A reality based independent journal of observation & analysis, serving the Flathead Valley & Montana since 2006. © James Conner.

21 September 2018 — 1537 mdt

Republican poll reports Tester and Rosendale tied at 44 percent

A poll conducted by Axis Research for the Republican National Senatorial Committee earlier this week found Sen. Jon Tester tied at 44 percent with eight percent undecided and four percent preferring Libertarian Rick Breckenridge. The live caller — landlines and cells — poll sampled 480 likely voters and has a sampling margin of error of 4.8 percent.

According to the Washington Examiner, a conservative publication that had access to a copy of the pollster’s report, Tester’s support is eroding:

Less than half of Montana voters approve of the job Tester is doing in the U.S. Senate. Tester’s job approval has dropped from 53 percent in July to 48 percent today.

Less than half of Montana voters have a favorable image of Tester (48 percent). Tester’s image has declined from a net +12 favorable to a net +3 favorable.

Voters report hearing more unfavorable news about Tester and his campaign with 42 percent seeing, reading, or hearing something that gave them a more unfavorable impression of the senator.

The Examiner also reports the poll found President Trump’s popularity in Montana is above water:

Trump’s job approval in the state is solid, ranging from 53 percent to 58 percent in the four recent polls Axis has produced for the NRSC — two in July, one in August, and this latest one.

Axis is not rated by FiveThirtyEight. Its memo on the Alabama Special Election that Roy Moore lost suggests Axis doesn’t always get it right. The Montana poll’s sampling methodology is sound. Axis did not reveal how the poll was weighted. The CBS News poll did:

The sample was weighted according to gender, age, race, education, and geography based on the American Community Survey, conducted by the U.S. Bureau of the Census, as well as 2016 Presidential vote. The weights range from 0.1 to 7.0, with a mean of 1 and a standard deviation of 1.2.

A third source of uncertainty in a poll is the choice of questions and how the questions are asked.

The RNSC did not release the full set of questions asked or the cross tabulations of the results. But the GOP did, reports J.M. Brown at The Western Word, send out an email pointing to the poll and asking donors for more money.

Tester v Rosendale is very close. Today, the Cook Political Report downgraded its rating from “likely Democratic” to “leans Democratic.” At Flathead Memo, we continue to consider the contest a toss-up.