While you can automatically set up subject tests from the create page by entering more than one subject, if you want to test something else (like content or audience), you'd use the random limit box, located at the bottom of the targeting create page.
Entering a number in the random limit box will take the total number of activists targeted with your includes and excludes and cut it down to the number you entered by choosing random activists in your universe. For example, if you targeted 5,000 activists using includes and excludes and then set a random limit of 1,000, your email would be targeting a random 1,000 activists in your larger 5,000 person universe.
Random limits enable easy email testing. For example, let's say you wanted to test two different images in your email to see which did better with your activists. You'd start by writing your first email, with the first image, and then targeting it to the set of activists you'd like to receive it. Then you'd enter in a random limit of, for example, 1,000 to target your first subject line to 1,000 random activists in your targeted universe.
Next, you'd save that email as a draft and duplicate it, to set up your second email. Change the image to your second version, and then in the targeting create page exclude the first email you set up. This will target the second email to the same universe with the same random limit of 1,000, but exclude anyone targeted in the first email. So you end up with two emails, each targeted to 1,000 random activists that are part of your universe, with no overlaps so no activist will get an email twice.
Now, you can send both emails at the same time and watch the statistics to see which one did best. When you're ready to send the winning version to the remaining 3,000 activists in your universe, you'd duplicate the winning email, make sure you're excluding both of the test emails on the targeting create page, and remove the random limit. Now your final email will be sent to all activists in your universe who didn't receive either of the first two emails -- the remaining 3,000.
(You can get a head start on this by using the "Duplicate With Exclude" button, which will duplicate the email, exclude the first email from the duplicate, and remove the random limit. You'll still have to exclude the second test you set up manually to get the right targeting, however.)
In this way -- with limits, duplications, and excludes -- you can quickly create complex A/B email tests.