Posts tagged statistics

why-animals-do-the-thing:

average United States contains 1000s of pet tigers in backyards" factoid actualy [sic] just statistical error. average person has 0 tigers on property. Activist Georg, who lives the U.S. Capitol & makes up over 10,000 each day, has purposefully been spreading disinformation adn [sic] should not have been counted

I have a big mad today, folks. It’s a really frustrating one, because years worth of work has been validated… but the reason for that fucking sucks.

For almost a decade, I’ve been trying to fact-check the claim that there “are 10,000 to 20,000 pet tigers/big cats in backyards in the United States.” I talked to zoo, sanctuary, and private cat people; I looked at legislation, regulation, attack/death/escape incident rates; I read everything I could get my hands on. None of it made sense. None of it lined up. I couldn’t find data supporting anything like the population of pet cats being alleged to exist. Some of you might remember the series I published on those findings from 2018 or so under the hashtag #CrouchingTigerHiddenData. I’ve continued to work on it in the six years since, including publishing a peer reviewed study that counted all the non-pet big cats in the US (because even though they’re regulated, apparently nobody bothered to keep track of those either).

I spent years of my life obsessing over that statistic because it was being used to push for new federal legislation that, while well intentioned, contained language that would, and has, created real problems for ethical facilities that have big cats. I wrote a comprehensive - 35 page! - analysis of the issues with the then-current version of the Big Cat Public Safety Act in 2020. When the bill was first introduced to Congress in 2013, a lot of groups promoted it by fear mongering: there’s so many pet tigers! they could be hidden around every corner! they could escape and attack you! they could come out of nowhere and eat your children!! Tiger King exposed the masses to the idea of “thousands of abused backyard big cats”: as a result the messaging around the bill shifted to being welfare-focused, and the law passed in 2022.

The Big Cat Public Safety Act created a registry, and anyone who owned a private cat and wanted to keep it had to join. If they did, they could keep the animal until it passed, as long as they followed certain strictures (no getting more, no public contact, etc). Don’t register and get caught? Cat is seized and major punishment for you. Registering is therefore highly incentivized. That registry closed in June of 2023, and you can now get that registration data via a Freedom of Information Act request.

Guess how many pet big cats were registered in the whole country?

97.

Not tens of thousands. Not thousands. Not even triple digits. 97.

And that isn’t even the right number! Ten USDA licensed facilities registered erroneously. That accounts for 55 of 97 animals. Which leaves us with 42 pet big cats, of all species, in the entire country.

Now, I know that not everyone may have registered. There’s probably someone living deep in the woods somewhere with their illegal pet cougar, and there’s been at least one random person in Texas arrested for trying to sell a cub since the law passed. But - and here’s the big thing - even if there are ten times as many hidden cats than people who registered them - that’s nowhere near ten thousand animals. Obviously, I had some questions.

Guess what? Turns out, this is because it was never real. That huge number never had data behind it, wasn’t likely to be accurate, and the advocacy groups using that statistic to fearmonger and drive their agenda knew it… and didn’t see a problem with that.

Allow me to introduce you to an article published last week.

Are 20,000 Big Cats Caged in the U.S.? Highly Unlikely.

This article is good. (Full disclose, I’m quoted in it). It’s comprehensive and fairly written, and they did their due diligence reporting and fact-checking the piece. They talked to a lot of people on all sides of the story.

But thing that really gets me?

Multiple representatives from major advocacy organizations who worked on the Big Cat Publix Safety Act told the reporter that they knew the statistics they were quoting weren’t real. And that they don’t care. The end justifies the means, the good guys won over the bad guys, that’s just how lobbying works after all. They’re so blase about it, it makes my stomach hurt. Let me pull some excerpts from the quotes.

“Whatever the true number, nearly everyone in the debate acknowledges a disparity between the actual census and the figures cited by lawmakers. “The 20,000 number is not real,” said Bill Nimmo, founder of Tigers in America. (…) For his part, Nimmo at Tigers in America sees the exaggerated figure as part of the political process. Prior to the passage of the bill, he said, businesses that exhibited and bred big cats juiced the numbers, too. (…) “I’m not justifying the hyperbolic 20,000,” Nimmo said. “In the world of comparing hyperbole, the good guys won this one.”

"Michelle Sinnott, director and counsel for captive animal law enforcement at the PETA Foundation, emphasized that the law accomplished what it was set out to do. (…) Specific numbers are not what really matter, she said: “Whether there’s one big cat in a private home or whether there’s 10,000 big cats in a private home, the underlying problem of industry is still there.””

I have no problem with a law ending the private ownership of big cats, and with ending cub petting practices. What I do have a problem with is that these organizations purposefully spread disinformation for years in order to push for it. By their own admission, they repeatedly and intentionally promoted false statistics within Congress. For a decade.

No wonder it never made sense. No wonder no matter where I looked, I couldn’t figure out how any of these groups got those numbers, why there was never any data to back any of the claims up, why everything I learned seemed to actively contradict it. It was never real. These people decided the truth didn’t matter. They knew they had no proof, couldn’t verify their shocking numbers… and they decided that was fine, if it achieved the end they wanted.

So members of the public - probably like you, reading this - and legislators who care about big cats and want to see legislation exist to protect them? They got played, got fed false information through a TV show designed to tug at heartstrings, and it got a law through Congress that’s causing real problems for ethical captive big cat management. The 20,000 pet cat number was too sexy - too much of a crisis - for anyone to want to look past it and check that the language of the law wouldn’t mess things up up for good zoos and sanctuaries. Whoops! At least the “bad guys” lost, right? (The problems are covered somewhat in the article linked, and I’ll go into more details in a future post. You can also read my analysis from 2020, linked up top.)

Now, I know. Something something something facts don’t matter this much in our post-truth era, stop caring so much, that’s just how politics work, etc. I’m sorry, but no. Absolutely not.

Laws that will impact the welfare of living animals must be crafted carefully, thoughtfully, and precisely in order to ensure they achieve their goals without accidental negative impacts. We have a duty of care to ensure that. And in this case, the law also impacts reservoir populations for critically endangered species! We can’t get those back if we mess them up. So maybe, just maybe, if legislators hadn’t been so focused on all those alleged pet cats, the bill could have been written narrowly and precisely.

But the minutiae of regulatory impacts aren’t sexy, and tiger abuse and TV shows about terrible people are. We all got misled, and now we’re here, and the animals in good facilities are already paying for it.

I don’t have a conclusion. I’m just mad. The public deserves to know the truth about animal legislation they’re voting for, and I hope we all call on our legislators in the future to be far more critical of the data they get fed.

Probabilistic modeling and inference are core tools in diverse fields including statistics, machine learning, computer vision, cognitive science, robotics, natural language processing, and artificial intelligence. To meet the functional requirements of applications, practitioners use a broad range of modeling techniques and approximate inference algorithms. However, implementing inference algorithms is often difficult and error prone. Gen simplifies the use of probabilistic modeling and inference, by providing modeling languages in which users express models, and high-level programming constructs that automate aspects of inference. Like some probabilistic programming research languages, Gen includes universal modeling languages that can represent any model, including models with stochastic structure, discrete and continuous random variables, and simulators. However, Gen is distinguished by the flexibility that it affords to users for customizing their inference algorithm. It is possible to use built-in algorithms that require only a couple lines of code, as well as develop custom algorithms that are more able to meet scalability and efficiency requirements. Gen’s flexible modeling and inference programming capabilities unify symbolic, neural, probabilistic, and simulation-based approaches to modeling and inference, including causal modeling, symbolic programming, deep learning, hierarchical Bayesiam modeling, graphics and physics engines, and planningand reinforcement learning.

via https://probcomp.github.io/Gen/

Bayesian Methods for Hackers is designed as a introduction to Bayesian inference from a computational/understanding-first, and mathematics-second, point of view. Of course as an introductory book, we can only leave it at that: an introductory book. For the mathematically trained, they may cure the curiosity this text generates with other texts designed with mathematical analysis in mind. For the enthusiast with less mathematical-background, or one who is not interested in the mathematics but simply the practice of Bayesian methods, this text should be sufficient and entertaining.

via https://camdavidsonpilon.github.io/Probabilistic-Programming-and-Bayesian-Methods-for-Hackers/

IndexMundi contains detailed country statistics, charts, and maps compiled from multiple sources. You can explore and analyze thousands of indicators organized by region, country, topic, industry sector, and type.

via https://www.indexmundi.com/

Software 2.0 is written in neural network weights. No human is involved in writing this code because there are a lot of weights (typical networks might have millions), and coding directly in weights is kind of hard (I tried). Instead, we specify some constraints on the behavior of a desirable program (e.g., a dataset of input output pairs of examples) and use the computational resources at our disposal to search the program space for a program that satisfies the constraints. In the case of neural networks, we restrict the search to a continuous subset of the program space where the search process can be made (somewhat surprisingly) efficient with backpropagation and stochastic gradient descent. It turns out that a large portion of real-world problems have the property that it is significantly easier to collect the data than to explicitly write the program. A large portion of programmers of tomorrow do not maintain complex software repositories, write intricate programs, or analyze their running times. They collect, clean, manipulate, label, analyze and visualize data that feeds neural networks.

via https://medium.com/@karpathy/software–2–0-a64152b37c35

What does “causality” mean, and how can you represent it mathematically? How can you encode causal assumptions, and what bearing do they have on data analysis? These types of questions are at the core of the practice of data science, but deep knowledge about them is surprisingly uncommon.

via https://medium.com/@akelleh/a-technical-primer-on-causality–181db2575e41

We’ve all heard in school that “correlation does not imply causation,” but what does imply causation?! The gold standard for establishing cause and effect is a double-blind controlled trial (or the AB test equivalent). If you’re working with a system on which you can’t perform experiments, is all hope for scientific progress lost? Can we ever understand systems that we have limited or no control over? This would be a very bleak state of affairs, and fortunately there has been progress in answering these questions in the negative! So what is causality good for? Anytime you decide to take an action, in a business context or otherwise, you’re making some assumptions about how the world operates. That is, you’re making assumptions about the causal effects of possible actions.

via https://medium.freecodecamp.com/if-correlation-doesnt-imply-causation-then-what-does-c74f20d26438

I started a series of posts aimed at helping people learn about causality in data science (and science in general), and wanted to compile them all together here in a living index.

via https://medium.com/@akelleh/causal-data-science–721ed63a4027

The Gini Coefficient, which can measure inequality in any set of numbers, has been in use for a century, but until recently it rarely left the halls of academia. Its one-number simplicity endeared it to political scientists and economists; its usual subject—economic inequality—made it popular with sociologists and policy makers. The Gini Coefficient has been the sort of workhorse metric that college freshmen learn about in survey courses and some PhD statisticians devote a lifetime to. It’s been so useful, so adaptable, that its strange history has survived only as a footnote: the coefficient was developed in 1912 by Corrado Gini, an Italian sociologist and statistician—who also wrote a paper called “The Scientific Basis of Fascism.”

http://www.psmag.com/magazines/january-february–2013/gini-coefficient-index-poverty-wealth-income-equality–51413/