Why are alexa and webpagetest results different?

According to webpagetest results SiteA is faster than SiteB from several different locations. However on Alexa siteinfo page SiteA is showing as slower than SiteB.

What are the parameters due to which this can happen?

Alexa numbers come from whatever users visit either site and happen to have the alexa toolbar installed. There are a LOT of demographic variables to control for that make comparing numbers between sites difficult. I also have no idea what the mix of the browsers are for people that have the toolbar (is it mostly IE6 users?).

Depending on the magnitude of the difference it could also be within the noise for the alexa data. Real-User Measurement (RUM) data for performance requires a LOT of data points to have high confidence.

Finally, make sure to also look at the repeat view performance. The real user performance will be a mix of first view and repeat view (as well as same-session browsing) data.