Thread by @Peoples_Pundit

Yes, yes. I know what you're all going to say: "It's seven points. Democrats can't be that unstated in polls."

In fact, they can. CNN: Trump +6. Do I need to go back to 2010? That said, with the exception of 2010, the other cycles didn't present a consensus like this. So...

https://twitter.com/EmersonPolling/status/1051819983494402049

So, we can't dismiss this @EmersonPolling result, though many will try. At the very least, Heller is again on top. It's also more confirmation the confirmation hurt Democrats, considering both McSally and Heller saw their leads grow, the latter to 7 points.

Nevada's Second District looks shockingly competitive (should be likely D), though they're small samples. It's worth noting, however, because President @realDonaldTrump is headed to The Silver State on Saturday, and he brings a bounce with him. ppdnews.us/uEMmz6Z

Yes, we do factor in voter registration changes. I've seen this tweet 1 too many times. It's literally on the model and it changes constantly. Not only do we factor it, but we know about the changes before they are public. Voter files are updated constantly. Come on now.

That said, voter registration can trick you. And it tricks most each year. Party ID changes more frequently than registration. If registration ONLY mattered, then Republicans would lose NC & FL, Democrats would lose CO & NH. It's of course helpful, but it's not end all be all.

For example: Poor Fox had North Carolina light blue right up until Election Day in 2016. Why? @David_Bossie told them in an interview a day before. Older white Democrats were being counted as Hillary votes by the decision desk. Not picking on Fox. Almost everyone (wrongly) did.

Fox had their polling to justify that rating. Problem was cell collection without an online component. They called too many non-rural Democrats on their cellphones and it threw off the crossover margin. Region, Chris. Region. But that's the danger of cell-mode data collection.