Memphis pollster Berje Yacoubian has released results of a poll his firm conducted last Tuesday and Wednesday that found the county mayoral race will be, as expected, very close. The topline results show Shelby County sheriff Mark Luttrell with a slim lead over interim mayor Joe Ford, 45 percent to 42 percent, down slightly from an earlier Yacoubian poll that showed Luttrell up 46 to 41. Undecideds remained the same -- 13 percent. The poll had 457 respondents, 147 of whom had already voted and 310 who were "definitely" or "probably" going to vote. It had a margin of error of plus-minus 5 percent at the 95 percent level.
Supporters of either candidate could find things to emphasize in the poll, but the Joe Ford camp is cheering the apparent uptick in support from white voters for Ford. Yacoubian said that white support for Ford rose 18 percent and Luttrell's white support dropped 21 percent. That's somewhat misleading because Ford's numbers among white voters were so low in the first place, but it is giving Ford's supporters reason to believe.
Citing various ingredients, Yacoubian concludes: "If Democratic voter turnout reaches 60 percent of the total August 5th (compared to 56% in the early vote), Ford would win by 51% to 49%."
Given the motivation Republican voters have shown, that is one very big "IF."
Why hasn't Yacoubian defended his polling, in the face of Herenton's challenge.
Your analysis was fine, Zack, as far as it went, but Berje is the expert, and it seems to me he should tell us why he is right and Herenton is wrong.
Same, by the way, goes for Bakke and Ethridge.
Please give me a physical address so I can mail you a text on statistics, Zack. If the poll's margin of error is 5%, then talk of "leads" that are less than five percent is meaningless. Potentially even dishonest.
If a poll says X is at 50%, with a margin of error of 5%, then the actual number is between 45 and 55%. Polling is an uncertain science, hence the "fudge factor" of a margin of error.
Start here: http://en.wikipedia.org/wiki/Margin_of_error