Comments on: In Defense of a Fluke

By: Dara

Dara — Tue, 22 Jul 2008 03:47:30 +0000

Counterexample? Preferably one that actually honors the context of the quote, i.e. social forces and/or causal chains?

By: David Broockman

David Broockman — Tue, 22 Jul 2008 03:35:13 +0000

P.S. “It’s impossible to gather enough data to meet statistical standards while understanding the context that generated each point in that data.” is completely and utterly untrue.

By: David Broockman

David Broockman — Fri, 18 Jul 2008 20:04:42 +0000

The only time I wish I were able/had to put a subject, which would read, “Agreed”.

1. Yes
2. Yes
3. Very Yes

The only thing I’d add on point three is that I think there’s a lot lot lot of areas where people assume empiricism can’t belong but where it can yield real insight, even if it’s only as a component of a larger theoretical undertaking.

By: Dara Lind

Dara Lind — Fri, 18 Jul 2008 19:44:05 +0000

Points I will concede:
-”You can tell a great story about a lot of things that aren’t true.” Absolutely, and I struggle with it. But sometimes it takes hearing a story told at this scale to realize that what’s being described is true, because it resonates, even when you didn’t understand or see it before. (This is particularly true when moving from “Statistics tell me I’m probably X” to “oh, wow, this is exactly what happened to me!”
-”Statistics does often tell us things we don’t know before.” Yeah, I overstated that point.
-”Statistically based knowledge is without a doubt richer.” Than knowledge that could be statistically based but isn’t? Sure. Than knowledge that couldn’t be verified quantitatively if we tried? Ehh, apples to oranges (and this gets to the “anything we care about” thing).

By: David Broockman

David Broockman — Fri, 18 Jul 2008 19:33:08 +0000

This is particularly appreciated with The Flaming Libs discussing something along the same lines today, too, inspired by the XKCD cartoon, but on the wrong side.

First, as much as I truly pity those that prize empiricism above creativity or think the only way we can talk about understanding the world is through proven facts, I don’t think you’re giving statistics their due. In fact, you’re vastly underestimating and completely strawmanning statistics pretty irresponsibly.

Yes, to start thinking about why not just what, you need to start in the land of conjecture — but while you can tell me a great story for a lot of things that aren’t true, well-done statistics do show causality and can describe the here and now, even many aspects of culture. Also, statistics does often tell us things we didn’t know before, it’s not just a method of verification. Finally, statistically based knowledge is without a doubt richer — when you abstract 50 levels from an original hypothesis proving the empirics of every step along the way, your 51st step is much more valuable than if you’ve simply sat there talking.

My usual line of argument in these situations is that numbers can’t describe everything we care about, which you do make towards the end. But I can’t get on board with the rest.