Five reasons to doubt the new Loras College Iowa poll

A new Loras College poll shows Hillary Clinton enjoying a double-digit lead over Donald Trump: 48.2 percent to 33.8 percent with no other candidates named, and 44.0 percent to 30.7 percent in a field including Libertarian nominee Gary Johnson (6.0 percent) and Green Party nominee Jill Stein (2.2 percent).

The same poll of 600 Iowa registered voters finds Senator Chuck Grassley barely ahead of Democratic challenger Patty Judge, 45.8 percent to 44.5 percent.

Unfortunately for optimistic Democrats, this poll appears to be an outlier.

You can find the Loras College polling memo here and the full questionnaire here. There’s no obvious problem with the methodology.

Note on Methodology: The Loras College Poll surveyed 600 likely Iowa 2016 presidential election voters by live operator. The survey was conducted June 24-28, 2016. Margin of error for full sample results is +/- 4%. Margins of error for subgroups are higher. All results calculated at a 95% confidence interval. The partisan composition of the sample was 35 percent Republicans, 33 percent Democrats, and the remaining 32 percent No Party/Other. Poll results for the complete sample have a margin of error of +/- 4%. Percentages may not add to 100% due to rounding.

• Survey conducted with a random sample of registered voters (phone numbers drawn from official Iowa Secretary of State voter files)
• The statewide sample was balanced for standard demographic variables such as age and gender, with party composition to approximate 2012 presidential electorate in Iowa.
• Survey included both landlines and cell phones (55 percent and 45 percent, respectively).
• Screen for likely voter is report of “definitely,” “very,” or “somewhat likely” to vote in presidential election in November. Those indicated only “somewhat likely” to vote were required to pass through an additional screening question related to their interest in politics.
• The survey was conducted using live operator interviews through a contracted professional call center.
• Script development and methodology used for the survey received input from Republican campaign consultant Brian Dumas and Democratic campaign consultant Dave Heller.

The age and partisan composition of the Loras poll is not far off from the statewide statistical report from the 2012 general election.

However, even with the best efforts to find a random sample, approximately one poll in twenty will be off by more than the margin of error, just by chance. This appears to be one of those times.

1) No presidential candidate has carried Iowa by double digits since Bill Clinton defeated Bob Dole by a little more than 10 points in 1996.

2) Every election forecaster sees Iowa as competitive. If Hillary Clinton’s campaign believed she was comfortably ahead of Trump in Iowa, there would be no large early television buy here. Public Policy Polling’s most recent Iowa surveys showed Clinton with a much smaller lead.

3) Grassley has never had such a narrow lead over any challenger in any poll. The controversy over the U.S. Supreme Court vacancy appears to have cut into his popularity this year, but I am skeptical Judge is within the margin of error. Public Policy Polling conducted two Iowa polls this month, both commissioned by progressive advocacy groups, and found Grassley 7 points ahead of Judge.

4) The new poll shows President Barack Obama’s job approval rating at 51 percent, with 45 percent disapproving. I haven’t seen approval ratings that high for Obama in Iowa for a long time. While he may have become more popular this year, it seems more likely this survey oversampled voters who favor Democrats.

5) Loras College hasn’t been doing polls for very long, and their track record is not great. Their last poll of the 2014 U.S. Senate race showed Bruce Braley ahead of Joni Ernst, but he lost statewide by more than 8 points. Their last Democratic survey before the Iowa caucuses overstated Clinton’s lead by more than any other pollster.

Any comments about the presidential or U.S. Senate race are welcome in this thread.

You need to signin or signup to post a comment.