• - WebmasterWorld
• -- Ecommerce
• ---- Trust Seals

arubin - 5:06 am on Nov 20, 2007 (gmt 0)

Thank you for posting. I love this statistical analysis and post the results of the sort of tests that I conduct on my sites on my blog. But the sample size doesn't allow us to draw any meaningful conclusions. I'll show you why:

Using your second, larger set of numbers:

For your seals off case, at a 90% confident level your margin of error is +/- 4.677% So you can be 90% confident that your actual conversion rate is from 0 - 7.51%

For your case with the seals, at a 90% confident level your margin of error is 4.674% So you can be 90% confident that your actual conversion rate is from 0 - 8.964%

If you want to graph it, the area of overlap is majority of the possibilities.

May sound boring, but basically the variation your seeing will actually translate to a flat or decreased conversion rate a large percentage of the time.

It doesn't take that large a sample for a jump in conversions of the size you are reporting. If you can keep the test running longer you can get some meaningful data. If you can post it again then you'll be doing everyone a service.