuring, what confidence level you want to achieve, how long you want to run the test, and it will also ask for your control page URL and your Experiment Page URLs. It will then provide some code for you to add to your page. This code will split your traffic although it's not necessarily going to be a 50/50 split. Google periodically analyze which page is reaching your Goal and will adjust the percentage of traffic that is being sent in order to favor the page that is winning. Once the Experiment runs its course you should have some good data to help optimize your pages.   This link will walk you through everything. Good Luck!!

I would be interested to see whats going on with your "long tail game". Long tell keyword for me = money, So... I wouldn't jeopardize ranking for "Red Polo t-shirts" just to rank better for Polo t-shirts.  I would suggest maybe boosting up the unique content on your Product pages to make them rank better for specific long tails. I like the fact that all of my product pages are indexed individually just for the simple fact that I have a customer base that google's Part numbers. Those product pages show up every-time and almost every-time it rings the cash register.  This just might be something to think about... but if all of your product pages are duplicates then they are not helping with the long tail so I would say canonicalze them and redirect them to get rid of the dups and see if it helps with those head terms.

Hope this helps
Chad