Herding is for horses. Not for pollsters doing horserace polls. Neither should the media herd the field in a political horserace via debates. Why? Here is my take on the…
10. Many candidates will fall within the error margin. Rankings become statistically meaningless.
9.1 Using decimal points makes statistically meaningless rankings even more meaningless.
7.9 – 8.1 Poll strew doesn’t necessarily taste very good. Some polls probe undecided voters to include “leaners,” others don’t. Some polls will be based on “likely” voters, others on registered voters. Poll results also vary when it comes to live or automated modes of data collection, proportion of cell phones vs. landlines, and weighting and analyzing data.
7. More problems. Some national polls take out “undecided” voters and recalculate based upon 100%. This wreaks havoc on averages.
5. Ok. I know I skipped number 6, but, then again, there’s no guarantee all polls will ask all candidates either.
5.1 (Couldn’t figure out where to place this item because it is not actually higher or lower than 5, statistically speaking). Some polls use push-button phones to record preferences. It’s tough to include 18 names when only numbers 1 thru 9 are usable.
4. “HELLLLOO” house effects.
3. Given that early caucus and primary states punch a candidate’s ticket to continue, why use a national average to determine debate participation?
2. Name recognition unduly influences results of early primary horserace polls. Lesser known candidates will now frontload their efforts to try to make the cutoff. Public polls altering campaign strategies? BAD!
1. And, finally, do you really want public polls this involved in a picking presidential nominee?
Try this on for size. How about a random drawing of half the field of announced candidates for the first hour of a debate and the second group for the second hour. More manageable. More equitable. And, doesn’t require a top 10 list!