You could very well be correct, however having seen the database while it was open to the world, it's simple to see which records contain the completed form and which do not. If they wanted a close-to-accurate number, all they have to do is count records which contain all fields filled in, instead of only registration ID # and email address (A record was added with the email address immediately on its submission, with a registration ID # being auto generated, and the ID # was used in the URL sent to people to complete the form). Of course, this number would include anyone who completed the form multiple times with different email addresses, but at least it wouldn't include the email-only records.