Wednesday, August 9, 2023

When it comes to politics, there's almost no real data journalism, just narratives and drunkards' lamp-posts*

* "They use statistics like a drunkard uses a lamp-post, not for illumination but for support."   **

Actual data journalism, where the story comes from the numbers and the journalist tries to explain what they're saying, is the exception in political reporting. What we generally get is narrative journalism where the data serves strictly as a drunkard's lamp-posts. Two recent examples, the first from the NYT.

[This picks up directly from Friday's post, so if you missed that you might want to go back and get caught up before proceeding.]

A year ago, the political press was all abuzz about how badly Trump was doing and how great DeSantis was doing. Their primary evidence was a poll that had the former president leading the governor 49 -- 25. Recently, the same crowd was abuzz over a new poll which had Trump over DeSantis, 54 -- 17.

 In absolute terms, this is not that much of a change. Trump improved by five points; DeSantis dropped by eight, but somehow when journalists, often the same journalists, analyzed these two polls, they managed to come to almost completely opposite conclusions. 49 meant that Trump was imploding, he was effectively out of the game, it was all over but the shouting. 54 meant that Trump was triumphant, he had the nomination wrapped up, it was again all over but the shouting.

 As we pointed out at the time, that 49-25 didn't actually look very good for DeSantis. By the same token, the 54-17 doesn't look as bad for the governor as everyone seems to think. After months of pretty much being the only rival that Trump has been focusing on, despite blistering attacks, DeSantis remains securely in the beauty pageant runner-up slot. If for any reason the pageant winner is unable to fulfill her duties as Miss GOP, the first runner up will step into the job. Since there is a real chance that for reasons of health or incarceration, Trump might drop out or be forced out, this is not a bad place for Ron to be. 

When you add in a rally round the flag effect that may or may not be fleeting, the polls haven't actually shifted that much but they are being used to serve a completely different narrative. Keep in mind, in 2021 and 2022, at least after January 6th, the general consensus was that Trump was effectively dead in the water. Even if you had put aside his legal troubles, he was now branded a loser and conventional wisdom had it that his loyal cult would now start to abandon him. Around the same time, the Rupert Murdoch empire with considerable help from the establishment media of the New York Times, Politico, Slate, etc decided that Ron DeSantis was the perfect choice to replace Trump as a far-right but still conventional candidate. There were a few voices in the wilderness such as Michael Hiltzik of the LA Times pointing out the flaws in this line of reasoning, but they were almost entirely ignored. Now things are different.

Print the narrative. When the narrative changes, print the new narrative.

Perhaps the mother of all modern political narratives is Dems in disarray. We could find countless examples of bad data analysis supporting this one from the pages of the New York Times, but it is hard to beat this recent example from Politico. Emphasis added.

By Jessica Piper [Sam Stein contributed to this report.]

One of the best online fundraising days for Democrats this year was the day of Joe Biden’s campaign launch — but even that day’s haul was meager compared to his campaign kickoff four years ago.

That’s among the findings of an analysis of fundraising for the first half of the year through ActBlue, the party’s primary donation processor. Small-dollar giving at the federal level totaled $312 million in the first half of 2023 — a drop-off of more than $30 million compared to this point in the 2020 cycle. The platform also had 32 percent fewer donors in the second quarter this year compared to four years prior, although its total fundraising increased slightly due to several factors, including more recurring donors and greater giving to non-federal groups.

How do you go from a slight increase to a a full-bore the-sky-is-falling story about a funding catastrophe? You employ those two old standbys of data cooking, arbitrarily limited range and comparing to a spike. In this case, the tip off is "federal." If you look at the big picture, fundraising is going pretty well. If you only look at the federal level, there is a drop, but only because the situation four years ago was completely different.

Think back to 2019. We were in the middle of perhaps the most wide open democratic primary in living memory. There were at that point over 20 reasonably serious competitors, all of whom were aggressively fundraising. 

This pretty much explains the "massive" hit. The money that was going to the primary has now shifted to places like Wisconsin where it can do the most good. If the journalists were following the data, the narrative would be "dems in array."

 Print the narrative.

 

** The original analogy goes all the way back to A. E. Housman who was talking about manuscripts rather than statistics. That isn't relevant to this post but is too cool to leave out.

Posted by at

No comments:

Post a Comment

Subscribe to: Post Comments (Atom)