Why Bloggers (and Reporters!) Should be Statistically Literate.
I’m going to leave the political dimensions regarding the prognosticating about the Medicaid study to my co-blogger Elias, and instead think about the importance of statistical literacy among bloggers.
Over the past 18 months we’ve had a vivid demonstration of the problems associated with writing about reports that deal in statistics with people who are, in essence innumerate. Everything from the ridiculous non-sense over “unskewing” poll results, to the R&R study’s faulty number crunching and finally the trumpeting of a Medicaid study that can’t even give us a decent sample size has shown us not the limitations of quantitative modeling, but a severe lack of critical reading skills in media new and old on statistics.
The simple fact is, in a world that’s becoming increasingly dominated by coverage of statistical information, the sheer lack of ability to parse that information and point out flaws by the fourth estate is troubling. It’s helping to lead to the stenography type coverage that makes the whole of reporting so problematic, particularly in policy matters.
I have only one problem with your argument – it’s too narrow in scope. Everyone should have some fundamental statistical literacy. Numbers are like magic to a disturbingly large fraction of the population.Report
This is true, but my point was more that at the very least people who are trying to translate statistical analysis to “plainspeak” should be absofuckinglutely required to know what things like margin of error, sample size etc. mean before they’re allowed anywhere near a word processor to write about say political polling or analyses like the Medicaid paper.Report
I decided to do a little web surfing before I started studying for a final I have later this week. Lo and behold I see a post about stats, which just so happens to be the final I will be studying for after I type this little comment up. My instructor has been drumming into our heads all semester that he knows we won’t be statisticians, his goal is that we become good consumers of data. Right now we are working on hypothesis testing, before this I had no idea what margin of error really meant or what a 95% confidence level was.Report
I can’t argue with that.Report
Do you have any data to support this claim?Report
Oooh, nicely done!Report
I’d say statistics is my third or fourth language at best, but even I can make them say anything.
The standard by which we evaluate results extrapolated from statistics should be the same as the standard we use for everything else: does it have predictive power? Are the predictions suggested by certain data sets useful? As for the fourth estate, it’s grasping at whatever it can to stave off its own death: the irrelevance of the mass media becomes more and more obvious every day.Report
Well reporters are notorious for reporting on subjects, areas, fields, etc. in which they have no education or experience, but that never stopped them. Why would this be any different? Hell, often times they don’t even get the FACTS correct.Report
“Over the past 18 months we’ve had a vivid demonstration of the problems associated with writing about reports that deal in statistics with people who are, in essence innumerate. ”
I don’t think that Tyler Cowen is innumerate, although you wouldn’t know it from his posts on the subject.Report
BTW, Kevin Drum has looked at the study:
http://www.motherjones.com/kevin-drum/2013/05/what-we-know-oregon-medicaid-study
http://www.motherjones.com/kevin-drum/2013/05/no-oregon-medicaid-study-did-not-show-no-effect
(cross-commented on Elias’ post, if you’ve read the comments there)Report
Do you have anyone in particular in mind with this?
Or are we just raging against imaginary ignoramuses? (Or is it ignorami?)Report
My initial draft post was titled: “Why Ross Douthat needs to take a stats course”…but I decided it might be a little too personal.Report
“My initial draft post was titled: “Why Ross Douthat needs to take a stats course”…but I decided it might be a little too personal.”
Or fruitless; he’d be on my list of people probably truly incapable of learning statistics (or anything requiring an ability to think).Report
In (dubious) fairness to many conservatives, I think statistical analyses are often understood and rejected because the conservative project is primarily concerned with normative politics and not positive analysis or feasibility. The standard conservative critique that statistical analyses fail to account for unintended consequences doesn’t really incline itself to premises that rely on statistical backing.Report
The Green Lantern theory of politics meets model specification!Report
Well put.
But I do think conservative skepticism of statistical models can be of use to greater society.Report
I disagree. The conservative criticism of, say, health care reform or gun control is as grounded in “what is” as “what should be”.Report
From my vantage point I certainly saw a whole lot of liberals pointing to statistical analyses suggesting Obamacare would save the country money in the medium and long terms and a whole lot of conservatives shouting “socialism” and “death panels” in response.Report
From my vantage point I see a whole lot of conservatives pointing to budget analyses that show Obamacare’s costs being much higher than predicted as well as shouting “socialism” and “death panels”. And I see a whole lot of liberals mangling international statistical comparisons and shouting that health care is a right and a moral imperative.Report
Mangling international comparisons is a bi-partisan sport. If we’re going to look at stats then you have to honestly admit what you are touting. Cost estimates are nice, but are not the same as physics. Cost estimates are dependent on the assumptions built in and it isn’t really a surprise that a conservative think tank will build in assumptions that back up R positions. Same thing for a liberal think tank. So estimates can be useful but are not exactly the same as a fact or even factoid. International comparison can also be useful but can’t prove everything. I know i’ve used int. comparisons to prove that a concept can work ie: Conservative says gov provided uni health care cannot possibly work. Me: that is objectively wrong since here are many countries where is does work. Does that imply what is morally right or what we should do, no. But those are separate questions.Report
“Mangling international comparisons is a bi-partisan sport”
here here, but it can be done well, and it can be done poorly.Report
Agreed. Both sides engage in fair and unfair use of normative and positive points.Report
Yet, in general, in my experience, conservatives tend towards know-nothingism when it comes to statistics, whereas liberals tend to believe anything with numbers attached.Report
Chris, that strikes me as not just wrong, but way-off wrong. (I’m thinking particularly of tax and trade policy, but I also have to note that the two most recent creepy anti-science trends, opposition to GM foods and vaccination, have found their home on the left.)
It seems so wrong that I have to suspect that there’s some confirmation bias on both our parts. We’re probably comparing the most rational thinkers on our sides to the worst populists on the other sides.Report
Generally speaking, statistical arguments don’t play well with populism.Report
4 out of 5 liberals say that Chris is a bit off here. The anti-vax and anti gm food crap is found more on the left. but most of the hard core anti-evo folks like to trot out “research” that they say supports them. Everybody like numbers and graphs and science kind of stuff when it supports what they already believe. The key difference is in how resistant a person is to having their mind changed or even open to learning before making a judgment on something. A YAC is going to disbelieve any evidence for evolution regardless. A person who doesn’t know much about GM foods but is open to the scientific process and evidence will read a bunch of different resources before deciding whether it will give them cooties or not.Report
Pinky, if by tax and trade policy you’re referring to regressing all of economics to almighty GDP you’ll get no argument from me that conservatives are the ones who look at numbers to justify their position. But these tend to be a particular breed of well-bred conservative, which ties in to Pat’s point.
As for opposition to GM food and the antivax movement, I actually find it all coldly rational if not excessively cautious and very selfish. New foods are my scrutinized the way new drugs are. Why is that? The antivax movement also has some vindication wrt the Fort Dix outbreak, although I think in general the antivax movement goes to show the dangers of a little knowledge.
As for me, I don’t really have a side on this. These are just some observations I’ve made, and I’m definitely willing to concede that my observations may not be representative of the whole.Report
Anyone have any thoughts about the Heritage Foundation article and subsequent thread? I’m struck that no one’s addressing the question of whether immigrants have lower IQ’s than the native white population of the US, and whether those IQ differences persist. They’re talking about politics and whether Heritage should have hired the guy.Report