Dem
50
Ties
1
GOP
49
When we went live with our Senate map on Monday, we explained that we are picky about the pollsters we include in our database because some pollsters don't have a track record, and others are partisan houses that might be less-than-honest with what information they do, and do not, release.
What we did not anticipate—though this situation IS covered by our finickiness—is that a pollster might be making their data up out of thin air. However, that is apparently what was happening with Median Strategies, a pollster with a very legitimate-sounding name, and a very, very short track record (first poll: late July of this year). That outfit released a poll of the Los Angeles mayor's race that had everyone talking, as it had Karen Bass up 12 points on Nithya Raman, with most of the former voters of Spencer Pratt gravitating toward the more moderate Bass. The Los Angeles Times looked closely at the cross-tabs, however, and thought something was awry. So, the paper contacted the pollster with questions, and the pollster admitted the data was 100% fake.
According to the e-mail that the people (person?) behind Median Strategies sent to the Times, they faked their data to make a point. Here is their explanation:
Median Strategies was created as a short term social experiment to examine how purported polling information could enter and spread through the political information ecosystem without independent verification. We are not seeking publicity or attribution for the individuals involved and are declining interviews.
In the same e-mail, the firm (one-person operation?) announced that it's permanently shutting down.
We are not entirely sure we buy this story. It strikes us as possible that Median was trying to get started as a legit pollster, but didn't have the funding, and so cut some corners, and then got caught with their hands in the cookie jar. Alternatively, it's possible they did the work, but screwed up their model so badly, it produced wacky results (we'll get to another of their polls, already proven wrong, in a moment). It is unlikely that anyone involved with Median will ever find work in the polling business again, but if they are holding out hope, then "we were conducting a social experiment" is much better than "we screwed up bigly."
That said, let us proceed with the assumption that the story is true, and the point was to show how easy it is to get fake polling into circulation. If so, the "experiment" was a moderate success, at best. The Bass campaign ran with the phony mayoral poll, and sent out a tweet repeating the numbers (the tweet was later deleted). The California Post, which is the new, west-coast version of The New York Post, ran a pretty big story about it. And... that's about it. Not too many others fell for it, while the Times smelled a rat, and discovered the deception.
And that brings us to the other poll from Median that got a lot of attention, a poll of the Wisconsin gubernatorial primary that had Francesca Hong leading David Crowley by 20 points. As everyone knows by now, that was a wild miss; Crowley actually won by about half a point. And this new development might seem like an answer to the question of "How did the pollsters get Wisconsin so wrong?" (i.e., they were making stuff up), but it's not. Each of three polls conducted after Mandela Barnes dropped out of the race also had Hong up by a similar number. And the other two pollsters, Slate Navigate (had Hong up 44%-22%) and PPP (had Hong up 42%-24%) are legit pollsters with actual track records. If it has been only one, unknown pollster who had the race as a blowout, then every responsible outlet (including us) would either disregard the poll or would do what we've done a million times, writing something like "If this poll is correct, then that's a big deal, but you never want to put too much faith in a single poll, particularly from a little-known pollster. So, take it with a fistful of salt or two."
In short, we don't really think Median Strategies fooled very many people, but we suppose they did give all the psephologists an opportunity to remind everyone that, thanks to the Internet, modern polling is kind of the Wild West, and there are a lot of unknown pollsters out there. So, as Obi-Wan Kenobi observed, we must be cautious.
Meanwhile, on a very related note, it's time for a pop quiz. Which of these are legitimate news outlets?
We will wait a moment while you lock in your selections. It is... instructive that all of them report on states with critical U.S. Senate races this year.
The answer, as many readers will have guessed, is that none of them are legitimate. They were all created by One Nation, a PAC that is the less savory partner of the more above-board Republican National Senate Committee. We have talked many times about push polling—polls meant to promote political messages, and not to collect actual data. Well, these outlets are all push journalism. Their purpose is not to cover the news, but to promote the GOP's messaging in a manner that appears, at a glance, to be fair and balanced and nonpartisan.
We aren't sure how many people are actually taken in by these outlets. First of all, because they are fake, they are not ranked highly in Google searches—and, in fact, Google's AI gives a warning that they are fake if you search for them. Second, because even if you manage to find, say, the Peach State Post, it's more than a little suspicious that they only produce a couple of stories a week, mostly about what Donald Trump is up to, with no bylines.
Still, as with the fake pollster, there really are completely phony news outlets out there (besides Fox, we mean). By and large, with local news, we prefer to link to local sources as opposed to national ones, since we assume the local sources know local subjects better. It's fairly easy when dealing with the major and semi-major newspapers. It's also pretty easy with TV stations that also post their stories in written form. However, we sometimes have to look very carefully at smaller outlets, or those that are "cleverly" named. For example, Yellowhammer News is a legit site for Alabama news; the Ames Voice is a legit site for suburban/rural Iowa.
"Trust... but verify," the old saying goes. We would say that the point of this item is that, these days, you should probably skip "trust" and jump right to "verify." (Z)