Scott RASMUSSEN: McCaskill +2, Scott +3, Braun +3, McSally +7,

[DataGuy]

People complain about how little polling there is this year, and then when they get a poll they find every which way to discredit it.

The proper response to every poll is to simply put it in the average, even if it doesn't "look right." It's a good thing when pollsters are honest about their numbers instead of twisting them to line up with the groupthink (known as "herding"). Gut feeling punditry is wrong time and again, yet people never learn.

[DataGuy]

Why does 538 have them entered like this? I'm confused.

They're getting the data from the 7-day rolling average on this page.

[DataGuy]

People complain about how little polling there is this year, and then when they get a poll they find every which way to discredit it.

The proper response to every poll is to simply put it in the average, even if it doesn't "look right." It's a good thing when pollsters are honest about their numbers instead of twisting them to line up with the groupthink (known as "herding"). Gut feeling punditry is wrong time and again, yet people never learn.

The problem is that such a scenario is all but impossible.  There's no way McCaskill is winning by 2-3% while Donnelly loses by 2-3%.  Missouri is much more conservative than Indiana and is less likely to vote for Dems downballot.

That hypothesis sounds sensible, but there is no limit to the strange and seemingly illogical things that can happen in elections. We may have our ideas of what should be happening, except (as one article put it after an upset some time ago) for the inconvenience of living, breathing individuals at the voting booths who may occasionally have different ideas.