I need to apologize. For multiple days, I’ve been aware of extra poll aggregating data. The Economist has a newish portal for this. Your lazy correspondent finally checked it out today. It’s amazing.
I’ve relied on the Real Clear Politics Average/list to see the mainstream surveys regularly referenced by the media and also used by the DNC to calculate debate qualifications. It’s easy. It’s straightforward. You can quickly compare a new survey to older ones from the same pollster.
For extra texture, there’s the FiveThirtyEight list. They also average the results, but more importantly, they cover all sorts of surveys that don’t make it to RCP. Sure, many of the pollsters are a bit more suspect. But that doesn’t mean they’re always wrong.
Sometimes they pick up outlier polls before similar results hit the mainstream. While the Monmouth result on Monday showing Joe Biden in third place at 19% was a shock to many, there were other signals in the FiveThirtyEight stack that make this a minority opinion as opposed to a single strange result.
You also get some state surveys that aren’t covered by RCP. For example, it’s interesting to me that Bernie Sanders leads a recent poll in Colorado. He did very well in very pro-Cannabis states in 2016, and if the trend continues in 2020, it’s a helpful edge.
So, having plenty to occupy me, I got a bit complacent and didn’t jump on the Economist numbers the minute I knew they existed. The most important feature is the aggregated and averaged crosstab data.
It was good that Monmouth told us Biden was sinking with each conceivable demographic group. Underlying data is super helpful. But they only included 298 voters. As meager as that sounds, it’s not an unusual or bad practice. But that means there were only 70 to 75 African American voters. Of which maybe 35 to 38 were over 50, and 17 to 20 female. You don’t really want to draw too many conclusions about what older black women think based on a maximum of 20 responses.
And that’s one of the larger groups. Try zeroing in on what under 30 year old Asian men think, or 30 to 44 year old Latinx voters with post graduate degrees who live in rural areas. Good luck.
All of this sort of data is available courtesy of the Economist. And they’re pulling together enough surveys that you can almost trust it. Unfortunately, they haven’t built the extreme example I did above. They do have breakdowns by gender, age, income, race, education, population density, and ideology. They also have some combinations, like age and race, gender and race, age and gender.
It goes far enough that you can decently infer the rest. And there’s semi-adequate proof now from a few things I’ve assumed previously. There are graphical representations so you can see where they fit compared to their competitors.
For example, Pete Buttigieg doesn’t have a giant problem with African American voters. It’s just not a strength. He’s 5th overall. There are only a few specific breakdowns where he’s any better than 5th. He tends to do better as education levels go up. He’s 6th among college educated black Democrats. The only extra candidate who jumps ahead of him is Cory Booker. Who like Mayor Pete does best with more educated voters. And is one of the most prominent black politicians in the country.
It’s not like he’s trailing Andrew Yang or Amy Klobuchar. I’m not saying he doesn’t need to improve here. But everyone not named Joseph Robinette Biden, Jr. does too.
I’ve already descended most of the way down the rabbit hole and will report back with various bits of analysis going forward. Anything listed on the Economist portal is enough sample size over a long enough time horizon to pay attention to.
We’re still going to talk about various individual surveys when they’re interesting enough. There’s just a really good new tool in the shed.