One: Separate the responses from the print and Electric subscribers (that is, if you want to poll "Electric" subscribers). This would allow you to better define your demographics for advertisers and allow everyone to see the opinions of the readership for both the print and online editions.
Two: Maybe have a mail-in card for print subscribers, each one numbered so fraud is difficult (nothing is impossible), or include a removable address sticker to put on the card (cards without address stickers are discarded).
Three: Do not mention any products in the categories, force people to write them down. While I understand this will make it more difficult to read the results, my experience indicates that some people just check what's available instead of taking the time to write down an alternative to what has been listed.
Finally, forget the 1996 poll altogether. The first has been invalidated by alleged "ballot stuffing" and the second was a survey with what most believe is an unrepresentative sample size.
Thomas B. Stanley
User Support Programmer
Oklahoma Climatological Survey
For questions or comments, email the Forums Editor