Pomp and Statistics (or: 95% of reporting is wrong)

Journalistic integrity

Imagine if a journalist said this:

The candidate’s recent scandal had no affect on his poll numbers. He lost support in the polls, but that was a coincidence. The movement in the polls was just randomness, and probably noise in how the poll was conducted.

Or this:

The candidate gained a huge amount of support in the polls. We can not explain why. It was probably caused broad, abstract trends that are outside the scope of our reporting.

They would have integrity, but loose their job.

On the subject of presidential elections, there are three main types of discussions:

1) Political philosophy

  • ex: “What is your stance on victimless crimes?”

2) Marketing strategies to appeal to voters

  • ex: “How do you think the speech in Pennsylvania will affect the polls?”

3) Long term demographic trends

  • ex: “Do you think Generation Z leans right?”

I am most interested in #1 and #3. Unfortunately, at least 95% of news is about #2. I have little interest in #2, for reasons I’ll explain.

There is a huge army of journalists dedicated to doing the #2 in a narratively satisfying way, pun intended. Each is tasked with saying something new and novel. Here’s the problem. Trends do not resolve themselves in narratively satisfying ways. Any attempt to ascribe a narrative to them is confining.

But this massive army is employed to ascribe a narrative to random fluctuations, for example:

“______ fluctuation in the polls is explained by _______ campaign news story!”

“______ fluctuation in the stock market is explained by ______ policy change!”

It’s too easy to “map on” news stories to movements in the polls, the stock market, or other data trends. But the matchmaking assumes causation improperly. In other words, it’s almost all fake.

A lack of information would be better than what is generated: a hoard of leftover misinformation.

This is the subject of the book Fooled by Randomness. Focusing on options trading, the book is about our temptation to “explain” randomness (to get attention). And fail. An excerpt: 

As I am writing these lines I see the following headlines on my Bloomberg:

—> Dow is up 1.03 on lower interest rates.

—> Dollar down 0.12 yen on higher Japanese surplus.

and so on for an entire page. If I translate it well, the journalist claims to provide an explanation for something that amounts to perfect noise. A move of 1.03 with the Dow at 11,000 constitutes less than a 0.01% move. Such a move does not warrant an explanation. There is nothing there that an honest person can try to explain; there are no reasons to adduce. But like apprentice professors of comparative literature, journalists being paid to provide explanations will gladly and readily provide them.

Romney

The success of the Trump campaign retroactively embarrassed the reporting of the Romney campaign.

Here are two commonly cited reasons for why Romney lost. First, Romney lost because he was rich, and therefore out of touch. Second, he lost because of misstatements, like the 47% comment that supposedly sunk his campaign.

Along came Trump. A person who regularly brags about his wealth. A man who makes “misstatements” so frequently that you have probably forgotten most of them.

Do you remember when Trump praised the Chinese response to the Tiananmen Square protests? You might not have even heard about it, because it got drowned out by other anti-Trump news. I only know about it because I watched the debate in which the statement was made.

People predicted that these factors would sink Trump’s campaign. They were wrong.

However, one could make the argument that it wasn’t Romney’s wealth or misstatements that sunk his campaign, it was his handling of them.

For instance, he acted like his wealth was something to be ashamed of, which made him seem inauthentic. If you try to downplay something, people think it is because you have something to hide.

When Trump spoke about Romney in an interview by Howard Stern between their respective campaigns, he argued that Romney should have puffed up his wealth more. People probably treated it as a joke, but his results speak for themselves.

To continue, each of Romney’s misstatements hurt him because he was ashamed of them. His cowered apologizing indicated that he could be controlled by the shame centers of the left wing media establishment. He appeared weak.

By contrast, Trump never apologizes. He says, “You don’t like what I said? Well fuck you!” This only makes him more popular, because he is one of the only people who cannot be controlled by the full force of the media machine.

So yes, Romney’s handing of these things was different than Trump’s. But, crucially, Romney only acted the way he did because he was operating under the poor advice that wealth and misstatements could hurt him in the first place!

The reporting of the Romney campaign is an example of journalistic malpractice. They thought they could “explain” the polls, and it became a self-fulfilling prophecy. When the election concluded, they wrapped it up in a little bow.

Until they were proven wrong by Trump. And after Trump, they scrambled to create “explanations” for his victory (racism, fake news, Russian collusion, social media, etc), to tie that election up into a bow. It’s confirmation bias.

Confirmation bias

It is much easier to “explain” something than to predict something. Especially when you pay no cost for giving explanations that are wrong.

That’s not surprising to most people. But this idea is more general than you might realize.

As an example, I found a random statistic from the Internet:

Millennials change jobs more often than other generations. About 21 percent of Millennials report switching jobs within the last year, and 60 percent are open to a different opportunity.

Again, I picked this statistic totally at random. Given this phenomenon, I can generate a great number of explanations:

  • Maybe jobs are becoming obsolete faster than ever, forcing people to move jobs
  • Maybe the economy has changed to favor short-term projects over stable, long-term ventures
  • Maybe the economy is more tolerant of failure, and millennial favor such ventures
  • Maybe millennials favor project-based jobs, where a “gig” is worked on until completed, like contract work
  • Maybe millennials are more risk taking and ambitious, always moving to higher-paying positions
  • Maybe millennials are more wishy-washy, looking for purpose, and lack the attention to devote themselves to a single mission
  • Maybe millennials have not had time to establish themselves into comfortable, tenure-style positions, and are therefore more volatile

I can go on forever. But how many of these explanations are actually true? Maybe all of them, maybe none of them.

Given any phenomenon, it is tempting to create a million “explanations” for it. Any creative person should be able to do so with almost any phenomenon.

That thing you just found out? You can explain it so well! You knew it all along!

But that is a type of confirmation bias. We are treated with the initial fact or statistic (I’ll call the “phenomenon”), and then we try to justify how other facts are compatible (I’ll call the “explanation”). The fact that you can associate a phenomenon and explanation does not attest to the veracity of either.

Here is a hardier test: does the explanation work in reverse? If you did not first know the phenomenon, could you have inferred or predicted it from the explanations that supposedly fit so well?

Probably not, owing to false positives and false negatives.

“False positives” are all of the facts that you are omitting from your explanation because they do not fit the phenomenon. The “false negatives” are the facts that supposedly do explain the phenomenon, but don’t.

In fact, you can spin the same set of facts to “explain” the exact opposite phenomenon!

This has happened to me. I read a statistic, and immediately assume a series of explanations. Moments later, I realize that I read the fact wrong, and in fact the opposite is true. Suddenly, all the same explanations magically work the opposite way.

 

This post is based on a series of Tweets I made a couple weeks ago. You can follow my Twitter account to find out when new posts come out, or to hear my interesting thoughts on the off chance I Tweet something else.