Submit Hint Search The Forums LinksStatsPollsHeadlinesRSS
14,000 hints and counting!

Click here to return to the 'pool structure biases to 'commented' level?' hint
The following comments are owned by whoever posted them. This site is not responsible for what they say.
pool structure biases to 'commented' level?
Authored by: Herve5 on Jun 19, '05 04:11:58AM

While this is quite off-topic, it really strikes me that those poll's levels that are associated to textual comments (whether they are positive or negative) attract *much* more votes than the other ones. I wonder whether one could deduce some "uncertainty bars" on the poll levels...


[ Reply to This | # ]
pool structure biases to 'commented' level?
Authored by: Han Solo on Jun 20, '05 08:59:49PM
It is certainly the case that the labeled responses attract more attention -- but also no surprise: this type of "framing" often affects surveys and other social experiments. However, these comments are, IMO, necessary in a survey like this: what does "6" mean exactly? Without some "guidance" or framing, one person's 3 or 4 may be another person's 7 or 8... and thus the aggregate results would be completely uninterpretable.

Error bars don't seem to be appropriate here: in no way can this survey be considered a random sample, so there is no statistically valid way to construct such error bands. Best one can do is look at the histogram and draw one's own conclusions. The median and modal responses both are 7 (53.2% of the mass, as of right now, was at 7 or higher), so most of the participants don't seem overly concerned. Notice that this is not how most the comments read, so it says something about people's willingness to click a button versus type in a text box. (Yes, one should always expect more extreme responses in a text box! ;) )

[ Reply to This | # ]