Wednesday, March 19, 2014


I was going to ignore Leon Wieseltier's get-off-my-lawn attack on young whippersnapper Nate Silver's preference for data-driven journalism, but this jumped out at me:
Many of the issues that we debate are not issues of fact but issues of value. There is no numerical answer to the question of whether men should be allowed to marry men, and the question of whether the government should help the weak, and the question of whether we should intervene against genocide. And so the intimidation by quantification practiced by Silver and the other data mullahs must be resisted.
But there is very much a "numerical answer to the question of whether men should be allowed to marry men" -- or at least the opponents of gay marriage strongly suggest that there is. Those opponents argue that gay marriage harms society -- specifically, they say that children suffer harm from not having two opposite-sex parents. How do we know this is nonsense? We can look at the lives of children raised by gay couples and compare their well-being to that of children raised by married heterosexuals. If gay marriage were harming the children of gay couples, we'd know it, but it isn't. And it's good that we have studies showing a lack of harm, because if we were high-mided and Wieseltierian and chose to remain above the tawdry collection of data on this subject, the anti-gay right would generate all sorts of anti-gay-marriage data and drive the debate with it. (Perhaps Wieseltier needs to be reminded of the preposterous statistics about gay people's health that have been circulating online and elsewhere for several decades -- "the lifespan of a homosexual is on average 24 years shorter than that of a heterosexual" and all that.)

There is also very much a "numerical answer" to "the question of whether the government should help the weak" according to critics of government aid programs:
The War on Poverty, which President Lyndon Johnson declared nearly 50 years ago, has "failed miserably," House Budget Committee Chairman Paul Ryan said Thursday -- and he wants to figure out what approaches would work to get Americans out of poverty....

"Next year marks the fiftieth anniversary of the War on Poverty. We've spent approximately $15 trillion and the question we ought to be asking ourselves is, 'where are we?' With a 15 percent poverty rate today -- the highest in a generation -- and with 46 million people in poverty, I would argue it's not working very well."
Rebuttals to this argument also rely on statistics. That's because we're not discussing this on the basis of morality, as Wieseltier airily suggests; we don't simply assume that government has a responsibility to help the weak because the right incessantly argues that it can demonstrate the failure of any such efforts -- with data.

And as for genocide: Does Wieseltier seriously believe that we regard intervention as an unquestioned duty? If so, how does he explain the fact that we intervene in some genocides and not others? Why does he suppose that is? I'd argue that our leaders consider morality, but also calculate the potential cost in blood and treasure, while pondering poll numbers for or against intervention. I don't what the hell Wieseltier's explanation would be.


Victor said...

Morality is subjective.

One man's moral patriot, is another mans immoral terrorist.

One religions moral profit, is another religions lying immoral heretic.

This is the same argument that that idiotic woman who was shown that she would save money with Obamacare, but said, "No. No, I don't believe that."

Or, when our stupid Speaker of the House, when questioned about his statement about 6 million people lost their health insurance because of Obamacare, if he really thought said, "Yes. I believe that."

The problem isn't that 'there are lies, damn lies, and statistics" - it's that you can't use only the statistics that support your preconceived notions, without weighing them against the statistics that don't support your preconceived notions.

That's called having wisdom.
It's a hell of a lot different from having morals.

Ron Chusid said...

While I agree with you regarding the facts and numbers involved, the fact remains that any discussion of gay marriage does also involve values--in this case the values of individual choice and separation of church and state in opposition of conservative values on this issue. I think this is what Wieseltier is getting into when questioning Silver's approach.

I think you might be interpreting Wieseltier's article a bit different from how I did, having read it just after Paul Krugman's defense of his style of opinion writing against what he saw as an attack from Nate Silver.

The debate between Krugman and Silver is one where neither side is entirely right or wrong and the differences between the sides are exaggerated when this turns into a blog debate. I don't think that either Paul Krugman objects to presenting the numbers or that Nate Silver really thinks that everything comes down to the numbers.

Roger said...

If there's a number of times that one can snort coke off Maureen Dowd's ass without suffering brain damage, Leon passed that number long ago.

Unknown said...

I think a lot of op-ed commentators are indeed filled with BS. A classic example is Thomas Friedman, who takes a bunch of anecdotes (many probably made up since I really doubt he talks to his cab drivers as much as he claims) and spins a story that matches his world view. Nate Silver's attack on such pieces is well deserved. OTOH, there are OP-ed columnists who don't do this, such as Krugman (although this may not be evident solely in his op-ed pieces)who presents the data that informs and drives his opinions extremely effectively and in depth in his blog, nearly every week.

Unfortunately, it seems from the initial pieces on the new fivethirtyeight that Silver has opted to go to the other extreme end. A good example again is the post Krugman mentions, talking about how the corporate cash numbers were revised downwards. This piece provided absolutely no context as to what that actually means.

Even worse, the site seems to be going towards a forcefully contrarian end where stuff is simply presented in a "hey, we found a datapoint which indicates the experts may be wrong...haha" without providing context that shows how that 1 datapoint fits in with all the other relevant data points (which may make the presented one appear nothing more than statistical noise or an anomaly).

Even worse is the direction it has taken by assigning Roger Pielke Jr. science duties, whose first article contains dishonesty (by cherry picking the start date from the raw data it points to, starting graphs from 1990 instead of 1980 as the original does) and worse, ignoring the salient data points (that only meteorological and weather related disaster costs have increased, while geophysical disasters like earthquakes have remained constant).

In sum, 538 appears to be a combination of Freqkonomics, Gladwell, and the op-ed columns Silver derides with a veneer of numbers thrown over it.

Quite disappointing actually since I was looking forward to it.

Steve M. said...

Yeah, the actual FiveThirtyEight has been a real letdown so far. But the weakness seems to be a temptation to publish contrarian linkbait, as you say, not the interest in data per se.

Varun Prasad said...

I agree. The focus on data is important and necessary and maybe I missed it in reading your article (I have no idea what Leon's argument is, tbh. I read both hours and Aron's interpretation in I), but I think the importance of data and numbers lie in how they can inform our values.

For example, I wouldn't be a gun control advocate if we didn't have the Australian example, which showed a significant positive effect when they introduced more stringent gun control. Of course, we don't have more relevant data for the US because the NRA pretty much succeeded in gutting any research money towards gun ownership and its effects. However, if the Australian data had shown no, or negative effects, I doubt I'd be a proponent of gun control.

Unknown said...

But you aren't "reducing" these issues to data, at least considering the way you're arguing we should use data to help us decide the issues you mention. The data is certainly relevant, but to say that you can "reduce" these issues to data is to say that the data is _all_ that matters. For instance, we can get some data about whether same-sex marriage harms children. Whatever the answer is, what do we do with that data? In what way does it matter? Answering that requires going beyond the data to ask about values. (And, even before that, the notion of 'harm' is already value-laden, so even to collect the data you've got to make some value judgments.)

Silver seems to be after some sort of impossible 'value-free' or 'objective' take on issues. And Wieseltier is right that that's impossible.

Of course, Wieseltier is an arrogant gasbag. But that doesn't mean he's wrong here.

Barry DeCicco said...

Steve M. said..

"Yeah, the actual FiveThirtyEight has been a real letdown so far. But the weakness seems to be a temptation to publish contrarian linkbait, as you say, not the interest in data per se."

The point is that contrarians are anti-data, as well as anti-truth.

Diptherio said...

I'm reminded of something Rabbi Dov Baer would say when confronted with two contradictory interpretations of scripture and asked to determine which was correct: "these and those are words of the living God."

Which is just a high-falutin' way of saying, they've both got a point. Silver wants to help people be more numerate and that would definitely be a good thing. Pundits might have a harder time playing fast-and-loose with statistics ("seasonally-adjusted" labor stats), and who can complain about spreading more and better math(s) knowledge? Of course, Wieseltier is also quite correct in pointing out that value and ethical judgments are not susceptible to mathematical formalization.

All too often, the pronouncements of our pundits (left, right and center) seem woefully divorced from grounding in real world data. See, for instance, proclamations about the devastating effects of raising the minimum wage, when the data is actually mixed and inconclusive; or the aforementioned labor stats, presented in seasonally-adjusted form without comment by the intrepid NPR news team--lookie here, we added jobs in January! More numeracy among the populace surely can't hurt our national discourse.

But we do need to be leery of quants who present math(s) as the answer to everything, not realizing that their unexamined philosophical biases are baked right into their regressions. We need to recognize that there are different tools for different jobs, and just as you wouldn't try to build a house with just a hammer or just a saw--but would use both and many more besides--so too we shouldn't try to build our understanding of the world using just math(s) or just moral reasoning.

I think there's probably plenty of room on the interwebs for both Silver's data journalism and Wieseltier's philosophically-grounded B.S. (that's "B.S." for "Belief System," btw), as well as a need for both.