The sample size is too small for that DVR percentage, 80% vs 84%, between watchers and nonwatchers to have any statistical significance. If DVR owners are being excluded from the numbers for all shows, then all numbers are a little low. Total viewership would increase, but share might or might not. We'd need a bigger sample to tell, but this says DVDTalk watchers and nonwatchers own DVRs in essentially identical numbers (which says next to nothing about the general public, as our DVR ownership is 10X the general publics)

I wasn't comparing the 80 percent to the 84 percent, but rather just showing that both sides have a ridiciously high percentage of DVR users. The real comparison that I was making was comparing the percentage of TVTalkers with DVRs to the nationwide percentage of DVRs.

Using DVR users might help ADs ratings, but this poll won't show (and won't <i>not</i> show) that whether it does or not. He was trying to show significance where there is none.