Nobel laureate psychologist and Economist Daniel Kahneman's book Thinking Fast and Slow posits that we have two ways of making decisions: fast and slow as you might suspect from the title, or as he calls them: system 1 and system 2. Reading it might just cause you to reevaluate a lot of things you've been led to believe because people who want to enlist your support, sell you their products, receive your donations and secure your votes make skillful use of this knowledge.
We have statistics thrown at us every day and we make decisions based on our statistical illiteracy, our intellectual laziness and the tendency to make decisions based on limited facts and wishful thinking. So much of Kahneman's work applies to how we choose investments, but it applies to virtually everything we prefer to approach with the unreliable "system 1" rather than to wade rigorously through volumes of data until our heads hurt. In a large part, "studies show" is enough for most of us. there's always a study, an authority, a book and it's usually enough. We prefer to judge, to evaluate on limited evidence and we prefer intuition over analysis and when we believe something to be true, whether it's because we identify with a party or an organized faith or an ethnic group, we are
"very likely to believe arguments that appear to support it, even when these arguments are unsound. If system 1 is involved, the conclusion comes first and the arguments follow."
That doesn't sound very profound or startling and yet we virtually always fail to detect such tendencies in ourselves, perhaps more so when we think we are highly intelligent. As Sheldon on the Big Bang Theory says "If I were wrong, don't you think I'd know it?
We tend to disallow legitimate science and valid statistical conclusions out of hand if we are engaged in politics that argue to the contrary and apparently intelligence has less than enough to do with it. The difference between that quoted statement of the obvious and the body of work summarized in the book is an enormous amount of documented science. ( I'll mention the Nobel prize once again.)
But this is not a book review and what I'm getting at is not another Lefty attempt to show the shoddy thinking and fallacy filled folly of the Right Wing. I'm making a case for divorce. I'm just not getting the kind of liberal thinking out of the Democratic party I need. I feel neglected and betrayed, ignored and saddened at the endlessly tainted logic, the misused statistics and conclusions supported by faith and gerrymandered facts. I want a divorce. Let me explain.
I got an appeal yesterday, to rethink my position on the legal modifications to prior self defense laws the press likes to call "stand your ground." Places that have such laws have more murders, states the pleading, and not just more murders but more murders of African Americans because non-African Americans tend to see dark skinned people as hostile and dangerous and likely to be carrying weapons. But don't take their word for it and you guessed it: studies show.
It so happens that I took part in the cited study which consisted of pressing a key with either the right or left hand to show whether the small black and white images of human eyes belonged to white or black people. Small images were also displayed which included items like crossbows and maces. Delays in pushing either key determined whether or not I was a racist and wouldn't you know it -- I am a racist. The study shows and never mind the various possible reasons for time delays. Never mind any other study.
The problem from my standpoint is that it was impossible for me to determine the 'race' of well more than half of the eyes much less analyze the emotion and there was a greater delay in pushing the key when I had that problem. The delay is said to be proof that I associated black people with weapons and hostility, albeit mostly medieval weapons. Seriously, but none the less, studies show.
Now from this study, taken to be conclusive and irrefutably so, comes the leap of fallacy that I cannot be trusted to determine whether or not someone is a threat to my life because I'm a racist and by extension, no other whit person can be trusted because white people are racists. I would expect Liberals to gasp and perhaps to gag, but Democrats don't. Democrats cling to this conclusion because studies show it's safer for to require me to run away from the attacker with a gun, white or black, and because they're Democrats after all. The conclusion comes first and the arguments follow.
Back when the Feds finally killed the 55mph speed limit, there were impassioned arguments from self-styled liberals like Alan Dershowitz telling us that there would be a bloodbath and indeed for a very short time there was a slight increase of highway fatalities. It was a statistical blip and those fatalities have continued to decline. Dershowitz made the mistake of taking statistically irrelevant data as proof of a trend -- of ignoring, as most of us do, regression to the mean and of choosing facts to fit the prior conclusion .
So my first reaction to the "states with SYG laws have more. . ." statement was the question: did they have more beforehand? and has the alleged increase been long enough to have statistical significance? Will Regression to the Mean make this increase go away? Without answering that question I'm supposed to hear Studies Show and jump on the bandwagon. Sorry, I can't and because it's nearly universal in political advertising to play games with statistics, I have to side with Dr. Kahneman again when he explains that a few facts and a plausible story tend to trump informed and rational conclusion. WYSIATI he calls the gambit: "what you see is all there is" and we're all prone, in our laziness and longing to belong, to fall for it. If we fear guns we won't ask questions. If we are invested in guns and gun rights, the opposite is true. First comes the conclusion.
The problem with interpreting statistics isn't just the pervasive ignorance of statistical method, the confusion between context and causality, but the laziness we're encouraged to cultivate in our consumer society. The vast majority of educated people will read the syllogism:
All roses are flowersSome flowers fade quicklyTherefore some roses fade quickly.
says Kahneman and agree, but of course it's not true, because it's possible that there are no roses among the some flowers that fade quickly, but my head hurts when doing the work needed to analyze it. It looks plausible, the statements before "therefore" are true even if they are not the only facts that need to be considered and therefore we settle for "it's true" and will defend any argument based on it's truth. What with our liberal backgrounds, it's easier to agree that all white people are racists and defend policies based on it than to question Liberal authority. Easier to explain gender differences on society than on science and therefore it's fine to suppress data and prevent studies which might show. the studies that agree with us are always right.
Because it gets cold outside and insulation is necessary to sustain life, no one should interfere with our right to own warm clothing.
Does that mean that we're not allowed to own a coat in Key West but only where it gets cold? You know what I'm getting at here. Context means more than what's actually said. We see what we want to see, believe what we're disposed to believe. We're not afraid of mittens, but still we think we're rational. We think we're guided by science and are objective, independent thinkers. If I am actually to be all that, can I still be a Democrat? Can I still be a Democrat if than now means I can't trust anyone in any way, that heresy is everywhere and the job of Democrats is to root it all out and punish it even if we have to cut a swath through truth, science and humanist values to bring the unbelievers to the stake? No, I think I need a divorce.
An interesting read. IMO, there are difficulties between the scientific method in the physical/biological sciences
ReplyDeleteand that in the social sciences, the latter relying necessarily on statistics. As you note, even a simple syllogism
can be misleading, while statistics is prone to more misinterpretation. An old German engineer once observed
that statistically, all Americans have one testicle and one breast.
Statistics are such a powerful tool and so easy to misuse, but I've become almost convinced that being rational is almost impossible and that even the smartest people can be mislead.
ReplyDeleteBB-Idaho has a point. The trajectory of a golf ball is described by a simple quadratic equation. The challenge of billiards is vector analysis. And the tools that measure human behavior and complex social systems are the blunt instruments of statistics, which I define as “methods of measuring meaningful distributions against a background of random noise.”
ReplyDeleteOne common misconception of statistics is that most people - especially journalists who report study findings to the public - think of statistics as deterministic; that such-and-such WILL happen as assuredly as the Sun rises and sets. Wrong! Statistics measure probabilities and central tendencies. There will always be sampling errors, research biases, iatrogenic artifacts, misattributed or unidentified study variables, statistical outliers, and outrageous liars.
The analogy I use to describe “statistical significance” is to visualize a cloud in the sky. The outer edges are bright white fading to blue – where data values are accidental or random in nature (correlation R = 0). Where the center is dark and menacing, statistically meaningful data rains down upon us (R = 1).
Now think of these clouds as Venn diagrams where circles, either intersecting or non-intersecting, represent topics under discussion: Milk, gluten, calories, income inequality, statistical illiteracy, science, Big Bang Theory, Left Wing, Right Wing, Democrats, Republicans, Gerrymandering, Stand Your Ground, race, gun violence, speed limits, highway fatalities, context and causality, climate and overcoats, mittens and divorce.
In trying to visualize these clouds on a hot and humid summer afternoon, I ask the Zen of Venn: Is there an intersection of circles that embody all aforementioned variables? And what percentage of the population is represented at this intersection? A million? Ten thousand? A dozen? One existential rage against the storm?
In this instance, N = 1.
Welcome home!
ReplyDeleteIndeed one of the major points of the book is our collective inability to assess probabilities and risks and the media certainly take advantage of that, getting us half hysterical about million to one odds while we ignore the nearly inevitable. Those mittens gonna getcha if you don't watch out and of course correlation and causation, association and determination are the same thing, aren't they? Minor factors loom while major ones are unnoticed. Then too, so many "studies" are attempts at self fulfilling prophecy: Obama's ruining the economy as it steadily improves.
Ah well, that old existential road rage is getting to me. Post hoc ergo propter hoc and Hieronymo's mad againe
Almost home … but not quite. A two-week road trip extended to a month is still passing before my eyes.
DeleteTo give myself time to fully recharge my ink and return to (b)flogging full-time, my self-assigned deadline is MANOVA Day, the national holiday when all statisticians cease their labors to celebrate the standard deviations of labored statistics. Go figure!
Manova? I once heard someone shouting manova board but I ignored it. It would have taken too long to calculate the odds of saving him and of course I suspect he was deviated in some way.
Delete