Lies, Damn Lies, and Susie Neilson
In January, 2023, The San Francisco Chronicle famously declared war on fact-based journalism:
“The consensus among younger journalists is that we got it all wrong,” Emilio Garcia-Ruiz, editor in chief of the San Francisco Chronicle, told us. “Objectivity has got to go.”
At the same time, the new Chron is insistent we focus on data. We may feel like crime is rising in SF, for example, but the statistics show crime is down, and numbers don’t lie.
Except that numbers can lie, especially when journalists are freed from the tethers of unbiased reporting. Let’s look at a few Chronicle examples below.
Note: These articles are written, or co-written, by the Chron’s “data reporter”, Susie Neilson, who wears her hard-left biases on her sleeve. Susie’s lapped up just about every kooky progressive idea imaginable:
This includes, of course, a misguided hatred of law enforcement. Her Twitter feed is a stream of consciousness about just how awful our first responders are.
Given these biases, can you trust Susie’s reporting? Let’s look at a few examples.
Torturing the numbers
On Jan 23, 2023, disgraced former DA Chesa Boudin gleefully retweeted a piece by Susie purporting to show a conspiracy by SFPD to “needlessly” pull over black drivers. This accusation was based on transcripts of the words used in police stops. “We mined the texts”, she wrote. And data mining by Susie can yield only one result.
This article might have been quickly forgotten, except that CPA, author, and lifelong city resident “Lou B” smelled a rat. As Susie listed the source of her data as “SFPD”, Lou requested the same data set, so he could do some data mining of his own. Problem is, it didn’t seem to exist.
SFPD provided Lou with the same data they provided the Chronicle, covering the same period in Susie’s article. But the words didn’t match. For example:
Neilson/Phillips claimed, “In one field the data, labeled ‘basis for search narrative’ we found officers frequently employed the words ‘smell,’ ‘marijuana’ and other drug-related terms.” Actually, in the spreadsheet SFPD provided to the Chronicle, the word “smell” never appears.
Lou did his best obtain the data Susie claimed she used, but SFPD finally gave up:
“after speaking with several different people/units that could have possibly handled the data mentioned by the Chronicle, they were still unsuccessful finding out the source of the (Neilson/Phillips’) data.”
So Lou worked with data he had. And it was enough to find Susie had cherry picked individual words to prove her point, that, as the saying goes, All Cops Are Bastards.
Lou went on:
But like how Neilson/Phillips hid their database to eliminate context, they also deceitfully withheld the words that must have been used disproportionately against whites.
Ominously, he continued:
The size of this search creates the suspicion they had help from an advocacy group that had access to this government information which is apparently external to SFPD.
Is Susie a mouthpiece for radical decareral groups determined to destabilize our city? We have only suspicions at this point.
Data, shmata
Once you toss out objectivity, reporting accurate data becomes unimportant. You write the headline, then mold the numbers to fit it, not the other way around.
Case in point: This July 25, 2022 article by Susie Nelson and Roland Li, titled Richer people left San Francisco in the pandemic. And they took billions of dollars with them. The lede:
Between 2019 and 2020, the number of people listed on a tax return in San Francisco fell by 39,202, a drop of 4.5%, according to data from the Internal Revenue Service. Residents who left made an average of about $138,000 per year in 2019, up 67% from the prior year, when departing residents had an average annual income of around $82,000.
The problem is, virtually none of this is true. Here’s the actual IRS data (source) they used to reach this conclusion (see Notes below for data definitions):
The authors got the 2020 number right, but their 2019 figure is way off — $130,331 vs $82,000. This invalidates the whole thesis of the article, that “more wealthy, white-collar workers, many of whom could work remotely, left the city” between 2019 and 2020.
But the errors don’t stop there. This bit, for example:
Between 2019 and 2020, the number of people listed on a tax return in San Francisco fell by 39,202, a drop of 4.5%
Is disastrously wrong. How?
Well, first, they arrived at 39,202 by subtracting the number of people on returns who left SF in 2019 from those who moved here: 77,318 - 38,116 = 39,202.
Problem is, that doesn’t give you the actual change in people listed on tax returns, as among other issues, the non-migrant population changes from year to year.
The Chron then takes this erroneous number and arrives at that 4.5% drop by using US Census data showing SF had 870k people in 2020. 39,202 / 870,000 = 4.5%.
But you can’t mix IRS filing data with census population numbers, since not everyone appears on a tax return. SF did have a population of 870,000, but only 733,556 of those were known to the IRS in 2020.
The article continues:
The total income in 2019 of people who had left the city by the time they filed their 2019 returns was about $10.6 billion, compared with $3.8 billion for those that came to the city — a net loss of almost $6.9 billion.
Nope. Wrong Year. This is true for people filing their returns in 2021 for the 2020 filing year, as you can see from the data above, not for 2019.
There was a time when Chron editors checked data before it was printed for accuracy. But accuracy no longer seems to matter.
(What does this IRS data actually tell us about the past, present, and future of SF? Quite a bit, actually, if you know how to read it, and that’s the subject of my next Substack post.)
There are many similar cases. For example, on Feb 28. 2021, Roland and Susie declared people really weren’t leaving California, as reported elsewhere. The problem? They used USPS change of address data, which doesn’t accurately track moves. A few months later, they admitted their mistake.
In another example, Susie reported crime figures you could see at a glance were far too low. The problem? The numbers she’d posted were per 100,000, not city-wide. When the Twitterverse alerted the Chron, it silently fixed the issue, without posting a note about the correction.
Conclusion
As Jeff Bezos said famously:
“when the anecdotes and the data disagree, the anecdotes are usually right. There is something wrong with the way that you are measuring it”
When the non-objective media attempts to devalue your lived experience in favor of “crunching the numbers”, remember that numbers can be crunched all sorts of ways, and the final product may or may not reflect reality.
Notes
IRS Migration Data Definitions:
Year (2019 or 2020) - the filing tax year. But migration data is taken based on current address when taxes are filed, usually early in the succeeding year.
“Returns” are the number of form 1040s filed by individuals or families.
“Individuals” are the total exemptions claimed on those 1040. If you’re a family of 5, you’ll take 5 exemptions. But the IRS isn’t aware of people not listed on form 1040, and not everyone has to file, so they can’t provide general population data.
Non-migrants are people who didn’t arrive or leave SF in a given year (again based on the address when they filed their taxes, early in the following year).
IN Migration are people who had SF addresses in the current, but not previous year. OUT Migration is just the opposite.
Non-migrants, plus IN Migration, plus OUT Migration gives the total returns/individuals filed for SF in a given year.