MN: Harper Polling (R) - Biden +3

[TiltsAreUnderrated]

August 30 to September 1
501 LV
MOE: +/- 4.38%

Biden - 48% (-2)
Trump - 45% (+3)
Undecided - 4% (-4)

Changes from a May poll which doesn't appear to have been released (not on FiveThirtyEight at least) but it's referenced in the released.

https://www.scribd.com/document/474637138/MN-SEN-General-Election-Key-Findings-Memo-Harper-Polling

Then there probably wasn’t a May poll.

There was, but the sampling period and sample size were not in the release. 538 only adds polls for which the sampling dates are supplied.

[TiltsAreUnderrated]

August 30 to September 1
501 LV
MOE: +/- 4.38%

Biden - 48% (-2)
Trump - 45% (+3)
Undecided - 4% (-4)

Changes from a May poll which doesn't appear to have been released (not on FiveThirtyEight at least) but it's referenced in the released.

https://www.scribd.com/document/474637138/MN-SEN-General-Election-Key-Findings-Memo-Harper-Polling

Then there probably wasn’t a May poll.

There was, but the sampling period and sample size were not in the release. 538 only adds polls for which the sampling dates are supplied.

No, I just don’t buy that they polled this race in May.

So you think they're actually faking their numbers, or that they polled slightly earlier but had those numbers ready for a private release in May? The former prospect is quite a bold claim to entertain.

[TiltsAreUnderrated]

The Trump campaign has always been obsessed with expanding the electoral map even though it's practically impossible for him to do that. Remember Parscale's New Mexico fixation? MN is just the latest iteration of that delusion. With Biden investing here, unlike Hillary, the state is a comfortable Likely Dem.

Trump spent $0 in the final few months in MN and the state party was laughably disorganised, while the DFL was one of the best outfits in the country no matter what Team HRC was doing. In 2016, the Republicans were well below their ceiling. To an extent, they are pushing at what was once an open door by pitching for MN and it is a smart move, especially if MI snaps back and/or too many sunbelt voters are unreachable for Trump.

It's not so much "expanding the map" as moving it.

[TiltsAreUnderrated]

August 30 to September 1
501 LV
MOE: +/- 4.38%

Biden - 48% (-2)
Trump - 45% (+3)
Undecided - 4% (-4)

Changes from a May poll which doesn't appear to have been released (not on FiveThirtyEight at least) but it's referenced in the released.

https://www.scribd.com/document/474637138/MN-SEN-General-Election-Key-Findings-Memo-Harper-Polling

Then there probably wasn’t a May poll.

There was, but the sampling period and sample size were not in the release. 538 only adds polls for which the sampling dates are supplied.

No, I just don’t buy that they polled this race in May.

So you think they're actually faking their numbers, or that they polled slightly earlier but had those numbers ready for a private release in May? The former prospect is quite a bold claim to entertain.

Actually, the former is what I meant, but the latter is possible given they polled MN-02 as well. I just don’t see why it would get better for Trump.

Internals are often overoptimistic not because the pollsters themselves are unreliable but because they're selectively releasing. If Lewis' outfit has regularly polled, natural variations thanks to MoE will mean they can eventually pick two polls and suggest the race has shifted in his favour since a previous point in time.

As it stands, though, I think there is reason to believe it could get better for Trump from May to this particular part of August due to a convention bump.

[TiltsAreUnderrated]

August 30 to September 1
501 LV
MOE: +/- 4.38%

Biden - 48% (-2)
Trump - 45% (+3)
Undecided - 4% (-4)

Changes from a May poll which doesn't appear to have been released (not on FiveThirtyEight at least) but it's referenced in the released.

https://www.scribd.com/document/474637138/MN-SEN-General-Election-Key-Findings-Memo-Harper-Polling

Then there probably wasn’t a May poll.

There was, but the sampling period and sample size were not in the release. 538 only adds polls for which the sampling dates are supplied.

No, I just don’t buy that they polled this race in May.

So you think they're actually faking their numbers, or that they polled slightly earlier but had those numbers ready for a private release in May? The former prospect is quite a bold claim to entertain.

Actually, the former is what I meant, but the latter is possible given they polled MN-02 as well. I just don’t see why it would get better for Trump.

538 has presumably got in touch and are now listing the May poll as a May 26-28 survey of 510 likely voters, presumably with a similar margin of error.