Skip to main content

View Diary: 93-Year Old Republican Shows How Bad Bush Has Been For Republican Party (224 comments)

Comment Preferences

  •  30% of what? (1+ / 0-)
    Recommended by:
    fhcec

    poll smoking is lethal in these times.

    phone surveys aren't "scientifically" comparable -- regardless of trend line inferences -- because of disparate survey designs.

    sample discrepancies -- regardless of normative, normalizing statistical methodologies -- rely on unverifiable estimates of "registered" and "likely" voter classes, reported by respondents, randomly selected.

    (not to mention the number discarded for whatever reasons. after all, the "reported" ratio of "registered" voters nationally is ~ 1/3 indie, dem, thug. CI of "weight" discrimination across classes is low.)

    survey topical hypotheses are uniformly characterized by complex phrasing of prompts (semantics), retarded by scalar measurement of responses (0-10), in order to admit vagaries in inferences of "weighted" or "adjusted" empirical evidence. that is the design is designed to dilute absolute frequency of responses.

    MSM validates survey credibility by association with biased organizations, some being blatant partisans, others being vested market propagandists.

    Diversity is the key to economic and political evolution.

    by MarketTrustee on Fri May 11, 2007 at 10:17:43 AM PDT

    [ Parent ]

    •  Phone surveys are getting more unreliable (0+ / 0-)

      by the minute, because they exclude anyone who doesn't have a home phone. Most people still do, but the number of people who use cellphones exclusively is growing.

      (by the way, they also tend to exclude people who work odd hours, travel a lot, or just plain don't like answering the phone during dinner - so it's not just the young and disaffected unlikely voters that the people who do the surveys would have us believe are excluded by the no-cellphone sampling)

Subscribe or Donate to support Daily Kos.

Click here for the mobile view of the site