Numbers manipulated to create a narrative

[Yehuda57]

Before every season, fantasy football analyst Mathew Berry writes a column called "100 facts" where he makes this exact point. He shows how pundits cherry-pick and manipulate stats to create narratives about the players they like and don't like. I don't read any other fantasy football columns, but I like this one because it applies to every field.

[Yehuda57]

https://twitter.com/cholentface/status/1593277225473802243

[S209]

Perhaps a thread for examples of number manipulation is in order. I’m sure there are hundreds of examples but I just happened to come across this one:

https://www.nbcnews.com/politics/politics-news/evictions-are-piling-us-covid-era-protections-end-rents-climb-rcna54798

“Eviction filings have been on the rise and were above their historical averages in half of the 1,059 counties tracked by Legal Services Corp., a federally-funded legal aid group, during either August or September.”

Sounds like an eviction crisis. Not. If numbers are above average in half of the counties tracked, numbers must necessarily be BELOW AVERAGE in the other half. Even in the half of the counties that were above average, it was only when looking at EITHER AUGUST OR SEPTEMBER but not both. So half of all counties weren’t even above average in either August or September and the other half had above average numbers in either August or September individually but not all were above average for both months combined.  In other words, numbers for August and September were actually BELOW average in a majority of all of the counties! What we have are numbers showing historically low eviction rates being manipulated to attempt to portray the opposite narrative.

How do you know they are below average in the remaining counties, and not just about at the average (as they should be)? If the authors were striving for accuracy then that should indeed be the case.

If your point is that numbers *could* be manipulated, sure. You should read some of Taleb’s books, especially “Fooled By Randomness”. But if your point is that this *must* be manipulated, or is even likely, then I disagree.

[Moshe Green]

www.pravada.co.ru/race-with-american-capitalist-dogs

"Mother Russia sent comrade today in race against dirty American dogs. Soviet Russia is proud big victory as Boris Chukechusky came in 2nd in close race. American capitalist pigs finish second to last."

[jye]

How do you know they are below average in the remaining counties, and not just about at the average (as they should be)? If the authors were striving for accuracy then that should indeed be the case.

If your point is that numbers *could* be manipulated, sure. You should read some of Taleb’s books, especially “Fooled By Randomness”. But if your point is that this *must* be manipulated, or is even likely, then I disagree.

Because LSC has the data on their website. You can scroll state by state or county by county and view the bar charts.

https://www.lsctracker.org/alaska/state-wide

 Very few are at or near the average. Many are way below like Delaware:




Some are way above, like Arkansas




A very small number are right at the average but are the exception when looking at both August and September, like Indiana:



Many others are at the average one of the two months but under or over the average on the other Month. As per the reporters methodology, they were counted as above average if the other month was above such as Pima county Arizona:



Therefore they should be counted as below in the reverse scenario such as New Hampshire:





Clearly the data was manipulated and cherry picked to create “news”.

[yuneeq]

If your point is that numbers *could* be manipulated, sure. You should read some of Taleb’s books, especially “Fooled By Randomness”. But if your point is that this *must* be manipulated, or is even likely, then I disagree.

While I agree that you can’t assume that the rest were below average, regardless, the numbers are manipulated because the article is intentionally presenting a conclusion based on incomplete data.

It’s more egregious here because the full dataset is available, but not presented in the article, committing a stacking the deck fallacy.

[S209]

The LSC has the data on their website. You can scroll state by state or county by county and view the bar charts. Very few are at or near the average. Many are way below like Delaware:

Some are way above, like Arkansas

Several are right at the average bet are the exception especially when looking at both August and September, like Pima county Arizona:

I didn’t fact check or look at the article at all, I’m just nitpicking what you said in the OP. You said the other counties *must* be below their averages. Obviously that isn’t a must and should not be true unless the author was specifically looking to manipulate the numbers.

While I agree that you can’t assume that the rest were below average, regardless, the numbers are manipulated because the article is intentionally presenting a conclusion based on incomplete data. It’s more egregious here because the full dataset is available, but not presented in the article, committing a stacking the deck fallacy.

If that’s true it’s egregious for sure.

[yuneeq]

If that’s true it’s egregious for sure.

The quote from the article is very telling

“Eviction filings have been on the rise and were above their historical averages in half of the 1,059 counties tracked by Legal Services Corp., a federally-funded legal aid group, during either August or September.”

Note what they are and aren’t saying. They aren’t saying that they tracked 500 counties and this is what they found. Rather, they tracked and have data for 1000 counties, but are only presenting a conclusion based on half of the counties they tracked.

The remainder of the article does not provide the rest of the data.

[jye]

Just out of curiosity I took a random sampling from the drop-down menu of one county from each state they had data for, and averaged out the two months to see what the data would show without spin. Obviously one would have to go through each county to get accurate data, but this should provide a pretty decent indication of what the numbers really show. I came up with 12 counties above the average historical number of evictions, 13 below average, and seven right around the average. Not much of a story.

[S209]

The quote from the article is very telling

Note what they are and aren’t saying. They aren’t saying that they tracked 500 counties and this is what they found. Rather, they tracked and have data for 1000 counties, but are only presenting a conclusion based on half of the counties they tracked.

The remainder of the article does not provide the rest of the data.

Not at all. The assumption would be that they tracked all counties and found that half of them were above average (with the implication that the other half were average). No reason to assume what you’re saying.

If it is as you’re saying that’s an egregious lie by omission but this is NBC so we’re used to yellow journalism from them.

Just out of curiosity I took a random sampling from the drop-down menu of one county from each state they had data for, and averaged out the two months to see what the data would show without spin. Obviously one would have to go through each county to get accurate data, but this should provide a pretty decent indication of what the numbers really show. I came up with 12 counties above the average historical number of evictions, 13 below average, and seven right around the average. Not much of a story.

That may very well be so but you couldn’t just assume that from the numbers they presented.

[yuneeq]

Not at all. The assumption would be that they tracked all counties and found that half of them were above average (with the implication that the other half were average). No reason to assume what you’re saying.

If it is as you’re saying that’s an egregious lie by omission but this is NBC so we’re used to yellow journalism from them.

The implication (and literally the purpose of the article) is that evictions are rising. Presenting half the available data to prove that point is an egregious lie.

[David61]

[mevinyavin]

As commented by the holy XKCD:

[S209]

The implication (and literally the purpose of the article) is that evictions are rising. Presenting half the available data to prove that point is an egregious lie.

Right, if they did that it’s a lie. Are you agreeing or disagreeing with me?

If it is as you’re saying that’s an egregious lie by omission

[Randomex]

From https://aish.com/dave-chappelle-and-the-space-jews/ :

Maybe it is because Jews are rich? But according to two major national studies,* between 16 and 20 percent of Jewish households earn less than $30,000 a year, and around seven percent of Jewish households earn less than $15,000. Compare that to the national poverty rate reported in the 2020 census, listing 11.4 % of the US population suffering from poverty.

But what does the top end look like and how does it compare to others?

[Moshe Green]

https://www.zerohedge.com/political/four-mass-shooting-truths-shut-down-your-liberal-family-members-thanksgiving-dinner

[Euclid]

https://www.zerohedge.com/political/four-mass-shooting-truths-shut-down-your-liberal-family-members-thanksgiving-dinner

Not sure what #4 is proving; that guns kill way more people than the media reports?


TBH this article is more like "Narrative manipulated to create a narrative" - it just harps on a few minor talking points

[sguitarist18]

Not sure what #4 is proving; that guns kill way more people than the media reports?


TBH this article is more like "Narrative manipulated to create a narrative" - it just harps on a few minor talking points

I think the point is that there's  perception (likely encouraged by people/media of certain political persuasions) that mass shootings comprise a high percentage of gun-related deaths. And that's relevant because there's also a (very strong, IMHO) perception that perpetrators of mass shootings are almost exclusively white males.

Note: I think the article is written to be deliberately obnoxious, but also makes some interesting points.

[YitzyS]

The Asbury Park Press is frequently guilty of the offence of this thread, and today they did it again.

Just a drop of background: Many of the towns surrounding Lakewood saw a big funding cut for education this year, and are cutting personnel and programs in response. They are fuming at the state, and are up in arms about it. But Lakewood saw an increase, since it was more underfunded previously.

Now the APP has a headline that stresses that the Lakewood School District is LOWERING taxes for next year. This will probably infuriate residents in all those neighboring towns, who will think based on this headline that Lakewood is getting so much extra money that they can raise less money from the residents. But that is patently false, as shown in a section near the end of the article. The amount of money raised from Lakewood taxes is actually MORE than last year. But the average value of homes went up, so it comes out that the rate per property value went down slightly. But you won't see that from the APP headline, or anywhere clearly in the article.

I can't tell if the writer is just plain stupid (I suspect he is) and is swallowing the narrative of the BOE, or if this is a malicious attempt to spark hatred.

[YitzyS]

Another day, another APP story.

"Lakewood Cheder OK'd for 3-school campus in Jackson," reads today's headline.

It was Bais Faiga which got the approval for girls schools in Jackson. But the headline has to plaster the word "Lakewood" across it to amplify the fear that Lakewood is taking over Jackson.

This reporter is a real closet anti-Semite IMO, who uses subtle methods to stoke fear and anger.