Hi, this is Sarah Colling from the bottom of the Grand Canyon, where I just backpacked

nine miles and 5,000 vertical feet. This podcast was recorded at 12.40 pm on Monday, May 13th.

Things may have changed by the time you hear this. Keep up with all of NPR's political coverage

on npr.org, on the NPR1 app, and on your local public radio station. All right, here's the show.

I'm really impressed that she climbed to the bottom of the Grand Canyon. That's a huge issue.

It's a lot, yeah.

Congrats to you, Sarah. Yeah. Hey there. It's the NPR Politics podcast. I'm Susan Davis.

I cover Congress. I'm Danielle Kurtz. I've been political reporter and I'm

Dominican Montenegro political editor. Today we're going to talk all about polls. What makes a good

poll? How much should we be relying on what the polls are telling us and what are the polls telling

us right now about 2020? One reason I think it's a good idea to have this conversation as we go into

2020 is remember coming out of 2016 there was this sense that the polls got it all wrong and did

we overread the polls. So I think it's worth starting with just saying how are we smartly reading

the polls in 2020 and Danielle is someone who is both out on the trail and very much into the data.

What do you look for when we get news that a new poll is out? I'd say first things first, I look

at how the poll was conducted. So was it done by live callers people actually calling people up

on their phones? For example, as opposed to robo polling when you get a call on your phone and it's

like press one for so and so you get the robotic voice and all there is some dispute out there of

whether online polls are good or bad. I mean if they are scientifically conducted they aren't

necessarily a bad thing they can be good but live callers are sort of remain the gold-ish

standard as far as polling. A reasonably large sample size I mean a sort of short hand you can

look for with that is check the margin of error see how big the margin of error is you want a margin

...