Post

Dem Senate candidates have lost a little less than 0.5 pts of vote margin over last 2 weeks, while the generic ballot has trended up for them by about 1 point. Overall state of the race is unchanged: Dems have leads well below size of normal poll error fiftyplusone.news/forecast/sen...

I've noticed a number of right-leaning polls in the Fifty Plus One averages. Insider Advantage, Trafalgar, etc. Is that influencing the polling averages from what the more credible pollsters are showing?

There have been some suppression polls gop has tried to put out but showing Paxton or Roger’s up by 1 does not help gop I think it will make sure Dems come out

Thanks. I remain a polling skeptic. Some of these polls are such small samples. There's no way 400-1000 surveys can be statistically representative of a state. Then you throw them all into averages and weight them and predict who is likely to turnout. It's just hard for me to believe.

Sample size is the smaller problem. A clean random sample of 600 lands within about 4 points of the truth 95% of the time. The hard part is that the people who pick up aren't random, and no sample size fixes that. That's the error a forecast has to leave room for.

That makes sense to me. Thanks. I'm not a polling wonk but it all seems a little far-fetched. Do the candidates/parties do a better job of that for their internals?

Sometimes. Campaigns can pay for bigger samples, and they poll straight off the voter file, so they know who actually voted last time. The catch is that the internals you get to see are the ones a campaign chose to release, so I read those like a press release.

Thanks. So, these close Senate races are all going to come down to turnout.