May 11, 2015 — The University of New Hampshire is routinely among the most unreliable of public polling entities, and the institution’s new release in partnership with WMUR-TV in Manchester is no exception to that characterization. The poll is attracting attention because it is the first one in months to project former Florida Gov. Jeb Bush as leading his fellow Republican competitors in any early voting state.
The survey, conducted during the very long sampling period of April 24 – May 3, interviewed only 293 “likely GOP primary voters.” The 10-day questioning period is seven days longer than the optimum timetable, and the sample size is only half as large as what one would typically see in a state the size of New Hampshire.
The pollsters will argue that because they are testing only likely Republican primary voters, the sample size will be smaller than a poll studying the entire electorate. While this is true, not even reaching 300 people taints the results with a very high error factor. By the pollsters’ own admission, the error rate in this study is greater than plus or minus 5.7 percent, which means the results could vary by as much as 10 points per individual.
Continue reading >