I wrote a simple program to go through the first X hostnames in the alexa top million list, and query the sitedossier.com database to compare the count of inbound links pointing to www.example.com versus example.com. Here are the results showing the winners.
(Note that if either version of the hostname has less than 10 referers the site is excluded.)
Top 100: www.example.com: 85 (86.7%) example.com: 13
Top 1,000: www.example.com: 758 (86.3%) example.com: 120
Top 10,000: www.example.com: 4,989 (84.6%) example.com: 902 tied: 3
Top 100,000: www.example.com: 20,371 (79.8%) example.com: 5,090 tied: 63
The aim of the test was to decide which way the convention went in the real world. I see a lot of TV ads that mention "example.com", so I was curious whether this was actually the preferred domain, or whether it was just a shortcut to the "true" hostname.
Based on this test (and a bit of personal bias after using the net for several years) I'll now be 301'ing mydomain.com to www.mydomain.com.