[lbo-talk] Bush: 2 point bounce!

Tom Walker timework at telus.net
Tue Sep 6 16:13:17 PDT 2005



>/ Strongly Approve 23%
/>/ Somewhat Approve 23% />/ Somewhat Disapprove 14% />/ Strongly Disapprove 39% />/ />/ />/ To me "somewhat approve" is worse than "somewhat disapprove", but the way /they are
>/ ordered it seems like its a tiered system, of best to worst. I don't know
/how people taking
>/ polls intrepret it, and it worries me about the accuracy of these
/ WS:


>You are grasping for straws. The stupid fatheads that voted for him will
>support him even more, blaming everything on government bureaucrats. That
>is the Amrikun way - bad government good corporate and political bosses.

Hardly. The unambiguous 'strongly disapproves' are more than double the 'strongly approves'.One might also expect the milder views to more or less conform to the general shape of the stronger ones, so the result for the 'somewhats' is paradoxical. If you simply add the 'somewhats' together you get 37%, which is still less than the strongly disapproves. As someone who has designed, conducted and analyzed opinion surveys, I agree that the somewhat category is extremely ambiguous. Obviously someone who somewhat approves also somewhat disapproves, so the answer is very much subject to whim and misinterpretation of the question. Frankly, the whole exercise of polling is fraught with interpretative pitfalls. When a result is both ambiguous and paradoxical it doesn't make sense to put much stock in it. I would tend to split the difference of the somewhats and read the result as (roughly) 40% approve and 60% disapprove with a margin of error big enough to drive a truck through.

Sandwichman



More information about the lbo-talk mailing list