Sunday, January 05, 2020

Some Advice For Political Readers In 2020. Or How To Be Informed.

This may be more of a rant than an advice column, because it is based on my recent deep immersion in all sorts of online worlds (1), almost all political, and because that immersion made me despair about the human race (and so very glad that I have now decided to apply for citizenship in the Elf World).  Still, there might be a few crumbs in the rant that are useful for others.

Here it goes:

1.  When you read about the weird conclusions of some study or about the outrageous sentencing in some court case which makes you believe people have utterly lost their minds, check the original sources before you retweet or join the approval cavalcade.  Do this even if the weird conclusions or findings support your worldview.

This is extremely useful, because the intermediaries who carry you their conclusions have already interpreted the data for you, and by accepting their conclusions you accept their interpretations as valid.  There are times, of course, when you trust the interpreter and save time by just accepting the condensed messages.  But more generally you should always double-check.

I have seen one evolutionary psychology study widely interpreted as being about something it was not, and this was done by People Who Should Have Known Better.  It also could have been easily avoided by just reading the study, or at least the conclusions of the study which do not mention the popularized applications at all. (2)

And once I saw a widely liked and retweeted Twitter post about a study on sexual violence discuss a PAGE NUMBER in the original pdf as meaning something.  It was a page number, for goddess' sake.

That extreme example should serve as a helpful reminder that we don't really know if those who pretend to be experts (or goddesses) online actually have any expertise.  Self-preservation requires a skeptical approach.

2.  Because of the powerful information bubbles we now inhabit online, it has become dangerous to believe that something you see widely cited as factual actually is widely accepted as factual.  If you are a lefty, say, your information bubble probably filters away all studies which don't support lefty views, and it's also likely to keep re-advertising those studies which do support lefty views.  The reverse applies on the right.

Under these circumstances "everybody accepts" or "everybody knows" or "scientists agree" don't mean what they used to mean (though of course tilting the findings and biasing the discussion always happened).

It's not pleasant to go and read stuff in the "other" bubble(s), but it's important to do so.  A bit like seeing the dentists. And by those visits you can learn that your arguments indeed don't have any holes or you get your cavities filled and strengthen your bite.

3.  Number-blindness is one of the worst information epidemics I come across online, from all sorts of people and supporting all sorts of political issues.  Too many people get qualitative arguments but fail in understanding the importance of quantitative arguments: 

It makes a huge difference if, for instance, a group consists of half men and half women or if a group consists of 96% men and 4% women (3)!  The two are not the same kind of "gender neutral" groups.

The meaning of a hundred percent increase in the incidence of an infectious disease should have different interpretations if the country-wide cases of that disease went up from 100 per year to 200 per year than if they went from 100,000 per year to 200,000 per year.  In other words, to properly understand the importance of percentage changes we need to know what the base figures are (4).

It matters if a particular political project would improve the lives of the numerical majority, or the lives of one percent of all people, and it also matters greatly where that one percent is initially located along some relevant measure (say wealth or health).   It matters, in electoral politics, how to get the majority (more than fifty percent of the electorate) to vote for your candidate, and to figure out how to get that overall majority you may need to appeal to many different minorities (5) at the same time, not just one group.  If you want to win the election, that is, rather than remain ideologically squeaky clean.

I'm going to start a file on all the terrible mistakes I see online when it comes to understanding statistical data and also on how very common quantitative ignorance is.


(1)  Reddit.  There is a site called nofap, with hundreds of thousands of members, mostly men.  It's about stopping the use of online pron for masturbation.  Or really about stopping the use of pron as a better substitute for real-world sex, especially for men (given that most pron is created for heterosexual men's viewing desires).  That site made me read a lot on erectile dysfunction among young men, and the question whether increases in it might correlate with the consumption of certain kinds of pron.  The scientific jury seems still be out on that, but then the saturation of the online pron market is fairly recent and future studies will tell us more.

That wasn't the only sex- or gender-related finding I brought home from my travels.  I also realized, after various excursions to not only Reddit sites but to other places, too, that I have lived in an Echidne-bubble where women can actually have short hair and wear work boots and so on, while in quite a few of those places we are very firmly back in the 1950s gender values.  Even some of the Woke World sites tend to be pretty accepting of sexist stereotypes.  Or at least of gender stereotypes.

More generally, I found that many political sites tend to develop their own lists of Approved Study Findings and that debates about those findings are not encouraged but must be accepted without questions.

(2)  I have been too exhausted, after all that travel, to dig up the actual reference I talk about here (it was some years ago).  It had something to do with daughters, sons and divorce, I think.  If there are many demands for more information on this, I might look it up, provided I get donations first!

Just kidding on the donations.

(3)  That example was caused by one UK government authority arguing that penetrative offenses in sex crimes are gender-neutral, apparently because boys and men are among the victims.  This argument was then extended to the perpetrators being treated with gender-neutral terminology, too.

But the actual figures of men and women in the sexual crime perpetrator statistics are like the latter ones, not the former ones.  To me "gender neutral" means something much closer to that fifty-fifty case.

(4)  This is especially clear when a study finds, for instance, that the increase in one's chances of getting some kind of a rare cancer is fifty percent if one has certain risk factors.  That increase does not mean that the new cancer risk someone with those risk factors faces is fifty percent.  It's an increase of fifty percent in the initial (and very tiny risk, given that we are talking about a rare cancer) likelihood.

(5)  By "minorities" here I don't mean just racial or ethnic or religious minorities and so on, but any group with certain shared interests who tend to vote in a similar manner but are not large enough to be a numerical majority of voters on their own.