Thursday, February 11, 2010

"This is not a scientific poll..."

Internet polls. If you just have the slightest knowledge about statistics, you know that internet polls are infamous for have large biases by itself. The design of the questions are often equally bad, with unbalanced wording and lacking of a "Don't Know" or similar option. And also where the poll is posted does matter (do you think a poll on gun legislation would yield the same result on the websites of Guns&Ammo and the Nobel Peace Price?). This has been discussed before, and not my point of this post. And this has been pointed out to the media several times.

But how does the paper deals with this? VG does this by adding a disclaimer under their polls: "This is not an scientific poll. It only reflects the view of those internet users that has chosen to participate." They also refer to it as "Just an assessment of the opinions, not a scientific survey".

Well, guess what, VG. "Not scientific" is not a property like the color of a car. Being "not scientific" means your results says s¤%t about the general opinion, especially on a biased question hidden in an online article where it's possible to vote as many times as you like...

1 comment:

Wilhelm said...

Just out of curiosity; what would a "scientific poll" look like?