Bush vetoes torture ban

[John Cuthber]

"Well, it depends. It does "work" if you want to get people to talk, but the data you get is extremely unreliable. You'd need to verify everyting that was said."

 

One problem with torture is that it can only possibly be justified when you need data that you cannot get anywhere else. If you can't do that then you can't verify the data from torture but, as you say, you can't trust data from torture victims unless it's verified.

 

Catch 22, you can only hope to justify torture under circumstances where you cannot make use of the data so obtained.

I still say it doesn't work.

[bascule]

And... surprise surprise, in a mostly party line vote, Republicans upheld Bush's veto:

 

http://www.reuters.com/article/newsOne/idUSN1163630920080312

[swansont]

That's not a true statement, swansont. The definition of "it works" was given in the very first post of that thread, which was that in that specific case waterboarding produced information that lead to lives being saved.

 

Now that I've answered your question (again), how about you admit that scientific evidence can be qualitative as well as quantitative, swansont? You said I wasn't presenting evidence because I didn't have a control, which is incorrect.

 

Sure, evidence can be qualitative. I don't think I implied otherwise.

 

Here's why that line of reasoning lacks rigor: If I hire a bunch of psychics to investigate some crime, or predict terror targets, and just one of them gives information that is correct, I get to say that hiring psychics "works." Because I can point to an event where a psychic produced information leading to lives being saved. In reality, though, we know that psychics do no better than guessing. That is why this is an anecdote, and does not meet the threshold of scientific evidence. It's counting the hits but not the misses. If you want to point to this as something that meets the threshold of scientific evidence, you have to do an accounting of the instances where it didn't produce information that saved lives.

[Pangloss]

You're just equivocating, swansont, and you did imply otherwise, and I don't think an argument about whether qualitative evidence "lacks rigor" is beneficial.

 

Have I, or have I not, provided SOME level of something that can be termed "evidence" (even if you wish to qualify it with the word "anecdotal") in the form of a SINGLE INSTANCE of expert testimony? Yes or no?

[bascule]

Have I, or have I not, provided SOME level of something that can be termed "evidence" (even if you wish to qualify it with the word "anecdotal") in the form of a SINGLE INSTANCE of expert testimony? Yes or no?

 

Yes, however you've also gone on to advocate the approach in general without normalizing successes across attempts. Some might call that specious reasoning.

 

By the way, I have a rock that keeps tigers away...

[Pangloss]

Are you an expert on tigers?

[swansont]

You're just equivocating, swansont, and you did imply otherwise, and I don't think an argument about whether qualitative evidence "lacks rigor" is beneficial.

 

You are misunderstanding my point. Since it's quite possible that I've done a poor job of phrasing it, allow me to clarify: it's not the evidence that lacks rigor, it's the analysis. (I already said that the incident is not in question). The datum you've provided is incomplete; it's only part of a larger data set, and only by looking at the whole data set can you hope to draw a scientific conclusion.

 

Have I, or have I not, provided SOME level of something that can be termed "evidence" (even if you wish to qualify it with the word "anecdotal") in the form of a SINGLE INSTANCE of expert testimony? Yes or no?

 

"anecdotal evidence" is not scientific evidence, because you have not produced data that would give a way of knowing how many events did not achieve the desired result. And after you do this, you have to decide what level of signal-to-noise is acceptable or useful. Put another way, what is the statistical probability that any piece of information gathered in this or similar fashion will save lives? You haven't provided enough data to make this determination.

 

As per my example and using your definition, using psychics "works." Why aren't we hiring psychics to ferret out the terrorists and their targets?

[Pangloss]

Again, the comparison is not valid. You still haven't acknowledged the fact that the testimony in this case is objectively valuable because it's coming from a recognized expert in the field. It is NOT the same thing as a random person claiming psychics work.

 

If you're so bent on teaching folks here how science works, why aren't you making the same point in response to the people posting expert testimony that torture doesn't work?

 

Regarding the point of scientific evidence, I think you're wrong there as well. Scientific research studies are done all the time based on input from people, without a control.

[iNow]

Evidence of a procedure's efficacy does not come from self-report. If it does come from self-report, further conclusions from those measures should be avoided. This is basic research methodology with which I'm quite familiar.

 

In order to extrapolate qualitiative results to other contexts, you'd better be damned well sure you have a huge n (aka, population size), otherwise, your results remain non-representative and limited in application.

 

 

Anecdotal evidence of one positive hit hardly meets this criteria. Your signal to noise ratio is also enormous... and equally undefined, since you're not tracking the overall data set.

 

This has nothing to do with control groups.

[John Cuthber]

This is a whole lot of discussion based on one (questionable) data point.

[swansont]

Again, the comparison is not valid. You still haven't acknowledged the fact that the testimony in this case is objectively valuable because it's coming from a recognized expert in the field. It is NOT the same thing as a random person claiming psychics work.

 

Again, the testimony's veracity is not the point in question here. The issue I'm raising has nothing to do with the credentials of the person providing the data point. You have a data point that says useful information was obtained, and I'm not challenging that. It's a boolean state of being "true". That the testimony came from an expert doesn't make it "more true."

 

The issue is how many data points are in the state of being "false," i.e. useful information was not obtained, or the information was bad.

 

If you're so bent on teaching folks here how science works, why aren't you making the same point in response to the people posting expert testimony that torture doesn't work?

 

Because they are using a different definition of "it works" that you are using, and under that definition, I think they've made their case — the information you get is unreliable. If the definition is "a reliable source of information" (i.e. it does so all or perhaps just most of the time) then it does not work, and you can't validly extrapolate "most of the time" from a single data point.

 

Regarding the point of scientific evidence, I think you're wrong there as well. Scientific research studies are done all the time based on input from people, without a control.

 

And this is completely beside the point that I am trying to make. From your link: "One way of differentiating Qualitative research from Quantitative research is that largely Qualitative research is exploratory, while Quantitative research hopes to be conclusive."

 

We are trying to reach a conclusion here, and I'm trying to demonstrate why I think your definition of "it works" is a poor one.

[Pangloss]

I've already acknowledged that the conclusion I suggested cannot be reached based on the single data point I raised.

[swansont]

I've already acknowledged that the conclusion I suggested cannot be reached based on the single data point I raised.

 

I apparently missed that.