What open source multivariate or a/b testing frameworks or libraries are available?

Google Website Optimizer http://www.google.com/websiteoptimizer and Visual 开发者_开发问答Website Optimizer http://visualwebsiteoptimizer.com/ both allow me to run experiments for individual sections of content on web pages. But, what libraries or frameworks are available for running tests on things other than snippets of HTML?

In other words, are there open source projects which provide the underpinning logic and calculations required to build my own multivariate tests?

One such library is ABingo http://www.bingocardcreator.com/abingo in Ruby. But, what other libraries/frameworks are available?

From this mashable article a-b-testing-resources I ended up finding vanity and genetify. Vanity has a prettier documentation but genetify looks like solid code.

Both are hosted on Github which is always a plus, and of course a lot of A/B testing is done on HTML snippets but I think both of these are extensible enough so that you could possibly enhance or customize them to suit your needs if they don't do so out of the box.

Vanity Screenshot

sidebar_test.png http://img690.imageshack.us/img690/7246/sidebartest.png

Genetify Demo screenshot-with-shadow.png http://img824.imageshack.us/img824/4526/screenshotwithshadow.png

I'd advise against using Google Website Optimizer and other javascript-based A/B packages. They drag your site's performance down and kill off any SEO you might have had, both of which affect your bottom line. Since you're probably using AB testing to make your site better at converting random strangers to money, it would seem counterproductive to do anything that damages that.

Look into something that runs on the server, such as ABingo for Rails and FairlyCertain for ASP.NET. Both of those are open source and free.

I recently found a Ruby on Rails gem Mountain Goat which provides in-house a/b testing and high-quality analytics for any Rails application. Mountain Goat provides an entire suite built into your application to manipulate and play around with your a/b test and conversion data.