I get Ruthless With David Leonard

(Dan here…lifted from Robert’s Stochastic Thoughts)

by Robert Waldmann

I get Ruthless With David Leonard

David Leonard picks cherries in a generally good op-ed. I agree entirely with his general conclusion that Democrats should run a populist campaign (no triangulation — he should have noted that Clinton ran on raising taxes on the rich and cutting taxes on the middle class in 1992 — he was a populist before he was a triangulator). He also says don’t talk about decriminalizing border crossing or eliminating private health insurance. I agree entirely. He relies on a Pew poll on issues. It is an interesting poll by a good pollster.

However, I think there should be a rule that any commentary on polls should consider all available still relevant polls. The norm of non data journalists writing about data is still to comment on one poll. This is nonsense. It is like election night coverage based on an interview with one voter. There is, I think, no excuse for looking at data other than averages of polls. I think fivethirtyeight.com can improve on the simple average, but that’s not my current assertion. I am asserting that any commentatory must justify (to an editor not the readers) every decision to not consider every poll which is not considered.

I was triggered by this passage justified by three picked cherries.

Yet Democrats are frittering away their advantage — and damaging their image. Last fall, most Americans had a favorable view of the Democratic Party, according to the Pew Research Center. That makes sense, because Democrats ran a populist campaign in the 2018 midterms, focused on pocketbook issues that dominate many people’s lives, like wages and medical costs.This year, the polling has flipped. Most Americans now have an unfavorable view of the party, no better than their view of the Republican Party. Likewise, slightly more voters say the “ideas being offered by the Democratic candidates” would hurt the country than say would help, according to the NPR poll.

 

 

I was surprised to learn how hard it was to find averages of party favorable ratings from 2019 (hard enough that I gave up anyway). However, I am confident that there hasn’t recently been a dramatic change, because the generic ballot shows a Democratic lead about the same as in 2018. The fivethirtyeight.com graph is based on dozens of polls. That’s the way to do it. That’s the only way which editors should allow. In fact, the highly anomalous party favorables in the Pew poll used by Leonard should have caused him to reconsider the issues polling. A crude but not pointless calculation would be to add 6.5% to medicare for all pro minus M4a contra. Saying an anomalous number on party favorability adds to the evidence from issue polling is to say that all polls but the latest Pew poll (and a briefly mentioned NPR poll) are irrelevant.

I stress again that I agree 100% exactly with Leonard’s conclusions and advice.

my comment cut and pasted 10:20 (I didn’t guess it would take as long as 20 minutes to refute his claim

I agree with your conclusion. I’d add (as your colleague Frank Bruni does today) that it is unwise to propose providing insurance to undocumented aliens (combining Medicare for really all and more than just a path to something good years from now for undocumented aliens). I happen to find myself in the minority which supports all three proposals. I also know that they are not going to happen — fuhggedaboudit, and that candidates admitting that they support those proposals helps Trump.

However, I consider the method of your argument to be unacceptable. You discuss two (2) recent polls (one very briefly) contrasting them with one (1) poll from 2018. This will not do, even though Pew is an excellent pollster. It passed as legitimate commentary way back in the 20th century, but people should now stick to the Silver standard.

I start the clock at 10:00 AM Rome time (4:00 AM in New york).

Pollingreport latest tweet is a link to this op-ed. Congratuationsl.

OK this is hard. I can’t find an average of polls of Democratic party favorability past 2018 (what is wrong with the web). I was wrong. What I find is generic ballot polls going back to January 2019 with a graphed average going back to April. I see a stable Democratic lead of around 6% if anything growing slightly with latest 6.5%. In 2018 the Democrats won the popular vote by 8.6% (not strictly comparable) https://projects.fivethirtyeight.com/congress-generic-ballot-polls/?ex_cid=rrpromo

Your cherry picked number is highly misleading

Comments (8) | |