Numbers, numbers, numbers...

 Michael – Thursday

I’ve written about the way numbers can be fake news before, and I suppose it is something of a fixation of mine. The reason is that all data is eventually turned into numbers and then the numbers are equated with information. Unfortunately, that can be as much of a leap as some of the way out conspiracy theories of QAnon. Garbage in, garbage out.

There’s no point in giving an example from the Trump era – almost everything he said needed to be taken with a lorry load of salt. So let’s move to the current incumbent. It’s obvious that he’s honest and committed, and has an impressive agenda. His second huge multi-trillion dollar initiative is a wide-ranging infrastructure bill that will not only update and repair existing US infrastructure, but modernize and green the economy and create 19 million jobs. “Independent analysis shows that if we pass this plan, the economy will create 19 million jobs — good jobs, blue-collar jobs, jobs that pay well.”

19 million? Huh? That is a staggering number. The US workforce is estimated at around 165 million and unemployment in 2020 was some 4%.  That’s about 6.6 million job seekers. Hmm. Where will the other 13 million workers come from. So I went back and read the report more carefully. Ah! It turns out to be over ten years. That makes more sense. Well, that’s growth of say 2 million jobs a year, and only in the areas relevant to the infrastructure program. It still seems inconceivable that the workforce can grow that quickly, especially with good jobs, blue-collar jobs, jobs that pay well, unless there's huge immigration of skilled people. That doesn’t seem to be on anyone’s agenda. 

So where does the 19 million figure come from in the first place? Where is the independent analysis from? It’s a report by Moody’s Analytics on the predicted growth of the US economy over the next ten years using economic models. Ten years is a very long time in the complex modern world. If there’s any salt left over from that lorry load, it may be good to keep it handy. Nevertheless, Moody’s is a serious source for numbers. They say 19 million new jobs from the infrastructure bill.

Except that’s not what they say at all. They are looking at the situation with the economy based at the start of the fourth quarter of last year i.e. devastated by covid. With no government intervention at all, they expect nearly 16 million jobs to be created over the ten years. The American Rescue Plan (the first multi-trillion bill) makes little impact over the ten years – its role is to kick start the economy over the next year or so i.e. to get people who lost their jobs back to work. When the dust settles, the infrastructure plan is expected to create an additional 2.7 million jobs, say 270,000  new jobs per year. That’s completely believable and, actually, terrific!

Note that the president’s statement is logically correct. In logic, if statement B is always true, then statement A implies statement B is also true whatever A is. But the “information” we draw from the way it’s worded is completely wrong.

There's always trouble when we try to reduce reality to numbers. In her new book, Counting, How we use numbers to decide what matters, Deborah Stone points out that we get what we count, so we better count the right things. She mentions Soviet textile factories that were given production targets by total length produced. They produced long narrow strips. And in the era of railroad construction in the US, payment for a project was sometimes based on the total length of track laid. A section in Nebraska took a wide, totally unnecessary, arc to add extra profitable miles of rail.

More subtle examples are found in many areas. In education, there’s an ever raging argument about how to assess students in a way that is fair, culturally neutral, and avoids rote learning. Collaboration is important both in learning and in almost all types of work these days. (Even writing mysteries in our case!) Naturally, reference materials and the internet are essential. Yet most examinations require students to write a paper they haven’t seen before or had time to think about, with no collaboration with any other person, and no use of any reference materials, computers, or the internet. Wow! Exactly how  we expect them to operate in the real world!

Then, of course, there’s IQ measurement. Since we don’t really know what intelligence is, it’s not surprising that we aren’t that great at measuring it. But IQ does correlate with things like income, longevity, professional success, so it is important. Early studies across geographies suggested differences in average IQs. For example, results from high-school students in South Africa and Botswana were below European averages. The numbers were correct. Except that the tests were administered to young kids in English instead of their home language. Garbage in, garbage out.

In this maelstrom of issues around interpreting numbers, trying to understand cause and effect versus correlation, and (sometimes deliberate) misinformation, we have to make important decisions. A discussion I had yesterday:

So would you say I should get a jab when they offer it?

Yes.

But don’t they cause blood clots?

Well, only very occasionally. And only one of them.

But they all do the same thing, right?

Well, no, not actually. It’s complicated.

Suppose it turns out that only a few people get clots immediately, but eventually everyone who has had the vaccine does?

Well, they don’t think that’s how the process works.

But they don’t know, right?

Well…