Something About Lies and Statistics...

Something About Lies and Statistics...

  • Comments 4

CNet is running a story about reader reaction to the new Windows Media Photo format. From the fine article:

In an unscientific poll asking whether a JPEG competitor is needed, almost half of the 5,621 voters said maybe, "but I'm concerned about it being a Microsoft product."

Kudos for labelling it as "unscientific", but this is really stretching it a bit. The poll results make it sound like 50% of the respondents are concerned about it being a Microsoft product, but the trouble is that the only three choices for the poll are:

  • No, JPEG works just fine.
  • Yes, e-mail and Web pages need smaller files.
  • Maybe, but I'm concerned about it being a Microsoft product.

So there was no ability to say "maybe" without also saying "but I'm concerned about it being a Microsoft product." I think that shows some clear bias on behalf of the poll authors.

My new poll:

Is this blog the best thing you've ever read?

  • Yes!
  • No, I want you to poke me in the eye with a sharp stick

Cast your vote below in the comments section!

  • The other thing I would question is whether the readers actually UNDERSTAND what is wrong with JPEG in the first place.

    Some people even seemed concerned about some sort of DRM in the file, but I had a quick look at the spec and it didn't seem to contain anything like that... I guess that's just a knee-jerk reaction to anything new in "media" these days...
  • No, I want you to poke me in the eye with a sharp stick. :)

    It's funny that you mention this b/c I was just reading an article about other public polls (like political polls, etc.) and how the pollsters can skew the numbers by engineering the questions.
  • OK Josh, where do you want to meet up for your eye-poking apointment? :-)

    But yeah, you can make a poll "prove" anything you want by selecting the right questions, the right audience, etc.
  • One of the recent episodes of "Penn & Teller: Bullshit" on Showtime titled "Numbers" went into a lot of detail about how polls are usually crafted to get the results they want, and how numbers can be skewed.  Very interesting to watch if you can find it on TV.
Page 1 of 1 (4 items)