Thursday, February 14, 2008

Old Methodology/Theory Debates

625 Comments:

Anonymous Anonymous said...

"social science stats classes"

For the record, there is no such thing as a social science stat class. There are statistics and there is social science -- two entirely different disciplines. Perstroikans frequently hold that many quantoid political scientists are really statisticians, and not political scientists. Given the complete lack of understanding of philosophy of science (much less political philosophy) manifest by the quants on this blog, this reasoning clearly has some validity.

3/31/2008 8:05 AM  
Anonymous Anonymous said...

there is no such thing as a social science stat class.

The chair of my department is going to be furious: he thinks I'm teaching a social-science stat class next Fall. More fool him. A semester off for me!

C'mon man: who cares what Perestroikans think? What is this? 2001?

I'm guessing there are more people who have first hand experience of the moon's surface than agree with all the 'determining cognition' and 'cannot generalize' cabaret above.

3/31/2008 8:17 AM  
Anonymous Anonymous said...

"For the record, there is no such thing as a social science stat class."
---------------------
For the record there are. You see YOU do not get to set the definition of things.


Seriously you don't.


I am not kidding.

There are social science stats classes. They exist and the teach all the things you hate.

You cannot change that anymore than you can find a single citation in the the philosophy of science that supports your bizarre claims.

3/31/2008 8:28 AM  
Anonymous Anonymous said...

"There are social science stats classes. They exist and the teach all the things you hate."

In the first place, there is nothing in the discipline I "hate." That is a strong word. Secondly, as someone vigorously argued, statistics are separate and distinct from social science subject matter. Was this person wrong?

Moreover, how does a social science stats class look differently than a non-social science stats class? Is it simply the data that is input into the statistical techniques?

3/31/2008 8:38 AM  
Anonymous Anonymous said...

Many political science graduate programs out-source their stat courses to other departments.

3/31/2008 8:51 AM  
Anonymous Anonymous said...

Moreover, how does a social science stats class look differently than a non-social science stats class? Is it simply the data that is input into the statistical techniques?
---------------------
You really don't know this? You constantly attack things you have no clue about.

Here is the way I see it. There are different data generating processes out in the world. the datat that they generate have different properties. Further our ability to measure the data introduces another set of complications. As such there are different statistical techniques developed to deal with the differences in data generating processes and measurement issues. Some of these techniques are developed by statisticians and some are developed by researchers with in a specific substantive area. Some areas have data and problems that are so unique they have distinct subfields devoted to developing and refining techniques that deal with the nuisances of the data they encounter in their field.

A social science stats class will teach students some subset of this wide world of statistics. Usually techniques that take seriously the underlying data generating process and measurement issues inherent in social science data. Much liek a bio-stat class will teach a sub set of stats that deals well with teh type of data they often deal with.

3/31/2008 8:59 AM  
Anonymous Anonymous said...

"You constantly attack things you have no clue about."

On one has "attacked" statistical techniques. What has been critiqued is the philosophy of science underlying much of their application in political science. Nonetheless, thank you for your thoughtful explanation.

I will add that much of what you describe sounds more like a methods course than a stats course. Of course, you can be teaching data collection with an eye toward breaking data down to be useful via statistical techniques. This still does not make it a stats class. Moreover, just because a particular statistical technique was developed or refined to deal with a specific social science issue, does not make it a social science stat -- again, there is no such thing.

3/31/2008 9:11 AM  
Anonymous Anonymous said...

RE: 8:59 a.m.

I think a problem we have seen on this blog is that people get so vested in statistical methods that they are unable/unwilling to think critically about their application. Most everything I have seen on the qualtoid side of the discussion is reasonable and logical, but the quantoids are almost inexplicably vehemently opposed to any criticism or limitation attached to their methods. I think their rigidness and dogmatism does a disservice to the quant project.

3/31/2008 9:21 AM  
Anonymous Anonymous said...

9:21,
The bluster and preening of the quants has made this blog fun. So leave them alone. The quant project is completely unaffected by what takes place on this blog.

3/31/2008 10:33 AM  
Anonymous Anonymous said...

Moreover, how does a social science stats class look differently than a non-social science stats class? Is it simply the data that is input into the statistical techniques?

I'm going to assume this was asked in good faith...

Random utility models are almost always taught (typically as part of GLMs) in social science statistics, but are not particularly common in statistics more generally (although GLMs are commonly taught).

You see the distinction in RUM-related problems like 'ideal point estimation'. Social scientists would assume an underlying utility structure whereas straight stat guys normally pursue some scaling/clustering approach. The latter can be model-based (in the sense of different underlying populations who might have different tastes etc), but there isn't usually (that I've seen) an explicit behavioral assumption underpinning the data.

Straight stats guys also don't do much in the way of structural estimation, but that's pretty common in social science. I actually don't know how stats guys would translate a theoretical model of human behavior in that
sense, but that'd presumably be familiar with reduced-form techniques.

Last, I tend to think of the behavioral economic econometrics as pretty much unique to social science. I'm talking about implementations of models which are underpinned by 'quantal-response equilibrium' or 'cognitive hierarchy' (or actually, models of 'cursed equilibrium' which I've not yet seen personally). Straight stat guys wouldn't be too interested in 'agent error' and the like.

There are probably more examples, but these are just some I know.

3/31/2008 11:06 AM  
Anonymous Anonymous said...

There are statistics and there is social science -- two entirely different disciplines.

If we're going to be splitting hairs--is social science a "discipline" just like statistics?

Sorry, but since you've lambasted quants for--what you believe--is a misnomer, then shouldn't you be held to the same standard? Social science isn't a discipline, is it? Political science is a discipline, sociology is a discipline, etc. Am I wrong?

3/31/2008 12:31 PM  
Anonymous Anonymous said...

As a general comparativist, I think the degree of disgreement between the qual and quants is completely misplaced. I do not want to sound trite, or seem like I am promoting compromise simply for the sake of promoting compromise, but both sides are right. The qualtoid is right that behavior is determined by human thinking (what else would it be determined by); we cannot generalize from one case (even if that case is the U.S.); lastly, human behavior is unpredictable (there is no denying that). These are achingly simple points, and philosophy of science 101. Where the qual(s) are decidedly wrong is that somehow better (or equal) social science is derived from qualitative description than from quantitative methods.

Going to the red light issue, as a general comparativist I would use some estimator of the rate of stopping at red light around the world. I then would seek to measure factors that would effect people's behavior at stop lights: official efforts to educate drivers, deterrence, fear of crime (i.e., car jacking), state legitimacy, accident rates at stop lights, etc. This is how we "measure" human cognition (to use the quallies' phrase) with regard to stop lights. How would the quals propose I conduct such a study of cognition?

3/31/2008 1:35 PM  
Anonymous Anonymous said...

I have a question for ILP. Where does epidemiology fall on the Immutable Laws of Physics/unpredictable continuum (is it even a continuum, or are those mutually exclusive and exhaustive categories)?

3/31/2008 2:26 PM  
Anonymous Anonymous said...

RE: 1:35 p.m.

Thank you for your thoughtful post. I am in general agreement with you. It should be noted that no qual has denied the utility of quantoid methods. As such, for whatever it is worth, I think your proposed course of analysis is a valid one. The only point I would add is that in your study I see opportunities for qualitative research to augment and complements your proposed research on stopping at red lights.

3/31/2008 3:06 PM  
Anonymous Anonymous said...

RE: 2:26 p.m.

Certainly there are aspects to disease governed by the laws of physics. There are other aspects that are governed by human behavior, and its attending unpredictability. This is why governments will choose quarantines as a way to protect public health.

3/31/2008 3:11 PM  
Anonymous Anonymous said...

That's not a very helpful response. Are epidemiology studies scientific or not according to your view of science? If there is unpredictability (e.g., a drug might cure an illness for one person but not another for reasons that we can't explain), then it can't be scientific, can it?

3/31/2008 7:26 PM  
Anonymous Anonymous said...

I hope I don't screw up any comp sci grad students' dissertations by saying this, or cost anybody some research prize, but it's now quite clear that the ILP is an entrant into a Turing Test competition.

The ILP device, actually nothing but a computationally sophisticated version of the old ELIZA machine, applies a rule to every blog post that determines which of a very large number of possible responses it will provide, and according to a pseudo-random number generator, it occasionally just repeats itself. That is actually how I discovered the ruse; the ILP device repeats itself so regularly that I tracked its comments and discovered it to be running repetitions based on a garden variety linear congruential pseudo random number generator.

In fact the ILP device has almost passed what is sometimes called the ultra-strong form of the test: it can not only fool human interlocutors into thinking its human, but can make them beg it to stop talking.

Let's congratulate the designers of the ILP device for a job well done!

3/31/2008 9:22 PM  
Anonymous Anonymous said...

9:22pm p00wnd ILP

3/31/2008 11:22 PM  
Anonymous Anonymous said...

11:22 is a n00b. It's spelled p00ned. But she (who are we kidding, we know that 11:22 is male) is right:

ILP was p00ned by 9:22

3/31/2008 11:25 PM  
Anonymous Anonymous said...

"a drug might cure an illness for one person but not another for reasons that we can't explain"

Then, yes, there are aspects of that drug that are beyond science. In other words, we do not understand the causal mechanisms determining the success or failure of the drug. Of course, this does preclude the possibility that there is clear mechanism determining the success rate drug, and we do not comprehend it yet.

4/01/2008 5:48 AM  
Anonymous Anonymous said...

RE: 5:48 a.m.

That should read: "Of course, this does not preclude the possibility that there is a clear mechanism determining the success rate of the drug, and we do not comprehend it yet" -- my apologies.

4/01/2008 5:57 AM  
Anonymous Anonymous said...

Of course, this does not preclude the possibility that there is a clear mechanism determining the success rate of the drug, and we do not comprehend it yet

Fair enough: this is basically what many people in clinicial medicine think of acupuncture, and what many (industrial) chemists believe about certain catalysts.

Essentially, they argue that given that the components are (at the end of the day) just chemical/physcial there must be some explanation that we haven't located yet. Once we know this, we would know why certain drugs/techniques work sometimes (and with varying degrees of success) and not others.

But couldn't this be true of political science too? That is, the true mechanism is out there we just haven't pinned it down perfectly yet?

4/01/2008 6:19 AM  
Anonymous Anonymous said...

"But couldn't this be true of political science too? That is, the true mechanism is out there we just haven't pinned it down perfectly yet?"

We know precisely what drives (causes) human social/political behavior: cognition.

4/01/2008 6:49 AM  
Anonymous Anonymous said...

We know precisely what drives (causes) human social/political behavior: cognition.

That's a bit like saying that 'lack of oxygen to the brain' was why the dinosaurs died out.

I am aware that cognition matters (indeed, what else could matter?) but as others have pointed out, we are not perfect (i.e. error free) at predicting behavior.

If we want to be perfect (i.e. error free) then perhaps we have to do more research: locate the precise causal mechanisms etc.

And thus, in this regard, the status of political science is akin to new research frontiers in medicine or chemistry.

4/01/2008 7:08 AM  
Anonymous Anonymous said...

"If we want to be perfect (i.e. error free) then perhaps we have to do more research: locate the precise causal mechanisms etc."

As a point of logical, we cannot categorically reject the position you posit. Nonetheless, there is no empirical nor theoretical reason to hold that anything other than cognition causes human behavior.

4/01/2008 7:13 AM  
Anonymous Anonymous said...

Nonetheless, there is no empirical nor theoretical reason to hold that anything other than cognition causes human behavior.

I must be new here: who is claiming something other than thinking causes behavior?

If the answer to this is 'no one', why did you make this completely empty claim?

BTW, I love lamp.

4/01/2008 7:22 AM  
Anonymous Anonymous said...

RE: 7:22 a.m.

Some have been struggling for months on this blog to establish this point. I am not certain whose intelligence is more in question, those who resisted it or those who fought for it.

4/01/2008 9:56 AM  
Anonymous Anonymous said...

Some have been struggling for months on this blog to establish this point. I am not certain whose intelligence is more in question, those who resisted it or those who fought for it.

Well...

As I read it, the debate is over

(a) can thinking be 'measured' by social scientists? and if so,

(b) how is this 'best' done?, or

(c) if we can't measure it, what methods should we use to analyze human behavior?

I don't think the debate is as coarse as you suggest.

Certainly (as you suggest), that would not be very interesting.

So, (a), (b) and (c) are what matter.

4/01/2008 10:09 AM  
Anonymous Anonymous said...

We know precisely what drives (causes) human social/political behavior: cognition.

That "cognition" is the sole psychological process driving behavior would come as quite a shock to people employed by psych departments.

4/01/2008 2:41 PM  
Anonymous Anonymous said...

What do you think they propose as driving human behavior? The laws of social/political behavior?

4/01/2008 2:47 PM  
Anonymous Anonymous said...

I hate to break it to you, Sparky, but "cognition" is not universally acknowledged as a synonym for "mental activity."

When you tell somebody who actually has even basic knowledge of psychology that politics is entirely driven by human cognition, you're telling them that other mental activities, such as affect, play no role in politics.

4/01/2008 3:48 PM  
Anonymous Anonymous said...

"cognition" is not universally acknowledged as a synonym for "mental activity."

Yes, I'd wondered about this too. Many moons ago, I tried to press the point and find out why the poster didn't try using 'thinking about politics' or the like.

My sense (unfortunately) is that 'determining cognition' is simply a catch-phrase to make the whole endeavour sound more impressive/scientific.

It's rather obscurantist, frankly, and surprising coming from Perestroika sympathizers who I assumed were a bit skeptical of quant technical language that they consider abstruse.

Note: I'm not a regular quantoid poster hell-bent on bad mouthing the quals (on here), I had just never come across this term in my career.

I genuinely have no idea why the poster continues to use it: to my knowledge there is no well-known literature in which the term needs no clarification (unlike e.g. 'standard error').

4/01/2008 5:09 PM  
Anonymous Anonymous said...

"When you tell somebody who actually has even basic knowledge of psychology that politics is entirely driven by human cognition, you're telling them that other mental activities, such as affect, play no role in politics."

Even if someone is voting for Obama because of "affect" (i.e., an emotional affinity for him), they are still cognizant of the fact that they are voting for Obama. Thus, "political affect" is mediated through cognition.

Additionally, when people are probed/questioned about their political behavior they normally given seemingly rational/logical reasons for their behavior. Benjamin Page has a whole body of research displacing the notion of the "irrational" public -- finding that the public, as a whole, is quite politically rational/logical.

4/01/2008 6:43 PM  
Anonymous Anonymous said...

Ah yes, the "miracle of aggregation." Page's work is little more than wishful thinking and handwaving.

4/01/2008 7:27 PM  
Anonymous Anonymous said...

"Additionally, when people are probed/questioned about their political behavior they normally given seemingly rational/logical reasons for their behavior. Benjamin Page has a whole body of research displacing the notion of the "irrational" public -- finding that the public, as a whole, is quite politically rational/logical."
------------------
In the alternate universe that is this blog, space and time bend and the qaualy begins to toe the Rat Choice party line.

4/02/2008 3:51 AM  
Anonymous Anonymous said...

Rational behavior flows from ideology, political values, and perceived long-term interests. It rarely follows from narrow specific cost/benefit analysis.

4/02/2008 7:13 AM  
Anonymous Anonymous said...

"Rational behavior flows from ideology, political values, and perceived long-term interests. It rarely follows from narrow specific cost/benefit analysis."

What do you mean by "It rarely follows from narrow specific cost/benefit analysis"? Can you just please clarify this?

4/02/2008 9:19 AM  
Anonymous Anonymous said...

Rational choice theory proceeds from the view that individuals operate according to a cost/benefit ratio. This is very close to the thinking underlying mainstream economics.

4/02/2008 10:43 AM  
Anonymous Anonymous said...

Rational choice theory assumes people choose what they most prefer among what is available, and their preferences are internally consistent. Full stop.

4/02/2008 10:49 AM  
Anonymous Anonymous said...

I do not doubt your explanation. Nonetheless, I have seen rational choice work that seemingly discounted political philosophy (i.e., political values), and ostensibly emphasized materialist (i.e., economic) considerations.

"Rational choice theory assumes people choose what they most prefer among what is available, and their preferences are internally consistent."

I also have to add that your description of rational choice theory does not present this theory as holding much originality. In other words, this notion is embedded in virtually every variant of political analysis -- going back to the Ancient Greeks.

4/02/2008 10:59 AM  
Anonymous Anonymous said...

> In other words, this notion is embedded in virtually every variant of political analysis -- going back to the Ancient Greeks.

Indeed. So the strange thing is that anyone at all objects to it.

4/02/2008 11:30 AM  
Anonymous Anonymous said...

"I also have to add that your description of rational choice theory does not present this theory as holding much originality."

I really hope that one of these day someone will explain to me what "rational choice theory" is. I always thought that there were several different models that assumed the rationality postulate, and I wasn't aware that there was "one" such theory.

4/02/2008 1:33 PM  
Anonymous Anonymous said...

In my view, rational choice theory only makes sense if the claim is made by its proponents that individuals are primarily motivated by an economic cost/benefit notion. Hence, rationality is narrowly defined as to emphasize economic calculations. Otherwise, I know of no theory that explicitly holds that people behave irrationally.

4/02/2008 2:55 PM  
Anonymous Anonymous said...

Hence, rationality is narrowly defined as to emphasize economic calculations.

What do you mean by economic calculations? I hope you're not one of those people who think that an Economic Theory of Democracy is about dollars and cents!

4/02/2008 6:12 PM  
Anonymous Anonymous said...

> In my view, rational choice theory only makes sense if the claim is made by its proponents that individuals are primarily motivated by an economic cost/benefit notion. Hence, rationality is narrowly defined as to emphasize economic calculations. Otherwise, I know of no theory that explicitly holds that people behave irrationally.

Whether you apprehend it or not is irrelevant. A model of behavior or action may properly be called a rational choice model even if it involves no "economic cost/benefit calculation," or even just a cost/benefit calculation, or even a calculation. This is true, to wit, for models that assert (a) individuals choose what they prefer from an available set, and (b) their preferences are internally consistent.

There are actually myriad theories of human action that do not link choice with preference and thus deny a key component of rational choice theory. For example, theories of behavior as determined by habit or by social roles do not specify preference relations to predict, understand, or explain behavior.

Now, a rational choice theorist might be interested in showing that adherence to one's social role is in fact rational, which means only that it is the most preferred choice for an individual holding a specific preference relation that the theorist specifies. In that sense, role adherence is not necessarily -irrational-, and indeed, no one choice possibly can be -necessarily irrational. Put differently, all empirical content of rational choice theory as such is captured by the weak axiom of revealed preference (and even that requires the untestable assumption of constant preferences), a test of which requires more than one choice to be made. Nevertheless, social scientists who use role theory to understand behavior usually do not do so with respect to any preferences, and are thus not employing rational choice theory.

I look forward, 2:55, to your further unsubstantiated assertions that, to the contrary, rational choice theory really does require economic cost benefit calculations.

4/02/2008 9:08 PM  
Anonymous Anonymous said...

In my view, rational choice theory only makes sense if the claim is made by its proponents that individuals are primarily motivated by an economic cost/benefit notion. Hence, rationality is narrowly defined as to emphasize economic calculations. Otherwise, I know of no theory that explicitly holds that people behave irrationally.

4/02/2008 2:55 PM
------------------------
Again YOU do not get to set the definitions of things at will. This is NOT what Rat Choice is.

As has been said here, Rat Choice assumes that people make choices that they believe are the best for them and that their preferences are internally consistent.

Thats it.

I myself have thought you talk of cognition being the sources of human behavior is Rat Choice 101.

4/04/2008 5:39 AM  
Anonymous Anonymous said...

but the quantoids are almost inexplicably vehemently opposed to any criticism or limitation attached to their methods. I think their rigidness and dogmatism does a disservice to the quant project.

You clearly have never been to a Polmeth conference. There, you will see criticism and discussion of limitations of those methods. You will also find that there are very important divisions within the subfield.

I think that the opposition to your view that you see here is mostly due to the realization, by the quantoids, that the criticism is coming from someone who doesn't have much knowledge in the very thing he is criticizing. Thus, you are met with vehement opposition, because what you say simply doesn't make sense. And since training in statisitcs is lacking, you don't know that it doesn't make sense. And you're amazed at the reaction and try to fault the quants for it.

What is really funny is that if you only knew how little sense you make when you discuss statistics, you'd probably chuckle... I'll give it to you that you know something about qual methods, and to some extent philosphy of science, but you know next to nothing about statistics.

I'm not just talking about you making basic mistakes about p-values, correlations, etc., but rather about your repeated misunderstandings of fundamentals such as stochastic terms, hypothesis testing, probability, etc.

4/04/2008 2:44 PM  
Anonymous Anonymous said...

Certainly there are aspects to disease governed by the laws of physics. There are other aspects that are governed by human behavior, and its attending unpredictability. This is why governments will choose quarantines as a way to protect public health.

So tell me, have you ever taken prescription duges (or would you ever consider taking prescription drugs if your doctor recommended it)? If so, why? Since the outcome is unpredictable, aren't you just rolling the dice with your life if you take these drugs (given side effects, costs, etc.)?

I mean, if I were you, I'd say no to any sort of prescription drugs since, as you've stated, human behavior is unpredictable.

4/04/2008 2:49 PM  
Anonymous Anonymous said...

After months of denial, ILP has finally conceded that his point is untenable. Or maybe he's studying for a comp right now...

4/05/2008 11:34 AM  
Anonymous Anonymous said...

"As has been said here, Rat Choice assumes that people make choices that they believe are the best for them and that their preferences are internally consistent."

Does "people make choices that they believe are the best for them" include political values and/or ideology? Or, according to you, is rational choice a theory purely of self-interest? This notion of people acting solely for "self-interest" comes closer to my understanding of rational choice.

4/06/2008 1:58 PM  
Anonymous Anonymous said...

RE: 2:49 p.m.

I am uncertain what you mean by "unpredictable" in your example. Do you mean doctors will mis-prescribe medication? Alternatively, do you mean that the drug companies will manufacture faculty medication? Sadly, both are regular occurrences.

4/06/2008 2:01 PM  
Anonymous Anonymous said...

RE: 2:44 p.m.

One does not need to comprehend statistics to intelligently engage in a discussion on philosophy of science. Additionally, expertise in statistics does not provide any particular insight into philosophy of science issues.

4/06/2008 2:06 PM  
Anonymous Anonymous said...

Does "people make choices that they believe are the best for them" include political values and/or ideology?
-------------------
Of course it would.


This notion of people acting solely for "self-interest" comes closer to my understanding of rational choice.
--------------
Where does this "understanding" come from?

The definition that we have been giving you is the standard one I am aware of. Is your understanding based on actual study of Rat choice or simply you own biased assumptions?

4/06/2008 3:29 PM  
Anonymous Anonymous said...

RE: 3:29

Perhaps you could provide some instructive citations to further your point?

4/07/2008 6:51 AM  
Anonymous Anonymous said...

Perhaps you could provide some instructive citations to further your point?

4/07/2008 6:51 AM
-----------------------
Perhaps you could.

Amazing. I asked you where your understanding that Rat Choice precluded political values and ideology came from. You reply by asking me for citations. All the while you have refused to provide us with a single citation that supports your claim that qualitative methods are a better tool for exploring political phenomena owing to their ability to examine cognition.

Never the less, since I do have a few books laying round the ole' office that happen to lay out the definition of Rat Choice I will give you a few.

Try either of these, I often use one of these with undergrads or grad students who are getting their first exposure to "Rational Choice":

Shepsle and Bonchek "Analyzing Politics"; Chapter 2

Bianco "American Politics: Strategy and Choice"; Introduction

4/07/2008 9:21 AM  
Anonymous Anonymous said...

I am uncertain what you mean by "unpredictable" in your example.

What did YOU mean by unpredictability (since that was in YOUR response to the question about epidemiology).

I think you know exactly what I meant. There are potential benefits to taking drugs, but there are potential costs as well (side-effects, monetary costs, etc.).

Since the benefits are unpredictable (the 'correct' medicine might work for me, but not for you) and the costs are real (we have to purchase it, and there is a real potential for side effects), why would you ever accept to take prescription drugs?

The doctor only knows what works 'on average.' That, according to your logic, is a 'guess' (I'd say uneducated guess, but you stated that there is no such thing as an uneducated guess--they are all guesses). Why would you take a drug based on a guess?

The results are unpredictable, and could leave you worse off than without taking the drugs. Why would you ever accept to take prescription drugs?

4/07/2008 10:08 AM  
Anonymous Anonymous said...

RE: 10:08 a.m.

The posts your are referring to presumably describe human behavior as "unpredictable."

I still am uncertain what you are asserting is unpredictable. Is it the effect of pharmaceutical drugs? If so, to a degree you are correct.

4/07/2008 10:45 AM  
Anonymous Anonymous said...

Is it the effect of pharmaceutical drugs? If so, to a degree you are correct.

OK, so we agree on this. Then answer my question: would you ever consider taking prescription drugs even though you know that their effects are unpredictable (and hence, that there is a non-zero probability that the costs will outweigh the benefits)? If so, why?

4/07/2008 12:52 PM  
Anonymous Anonymous said...

They are not as unpredictable as you suggest.

4/07/2008 12:55 PM  
Anonymous Anonymous said...

What is your broader point? What do you intend with these questions?

4/07/2008 12:57 PM  
Anonymous Anonymous said...

One does not need to comprehend statistics to intelligently engage in a discussion on philosophy of science.

Now you're just being silly. What you forgot to mention was that this was a discussion on philosophy of science as it applies to statistics. So you've got two elements a)philosophy of science, and b)statistics, whose relationship we're trying to uncover with this discussion.

When qantoids show a lack in their knowledge of a), you point that out as a shortcoming in their position (do I really need to point you to the posts where you've chided them on that?). Why is it that when called on your lack of knowledge on b), you dismiss the criticism as nothing more than peripheral, and immediately point to the quants' lack of knoweldge in a) as your defense. Is that a defense mechanism? Or do you actually believe that to have an enlighten debate, one ought to know a), but is not required to know b).

I thought the debate pertained to the role of statistics in trying to uncover cognition. What you're saying is that one need not know much about statistics (or even cognition, as others have pointed out) to engage in that debate? As long as I have a modicum of knowledge of philosophy of science, that makes me well-equipped? Is that your position?

4/07/2008 12:57 PM  
Anonymous Anonymous said...

They are not as unpredictable as you suggest.

Oh, but earlier you said that things were either unpredictable (i.e., human behavior) or predictable (i.e., immutable laws). Now you're saying there are varying degrees of unpredictability? How do you know that? Can you measure it? How so?

I'm getting at the fact that you agree that there is an inherent unpredictability in the effects of drugs (since they depend on human behavior, which is unpredictable--and don't try to recant, you've already admitted to their inherent unpredictability).

Given this unpredictability, the only reason one would agree to use these drugs is if one believes in the law of averages and in generalizations. I mean, why else would you agree to use drugs that have been tested on other people? If we can't generalize from these studies, then how else would you take these results as being generalizable to all humans?

Either you can generalize or your can't, given uncertainty. You agree there is uncertainty. If you take the drugs, you admit to the ability to generalize under uncertainty.

I'm curious to know your position on this, and how you reconcile it with your earlier view that human behavior is unpredictable, and hence that we can't generalize from it (since what happens to one subject might not happen to another patient).

You know EXACTLY why I'm asking this. There's no way you have a Ph.D. and can't figure it out...

4/07/2008 1:02 PM  
Anonymous Anonymous said...

RE: 12:57 p.m.

The question animating this discussion is the goal of social sciences. In the physical sciences the goal is to uncover and analyze the laws of physics -- in the myriad ways that these laws manifest themselves and operate. As a result, the utility of quantitative methods in this context is rather patent.

In the social sciences the goal is an analysis of human cognition -- the prime cause of human behavior. The question then what is the utility of quantitative methods in the social sciences?

To engage in this discussion does not require an intricate understanding of quantitative methods. Indeed, as indicated by this discussion, perhaps a less than stellar understanding of these methods could be beneficial – as being vested in these methods could create opposition to their critical treatment.

4/07/2008 2:09 PM  
Anonymous Anonymous said...

RE: 1:02 p.m.

I trust you see that you are equating human behavior with the effects of pharmaceutical drugs.

4/07/2008 2:11 PM  
Anonymous Anonymous said...

perhaps a less than stellar understanding of these methods could be beneficial – as being vested in these methods could create opposition to their critical treatment.

A strange view: if you want to criticize something and be taken seriously, it surely pays to have some clue of what you are talking about.

the goal is an analysis of human cognition -- the prime cause of human behavior.

Do you have some kind of sponsorship deal with the word 'cognition'? Ten cents for every mention?

Or do you think it makes your argument sound more interesting/impressive/scientific?

It does none of these: it just seems pretentious, lazy and misplaced.

4/07/2008 2:26 PM  
Anonymous Anonymous said...

1:02 PM is equating uncertainty with uncertainty.

4/07/2008 2:35 PM  
Anonymous Anonymous said...

LOL!

That is really outrageous. ILP makes claims (just go back a few posts--you've made them) and then when asked about the implications of his own words, he pretends he doesn't understand.

That's enough for me to infer that he concedes the point. Why else would he attempt to evade the question? So ILP believes that human behavior is generalizable. At least now we agree on something.

See, ILP, it took all this time but you've finally conceded your point.

4/07/2008 4:38 PM  
Anonymous Anonymous said...

"A strange view: if you want to criticize something and be taken seriously, it surely pays to have some clue of what you are talking about."

Quals have more than a "clue" in terms of understanding quantitative methods. The germane issue is can we infer causation from patterns of behavior? Quantitative methods are effective in detecting patterns of human behavior -- to the extent that such patterns exist. More than that, for purposes of this discussion, does not need to be known about quantitative methods.

4/07/2008 4:46 PM  
Anonymous Anonymous said...

RE: 2:35 p.m.

That is the serious flaw in your logical. There are hugely different levels of uncertainty. There is the uncertainty of trying to win the lottery versus the uncertainty of taking blood pressure medication.

4/07/2008 4:49 PM  
Anonymous Anonymous said...

That is the serious flaw in your logical. There are hugely different levels of uncertainty. There is the uncertainty of trying to win the lottery versus the uncertainty of taking blood pressure medication.

My first question is: how do you know that? I can model the probability of winning the lottery. Epidemiologists can also model the uncertainty regarding drugs. But I can only do so using quantitative methods. How would you go about doing that?

3/30/2008 3:10 PM:
What does "best guess" mean? A guess is a guess, educated or otherwise. My claim that at sea level water will boil at very close to 212 degree F tomorrow and a year from now, is not a guess. It is a scientific fact.

Again, because human cognition is autonomous the best we have are guesses about future human behavior. We are uncertain how much confidence to put in them.


But you said that a guess is a guess. What I'm curious to know is that if unpredictability precludes scientific generalization (YOUR claim, not mine), then how can you generalize results from epidemiology studies?

Is your position the following: "there are varying degrees of uncertainty. when uncertainty is relatively small, we can generalize"? If not, then WHAT is your position?

Look, you either have to abide by your own words, or recant. You can't have it both ways. Can we generalize or not when there is uncertainty (i.e., when we're not dealing exclusively with 'immutable laws')?

4/07/2008 5:33 PM  
Anonymous Anonymous said...

"Can we generalize or not when there is uncertainty (i.e., when we're not dealing exclusively with 'immutable laws')?"

I am uncertain what you mean by "uncertainty" (no pun intended). Nonetheless, it is true that generalization requires "immutable laws."

4/07/2008 6:22 PM  
Anonymous Anonymous said...

Uncertainty is the degree to which an event (the blood pressure medication working correctly or winning the lottery) depends on chance. Sure, there are different degrees of uncertainty, but it's uncertainty nonetheless.

4/07/2008 8:55 PM  
Anonymous Anonymous said...

There is a vast difference between uncertainty and unpredictability. Human behavior is inherently unpredictable. The effect of a drug on an individual person is uncertain.

4/08/2008 6:41 AM  
Anonymous Anonymous said...

Human behavior is inherently unpredictable

You already conceded this point in the discussion about stop lights several weeks ago. Stop pretending that you forget.

4/08/2008 8:09 AM  
Anonymous Anonymous said...

Human behavior is inherently unpredictable.

Why do you use 'inherently'?

You think it could never, under any circumstances, be predicted?

Clearly you don't believe that (as your life would be very difficult if it was true).

Conclusion: stop misusing qualifiers.

4/08/2008 8:12 AM  
Anonymous Anonymous said...

"You already conceded this point in the discussion about stop lights several weeks ago. Stop pretending that you forget."

I may believe that most people will stop at red lights tomorrow, but that is not a prediction. I understand and fully acknowledge that they may not. Conversely, I will predict that at sea level water will boil very close to 212 degrees F tomorrow, and a thousand years from now.

4/08/2008 9:37 AM  
Anonymous Anonymous said...

"You think it could never, under any circumstances, be predicted?"

"Clearly you don't believe that (as your life would be very difficult if it was true)."

I am using predication in a scientific sense. Moreover, life is often very difficult due to unpredictable human actions. Could we predict that George W. Bush the presidential candidate as President would invade Iraq? He seemingly said otherwise during his 2000 campaign (i.e., "no state building"). On a more personal level there are questions of adultery, accidents, crime, etc. that cannot be predicted. The end result is that inferring future human behavior from past behavior is unreliable, and the process is unscientific.

4/08/2008 10:45 AM  
Anonymous Anonymous said...

"Could we predict that George W. Bush the presidential candidate as President would invade Iraq? He seemingly said otherwise during his 2000 campaign (i.e., "no state building")."

So what you are saying is that we cannot explain what people will do or why they did it based on their statements? That's a pretty interesting admission from the person that believes that we can only understand human behavior (cognition for you?) by asking people why they did something or what they will do.


"The end result is that inferring future human behavior from past behavior is unreliable, and the process is unscientific."

Who here is talking about predicting behavior from past behavior? Predictions come from our theories, which may or may not be about the linkage between past and future behavior. That being said, I really don't think that it is too much of a stretch to predict that someone that voted democrat in the past 5 elections will vote democrat in this election, do you? Likewise, I also don't think that it is too much of a stretch to predict that someone that has never run a red light in 20 years will also stop at that red light tomorrow. But then again, I guess I'm just not "scientific".

4/08/2008 12:58 PM  
Anonymous Anonymous said...

On a more personal level there are questions of adultery, accidents, crime, etc. that cannot be predicted.

You believe this, do you? The bit about crime and accidents?

Would you be willing to live in a 'high crime' area? Do you believe such a thing exists? Or is human behavior so unpredictable that there can be no such thing as a high crime area? I guess crime might be equal in Compton and Beverly Hills this year: there's no way of telling, is there?

And the accidents? Do you think car insurance is a scam? I mean the way they charge new drivers more than older ones? You must, because you don't think accident behavior is predictable.

4/08/2008 1:14 PM  
Anonymous Anonymous said...

"But then again, I guess I'm just not 'scientific.'"

Seemingly not.

4/08/2008 1:23 PM  
Anonymous Anonymous said...

"Would you be willing to live in a 'high crime' area? Do you believe such a thing exists? Or is human behavior so unpredictable that there can be no such thing as a high crime area? I guess crime might be equal in Compton and Beverly Hills this year: there's no way of telling, is there?"

"And the accidents? Do you think car insurance is a scam? I mean the way they charge new drivers more than older ones? You must, because you don't think accident behavior is predictable."

I am uncertain what you are arguing.

4/08/2008 1:24 PM  
Anonymous Anonymous said...

I am uncertain what you are arguing.

!!!!!!!!!!!
!! BINGO !!
!!!!!!!!!!!

Sorry, I got overexcited.

Point is this: you know 'high crime' areas exist and you know certain parts of the population are more likely to have car accidents. That is why we see differences in policing budgets for different areas, and why car insurance rates differ (respectively).

All of this is based on the fact that human behavior is predictable and generalizable: Compton is high crime this year, and will be high crime next year.

If you still don't understand this idea, I'd suggest moving to your nearest ghetto: housing is very cheap (for a reason that, I guess, you don't understand).

4/08/2008 1:37 PM  
Anonymous Anonymous said...

"But then again, I guess I'm just not 'scientific.'"

"Seemingly not."

Your ignorance is surpassed only by your arrogance. Instead of dealing with the stated point, you simply make a snide remark.

The point was that YOU are the one that is constantly stating over and over again that the only way we can understand human behavior is to ask them why they did something or what they will do. BUT, then you also say that we can not predict human behavior because Bush says one thing and does another. I really don't think that Bush is alone in this respect and that a number of people don't do things for their stated purposes. So, once again, can you tell me how YOUR approach (i.e., asking people) is more scientific than anyone else's?

----------------------------------
"I am uncertain what you are arguing."

If you do not have an answer, just say so. It gets really tiring to hear you hide behind your uncertainty. Again, if you weren't so arrogant, you would understand that not having an answer to every point is ok. Just say so.

4/08/2008 3:43 PM  
Anonymous Anonymous said...

I need a good book on HLMs. Any suggestions?

4/08/2008 6:19 PM  
Anonymous Anonymous said...

"So, once again, can you tell me how YOUR approach (i.e., asking people) is more scientific than anyone else's?"

Human behavior cannot be predicted. No methodology can overcome this.

"If you do not have an answer, just say so. It gets really tiring to hear you hide behind your uncertainty. Again, if you weren't so arrogant, you would understand that not having an answer to every point is ok. Just say so."

If I do not have an answer to something, I would acknowledge it. If I am uncertain as to what is being argued, I will state that as well.

4/08/2008 6:54 PM  
Anonymous Anonymous said...

but only qualitative research can uncover the causal pennings of high crime or insurance rates.

4/09/2008 12:54 AM  
Anonymous Anonymous said...

I need a good book on HLMs. Any suggestions?

"Data Analysis Using Regression and Multilevel/Hierarchical Models" Andrew Gelman and Jennifer Hill, Cambridge University Press, 2007

is excellent IMHO. Covers HLM and H non-L M. Lots of examples, and R and BUGs code too.

BTW, I'm not Gelman, not Hill and am not paid to endorse their product!

4/09/2008 5:13 AM  
Anonymous Anonymous said...

need a good book on HLMs. Any suggestions?

"Data Analysis Using Regression and Multilevel/Hierarchical Models" Andrew Gelman and Jennifer Hill, Cambridge University Press, 2007
-----------------------

ditto

great book!

4/09/2008 6:16 AM  
Anonymous Anonymous said...

"but only qualitative research can uncover the causal pennings of high crime or insurance rates."

What does "pennings" mean?

In any event, poverty and crime are weakly correlated. Indeed, the most economically damaging crime is conducted by wealthy white collar criminals -- as are ostensibly most public health related crimes (e.g., dangerous products, poor working conditions).

More importantly, the logic put forth by the quant poster reflects some of the laziness and seeming anti-intellectualism of quantoid work. The prime/sole concern is with correlation, but not with causation. Thus, the quantoid is not interested in what actually causes crime, but in only showing that certain demographic factors are correlated with being convicted of a crime. This does not rise above demography 101.

4/09/2008 7:08 AM  
Anonymous Anonymous said...

the logic put forth by the quant poster reflects some of the laziness and seeming anti-intellectualism of quantoid work. The prime/sole concern is with correlation, but not with causation.

This switching of the argument every other post is an intriguing rhetorical strategy, but I think it's been overused here and makes you appear a bit odd. The discussion of crime was a response to a question about predictability, not causality. The latter was not being discussed. You know this, and it frustrates and embarrases you, so you accuse the poster of mistaking correlation for causation.

Don't believe me?
See 4/08/2008 6:41 AM through to 4/08/2008 1:37 PM.

AYBABTU.

4/09/2008 7:24 AM  
Anonymous Anonymous said...

"Human behavior cannot be predicted. No methodology can overcome this."

So once again, why is your qualitative analysis superior?

4/09/2008 7:34 AM  
Anonymous Anonymous said...

"The discussion of crime was a response to a question about predictability, not causality."

How can causality and predictability be separated?

4/09/2008 7:52 AM  
Anonymous Anonymous said...

"So once again, why is your qualitative analysis superior?"

On questions of predictability I am not saying that qualitative methods are.

4/09/2008 7:55 AM  
Anonymous Anonymous said...

How can causality and predictability be separated?

I'm not sure if this a serious question. I mean, this is one the fundamental social science research problems.

You can predict things (sometimes very accurately) without knowing anything about what caused them: this is why people worry about correlation/causation distinctions.

Can I predict Baltimore's murder rate for 2009? Sure: I'll use this year's figure (or take a mean over the last five years), and I'll be reasonably accurate. Can I tell you what causes that murder rate? Maybe I could have a guess, but that's a very different question...

4/09/2008 7:58 AM  
Anonymous Anonymous said...

"Can I predict Baltimore's murder rate for 2009? Sure: I'll use this year's figure (or take a mean over the last five years), and I'll be reasonably accurate."

I do not see why you or anyone else would place any value in your 2009 prediction?

4/09/2008 8:12 AM  
Anonymous Anonymous said...

The 2009 prediction amounts to little more than a guess. It does not even amount to a prediction. This does reflect lazy thinking.

4/09/2008 8:14 AM  
Anonymous Anonymous said...

The 2009 prediction amounts to little more than a guess. It does not even amount to a prediction. This does reflect lazy thinking.

Again, I see the same rhetorical strategy: you change the argument, and accuse the poster of laziness.

Predicting values at period "t" by using period "t-1" is not especially unusual. In fact there is a whole econometric literature on this idea: "lagged dependent variables"

But let's suppose it isn't a good prediction of the Baltimore murder rate: the original point (to which the post responded) was that garnering a prediction is very different to determining causation, and this was an example of the difference. Again, correlation is not causation as several qual posters are wont to note.

Do you genuinely not follow the arguments on here? I'd recommend you read more and post less if so.

4/09/2008 8:21 AM  
Anonymous Anonymous said...

"Predicting values at period 't' by using period 't-1' is not especially unusual. In fact there is a whole econometric literature on this idea: 'lagged dependent variables'."

If this involves predicting the future, it is junk science.

"But let's suppose it isn't a good prediction of the Baltimore murder rate: the original point (to which the post responded) was that garnering a prediction is very different to determining causation, and this was an example of the difference."

Without determining causation, predictions are nothing more than guesses.

"I'd recommend you read more and post less if so."

Simply because something makes into print (even in a so-called leading journal) does not make it valid. To believe so is the worst kind of conformism and intellectual laziness.

4/09/2008 8:28 AM  
Anonymous Anonymous said...

Without determining causation, predictions are nothing more than guesses.

You are seriously claiming that one must specify causal mechanisms to have a 'good' prediction (where 'good' is measured by difference between prediction and reality)?

Suppose the mayor of Baltimore asked you what the murder rate in his city would be this year: how would you advise him?

Anyway, your general assertion will come as quite a surprise to anyone interested in genetic research: there, predictive techniques are used regularly to try to isolate genes that seem to be associated with various disorders. We often have no idea about how causation works (in terms of the mechanism).

And can we please avoid the "well that's a physical science, and social science has no immutable laws" crap? The point is the distinction between prediction and causation.

4/09/2008 8:46 AM  
Anonymous Anonymous said...

OK. Now let me lead you through your logic to show just how inconsistent you are.

You just said (4/09/2008 8:28 AM):
"Without determining causation, predictions are nothing more than guesses."

So apparently, if we can determine causation, then we can predict. Right?

You have also said over and over again (too many places to cite, but most recently 4/09/2008 12:54 AM) that the only way we can determine causation is through the use of qualitative methods (i.e., to ask people). So apparently, your use of qualitative methods to determine causation can allow you to predict, right?

Then you said (4/08/2008 6:54 PM):
"Human behavior cannot be predicted. No methodology can overcome this."

So let me try to get your position straight about this. You are saying:
If we can determine causality, then we can make predictions. Your qual. methods allow you to determine causality. You should therefore be able to predict. But you say that you cannot predict.

But why can't you predict. Simple logic would lead us to believe that YOU could predict because you can determine causality through your qualitative methods. So, the only reason that you shouldn't be able to predict is if your qual. methods do not indeed allow us to determine causality.

Have you been feeding us a line of bs this whole time?

Say it ain't so ILP, say it ain't so!

4/09/2008 8:50 AM  
Anonymous Anonymous said...

"And can we please avoid the 'well that's a physical science, and social science has no immutable laws' crap? The point is the distinction between prediction and causation."

No, the point is that with immutable laws reliable predictions can made be.

"Suppose the mayor of Baltimore asked you what the murder rate in his city would be this year: how would you advise him?"

I would seek to identify the causes of crime in the City, and relate those causes to the "predicted" future crime rate. I would not say, "Well this was the crime over the last five, and based on this I predict the following crime rate." This comes close to being deterministic.

4/09/2008 9:00 AM  
Anonymous Anonymous said...

One general point is that due to the autonomy of human cognition human behavior cannot be scientifically predicted. This is in sharp contrast to the behavior of matter and energy.

Now, can we make "educated" guesses about human behavior? Yes. The best educated guesses tend to be made with the use of qualitative methods. Let us take the example of Hitler. There were observers that made accurate guesses about Hitler's rule (before he ever took office) based on his rhetoric and books. Thus, these accurate guessers familiarized themselves with Hitler's cognition (among other factors), and were able to divine his actions in office. Such accurate guesses could not have been made with quantitative methods.

4/09/2008 9:16 AM  
Anonymous Anonymous said...

"Let us take the example of Hitler. There were observers that made accurate guesses about Hitler's rule (before he ever took office) based on his rhetoric and books. Thus, these accurate guessers familiarized themselves with Hitler's cognition (among other factors), and were able to divine his actions in office."

Or I could look at his previous patterns of international behavior leading up to Munich to make an "educated guess" about what he would do after Munich, right? This would actually yield a very accurate guess about how he would press for further acquisitions in Czechoslovakia and eventually Poland, wouldn't it?

4/09/2008 9:20 AM  
Anonymous Anonymous said...

"I could look at his previous patterns of international behavior leading up to Munich to make an 'educated guess' about what he would do after Munich, right?"

What pattern are you speaking of?

4/09/2008 9:24 AM  
Anonymous Anonymous said...

"I could look at his previous patterns of international behavior leading up to Munich to make an 'educated guess' about what he would do after Munich, right?"

"What pattern are you speaking of?"

Lets see. The remilitarization of the rhineland, the anschluss, his continued pressure for more and more concessions and divisions of czechoslovakia before and during munich...

All of these things showed a pattern of behavior that could be used to predict Hitler's later behavior couldn't they? Even without knowing his "cognition", personality defects, etc.

4/09/2008 9:41 AM  
Anonymous Anonymous said...

BTW: I'm still waiting for you to address the point in 4/09/2008 8:50 AM.

4/09/2008 9:42 AM  
Anonymous Anonymous said...

"One general point is that due to the autonomy of human cognition human behavior cannot be scientifically predicted...Now, can we make "educated" guesses about human behavior? Yes."

So predictions have to be correct 100% of the time in order to be scientific? Otherwise, they are educated guesses?

4/09/2008 9:45 AM  
Anonymous Anonymous said...

"So predictions have to be correct 100% of the time in order to be scientific? Otherwise, they are educated guesses?"

No, but predictions have to be base on finite bases (causes). With human cognition explicitly unpredictable, that means that human behavior is explicitly unpredictable.

4/09/2008 12:08 PM  
Anonymous Anonymous said...

"The remilitarization of the rhineland, the anschluss, his continued pressure for more and more concessions and divisions of czechoslovakia before and during munich..."

I do not think any of these actions would necessarily lead one to think that Nazi Germany would invade Poland, France, the Soviet Union, etc. Indeed, many in Washington, London, and Paris thought that Hitler would end his expansionist policies with Czechoslovakia.

In any event, based on studying Hitler's cognition strong and clear "predictions" were being made that if he were to attain power the Nazis would crush domestic opposition (especially on the Left), and a German invasion of the Soviet Union would occur. Thus, using the method you propose "guesses" about Nazi foreign policy could be put forth only after Hitler took power, whereas qualitative methods did lead to accurate (and well grounded) guesses about the behavior of Germany under the Nazis.

4/09/2008 12:18 PM  
Anonymous Anonymous said...

The answer to 8:50 is 9:16. I apologize for not making that clear.

4/09/2008 12:26 PM  
Anonymous Anonymous said...

I would seek to identify the causes of crime in the City, and relate those causes to the "predicted" future crime rate.

How would you do this? Quant methods I assume?

In which case you either believe (a) quant methods can be used to make causal statements, (b) or you think they can be used for prediction.

If you believe (a), then you basically concur with the quants on this board.

If you believe (b) but not (a), you apparantly understand the distinction between prediction and causation. In which case, we are 'done' with this particular point.

Please clarify.

4/09/2008 12:47 PM  
Anonymous Anonymous said...

RE: 12:47

I would not run a regression analysis. You conflate counting (e.g., the unemployed) with quantitative methods. Moreover, in terms of gaining a grasp of future crime in any city, I would talk to community leaders to try to "grasp" the social fabric. In my view, such qualitative measures can allow one to make an "educated" guess about the direction the crime rate is going. It is this qualitative approach that underlies the theory of community policing.

4/09/2008 1:01 PM  
Anonymous Anonymous said...

In my view, such qualitative measures can allow one to make an "educated" guess about the direction the crime rate is going.

Really? That's what you would say to/do for the Mayor when he asks about budget/policing/legal staff requirements etc?

What if Margaret Spellings asked you about the US's education needs for 2018 (planning ten years in advance is pretty usual in policy circles)? What methods would you use to advise her?

4/09/2008 1:12 PM  
Anonymous Anonymous said...

RE: 1:12 p.m.

Outside of demographics, and setting education goals, what would planning for education involve?

4/09/2008 1:36 PM  
Anonymous Anonymous said...

ILP, did you write 4/09/2008 9:20 AM?

If so, can you explain the difference between an 'educated guess' and a 'guess'? In an earlier post (I could find it but it's a few weeks ago), you said that there were no such thing as educated guesses. They were all guesses (I think you brought up someone predicting that the Cowboys would win the Super Bowl as a guess).

Are you now saying that there are 'educated' guesses? I thought we couldn't model uncertainty in political science. Can we? How do we know whether our guesses are educated or not? Do you have some sort of criteria in mind?

Your about-faces are really interesting to follow. Boy wouldn't I love to take a look at your cognition...

4/09/2008 2:49 PM  
Anonymous Anonymous said...

"I do not think any of these actions would necessarily lead one to think that Nazi Germany would invade Poland, France, the Soviet Union, etc. Indeed, many in Washington, London, and Paris thought that Hitler would end his expansionist policies with Czechoslovakia."

Now you are starting to show that you are lacking in historical knowledge as well. After some early histories of WWII and Munich (e.g., Taylor), much of the more recent historical studies reveal that the major players (including Chamberlain) had few doubts about Hitler's ambitions. If you want some citations here, start with Kissinger (1994) then I can give you some more.

4/09/2008 3:33 PM  
Anonymous Anonymous said...

Outside of demographics

So, can we predict education demands based on quant methods (i.e. quant analysis of demographic data)?

4/09/2008 3:42 PM  
Anonymous Anonymous said...

"In an earlier post (I could find it but it's a few weeks ago), you said that there were no such thing as educated guesses. They were all guesses (I think you brought up someone predicting that the Cowboys would win the Super Bowl as a guess)."

From the perspective of science that is true.

4/09/2008 6:21 PM  
Anonymous Anonymous said...

"Now you are starting to show that you are lacking in historical knowledge as well. After some early histories of WWII and Munich (e.g., Taylor), much of the more recent historical studies reveal that the major players (including Chamberlain) had few doubts about Hitler's ambitions. If you want some citations here, start with Kissinger (1994) then I can give you some more."

I did not make any claims about specific individuals.

4/09/2008 6:22 PM  
Anonymous Anonymous said...

"So, can we predict education demands based on quant methods (i.e. quant analysis of demographic data)?"

"Educated guesses" about future population trends can be made.

4/09/2008 6:36 PM  
Anonymous Anonymous said...

"Educated guesses" about future population trends can be made.

Just so I understand then, for you a prediction is essentially a 'modus ponens' like statement?

Several comments if so:

1) that is not the way 'prediction' is commonly used in social science (and thus not the way it is used by US govt agencies interested in popn trends etc)

2) qual methods in the social sciences cannot offer predictions either---and hence I don't follow how one could predict policing demands based on a case study/speaking to a community

3) there is still a difference between prediction and causal mechanism description. I predict that the sun will rise in the east tomorrow (I will be 100% correct for all time), but I need have no good/sensible causal story as to why it will occur.

Thus, I am still at a loss to understand why you refuse to see the difference between prediction and causation. Another example: I predict an object when dropped will fall towards Earth, but I (and actually no one else either) have any notion of what the micro mechanism is (gravitons etc).

4/10/2008 6:01 AM  
Anonymous Anonymous said...

"there is still a difference between prediction and causal mechanism description. I predict that the sun will rise in the east tomorrow (I will be 100% correct for all time), but I need have no good/sensible causal story as to why it will occur."

"Thus, I am still at a loss to understand why you refuse to see the difference between prediction and causation. Another example: I predict an object when dropped will fall towards Earth, but I (and actually no one else either) have any notion of what the micro mechanism is (gravitons etc)."

Even if we do not understand a causal mechanism in the cases you invoke, we know they exist and prompt regular and predictable behavior with regard to energy and matter. Predictions are reliable made on this basis.

"Qual methods in the social sciences cannot offer predictions either---and hence I don't follow how one could predict policing demands based on a case study/speaking to a community."

The first point has been acknowledged. As to your second point, the presumption is that community leaders "know" the people in the community, and on that basis reasonable guesses can be made as to crime trends.

4/10/2008 6:49 AM  
Anonymous Anonymous said...

Even if we do not understand a causal mechanism in the cases you invoke, we know they exist and prompt regular and predictable behavior with regard to energy and matter. Predictions are reliable made on this basis.

I'm not disputing whether or not the predictions are reliable. Thank you for your clarification here: we are in agreement that prediction is different to causation. I had previously believed this was disputed (see 4/09/2008 7:52 AM)

We are getting somewhere -- good.

Now, my next question: can we logically have a causal story without a prediction?

That is, if we claim a prediction requires a modus ponens type relationship, what can it mean to have an account of causation that cannot produce such a predictive relationship?

My general point is this: you claim that neither quant nor qual methods can predict in the scientific sense, yet you believe they (or perhaps just qual) can explain in the scientific sense?

This strikes me as odd, and counter to standard scientific accounts: typically, properly and fully specifying the micro-mechanism is a sufficient condition for making a prediction (though, as you agree, it is not a necessary condition).

Making the prediction is not a sufficient condition for specifying the causal mechanism.

4/10/2008 7:04 AM  
Anonymous Anonymous said...

"I did not make any claims about specific individuals."

So you say that the only way that we could have made an accurate "educated guess" about Hitler's future behavior is through studying his cognition. I pointed to how leaders drew accurate "educated guesses" based on the consistency of his past behavior (not only through understanding his "cognition"). And your response is:

"I did not make any claims about specific individuals."

How does this response bolster your position or invalidate mine? I don't get it. Is using Hitler's past behavior leading up to Munich an effective way of estimating his future behavior or not?

4/10/2008 9:20 AM  
Anonymous Anonymous said...

RE: 7:04 a.m.

You are correct that some predictions are made irrespective of our understanding of causation. I think there are two relevant and salient points to be made in light of this observation. Firstly, the phenomena you point to (the rising sun and the effect of gravity) can be predicted sans an understanding of causation because of the operation of the laws of physics. (In other words, they work with such regularity that we can predict their effect without fully understanding them.) Secondly, the absence of such laws driving human behavior makes understanding causation especially important in terms of "guessing" about future human actions.

4/10/2008 10:10 AM  
Anonymous Anonymous said...

RE: 9:20 a.m.

Could a prediction of Germany's invasion of the Soviet Union in 1941 be derived from Germany's actions in 1938? I would argue no. The invasion of Poland in 1939 could reasonably be foreseen because Hitler up to 1938 was ostensibly targeting countries with substantial German speaking populations, and enclosed within Poland was actual official German territory.

Of course, this conclusion could only be derived from qualitative methods, since Hitler was making claims on Austria, Czechoslovakia, and Poland based on their containing German speaking populations. In short, I am uncertain of what can reasonably be concluded solely from Germany's invasion of the three countries in question. Similarly, are observers concerned about a possible U.S. attack on Iran because the U.S. invaded both Iraq and Afghanistan, or because of the verbal threats made by the Bush Administration against Iran?

4/10/2008 10:22 AM  
Anonymous Anonymous said...

"Of course, this conclusion could only be derived from qualitative methods, since Hitler was making claims on Austria, Czechoslovakia, and Poland based on their containing German speaking populations. In short, I am uncertain of what can reasonably be concluded solely from Germany's invasion of the three countries in question."

His initial demands were in terms of territories containing german speaking peoples. Things changed very quickly in later negotiations. Again, this is standard history. Do you deny that his actions showed a desire to dominate central/eastern europe.

"Similarly, are observers concerned about a possible U.S. attack on Iran because the U.S. invaded both Iraq and Afghanistan, or because of the verbal threats made by the Bush Administration against Iran?"

Have you been keeping up with the news over the past year? Isn't this exactly what people were worried about and why the most recent national security estimate was so important--proving that there is no reason for an invasion of Iran?

4/10/2008 11:41 AM  
Anonymous Anonymous said...

"Do you deny that his [Hitler's] actions showed a desire to dominate central/eastern europe."

Just looking at Germany's actions leading into 1939, one could not seemingly conclude that Hitler planned on "dominating" Eastern Europe. By this I mean it was not clear that Germany would invade the Soviet Union. Certainly policymakers in Moscow did not foresee this. They were caught completely by surprise by the 1941 invasion.

4/10/2008 11:59 AM  
Anonymous Anonymous said...

Firstly, the phenomena you point to (the rising sun and the effect of gravity) can be predicted sans an understanding of causation

Good, we agree on this part.

because of the operation of the laws of physics.

this is strange phrasing: 'laws' of physics are what, ultimately, allow us to predict anything (including biological processess etc). But, I don't want to labor the point, so...ok.

the absence of such laws driving human behavior makes understanding causation especially important in terms of "guessing" about future human actions.

I don't understand the logic here: strictly speaking (as we both agree) prediction (or "guessing", whatever the terminology) does not require understanding causation.

I sense you are claiming that understanding causation generally helps us to build better 'models' (these need not be formal, of course) of behavior. This is true, but the way we know how 'good' those models are is via their predictive (just bear with me) ability. That is, how well they 'fit' the world we see around us. Notice I'm not necessarily referring to statistical goodness-of-fit tests.

All this leaves me puzzled: you want better causal 'models', because they help us predict (and this makes perfect sense to me), but how do we compare the models to see which is best? I have some ideas, but I'd like to hear your views.

4/10/2008 12:25 PM  
Anonymous Anonymous said...

"All this leaves me puzzled: you want better causal 'models', because they help us predict (and this makes perfect sense to me), but how do we compare the models to see which is best? I have some ideas, but I'd like to hear your views."

What do you mean when you ask "how do we compare the models to see which is best?"

4/11/2008 6:05 AM  
Anonymous Anonymous said...

What do you mean when you ask "how do we compare the models to see which is best?"

Best at 'predicting' (or making educated guesses, whatever the terminology). Some models (quant, qual, whatever) will produce better (more accurate) guesses than others.

I am wondering how we compare one model/account to another in this regard.

4/11/2008 6:14 AM  
Anonymous Anonymous said...

I am still somewhat unclear what you mean by "compare." Are you asking for a general metric that can be used to assess the effectiveness of differing research to "predict" the future?

In any event, in my view qualitative approaches can more aptly yield "educated" guesses about future human behavior. By examining human cognition one can offer well-grounded supposition about how an actor(s) well act in the immediate future.

4/11/2008 6:20 AM  
Anonymous Anonymous said...

in my view qualitative approaches can more aptly yield "educated" guesses about future human behavior.

OK, that's fine, but given two (or more) qualitative accounts, how would you decide which is better (in terms of the educated guesses)?

Think of it in policy terms: you have two accounts of the community, by two different people, and you need to advise the Mayor about some policing/budget aspect. How would you decide which account is better (let's assume they suggest opposite things)

4/11/2008 6:24 AM  
Anonymous Anonymous said...

"Think of it in policy terms: you have two accounts of the community, by two different people, and you need to advise the Mayor about some policing/budget aspect. How would you decide which account is better (let's assume they suggest opposite things)"

Can you please give me more dimensions to your example?

Also, I am embarrassed to ask, but how do you italicize?

4/11/2008 6:46 AM  
Anonymous Anonymous said...

Can you please give me more dimensions to your example?

Just think of things this way: you are asked by a policy maker for advice on policing needs for 2009. You have two qualitative studies by two different people on which to base your report (suppose, as is often the case, there is no time to do your own research).

Now, how would you judge which of these reports is 'better'? In this case, better will mean 'which of these reports is more accurate in terms of its predictions about the community'?

Can this be done?


Also, I am embarrassed to ask, but how do you italicize?

Use HTML tags: you will need a little "i" in angle < > brackets before the words to be italicized and then "/i" in angle < > brackets after the words to be italicized.

(Obviously, you don't need the quote marks, just i or /i in angle brackets)

4/11/2008 6:54 AM  
Anonymous Anonymous said...

Thanks!

To answer your question let me give some specificity to the hypothetical studies you refer to. One study shows that during the last two economic downturns crime went up in the city by "x" and "y" percent. From this it is deduced that crime will similarly go up.

A second study, relying on interviews with community leaders, find that families, community organizations (e.g., religious groups), homeless shelters, informal welfare leaders (i.e., patronage providers), etc. learned from the last two economic downturns experienced by the city, and, as a result, are better prepared to offset the affects of the current economic downturn. Thus, crime in this hypothetical city will not spike up like during the last two recessions.

Again, these studies are just making educated guesses. Nonetheless, the second one does give us a more thorough understanding of the potential effects of an economic downturn, and a better basis to make a "prediction" about future crime rates.

4/11/2008 7:18 AM  
Anonymous Anonymous said...

Nonetheless, the second one does give us a more thorough understanding of the potential effects of an economic downturn, and a better basis to make a "prediction" about future crime rates.

Suppose you had two studies done in the same qualitative way as the second one you mention: with interviews etc.

The thing is, they come to different conclusions: one says they are better prepared, one says they are worse prepared.

How would we judge which of those studies is the one we should 'believe' or 'trust' more: this matters, because we are about to make budget decisions based on the findings.

4/11/2008 7:25 AM  
Anonymous Anonymous said...

How would we judge which of those studies is the one we should 'believe' or 'trust' more: this matters, because we are about to make budget decisions based on the findings.

My first response is that if there is an economic downturn more money should be spent on strengthening public services and public welfare -- as public need will be greater. An unwillingness to increase spending during an economic city-wide crisis will in-and-of-itself increase alienation, exacerbate social/political breakdown, and serve to push up crime. (What I am saying is that the political statement of not increasing public spending will itself have an adverse political/social effect.)

As to which of the two qualitative studies the Mayor would find more persuasive, he/she would (should) "know" which community leaders have the "pulse" of the city. Whichever study drew more heavily on those leaders who the Mayor thinks are "in the know" would would more likely persuade the Mayor.

4/11/2008 7:47 AM  
Anonymous Anonymous said...

As to which of the two qualitative studies the Mayor would find more persuasive, he/she would (should) "know" which community leaders have the "pulse" of the city.

The point is though, that the Mayor didn't do the studies himself (I assure you, Mayors never do!). So he is relying on accounts he is reading: which should he trust more, given both authors spoke to the same people?

My point here is: how can we tell which of two qualitative studies is 'better'. That is, how do we know which describes the world more accurately?

This isn't a trick or a trap, and I'm getting a bit frustrated I can't get a straight answer.

4/11/2008 8:01 AM  
Anonymous Anonymous said...

This isn't a trick or a trap, and I'm getting a bit frustrated I can't get a straight answer.

Part of the problem here is we are describing future predictions. As I have argued, these are inherently guesses.

Moreover, you are basically asking which of two hypothetical studies of a hypothetical city is more reliable in terms of guessing about the future of this made-up city. Again, there are so few dimensions to this scenario that I think you can see it is impossible to draw any definite conclusions.

4/11/2008 8:12 AM  
Anonymous Anonymous said...

Moreover, you are basically asking which of two hypothetical studies of a hypothetical city is more reliable in terms of guessing about the future of this made-up city.

No, I'm asking how we objectively judge which of two qualitative studies is better. It's a simple and reasonable question, and I think I've been pretty patient in trying to explain it (and why it matters).

The answer is, apparantly, that you "don't know".

Fine. There are lots of thing I don't know. It is nothing to do with whether or not the City is hypothetical or whether or not sufficiently numerous 'dimensions' have been established. Personally, I suspect there are no objective criteria and can never be.

That for me, is why qualitative studies are problematic: not because of the techniques they involve, whether they 'determine cognition' or whatever the phrase is, but because we cannot assess how good a qualitative study is in any reasonable, generally accepted way.

This is both absolute (we cannot say whether a study is 'good' or 'bad') and relative (we cannot say if one study is better or worse than another).

If you believe that we are generally trying to accumulate knowledge about cognition (or whatever) then this should strike you as troubling: if we have no criteria to compare studies, then we certainly have no way to accumulate knowledge in the discipline.

If you don't believe some or any of the above, feel free to object, but please avoid shirking questions or feigning misunderstandings/confusion when you simply don't have a (good) answer. It isn't helpful for debate on this board or, in fact, in the discipline as a whole.

To be honest, I've noticed that debate here is starting to dry up: I think that is because several posters have reached my situation: they ask questions, they give examples, and they just can't get straight answers so...what's the point. Prove me wrong.

4/11/2008 8:38 AM  
Anonymous Anonymous said...

If you don't believe some or any of the above, feel free to object, but please avoid shirking questions or feigning misunderstandings/confusion when you simply don't have a (good) answer.

I have never done that.

That for me, is why qualitative studies are problematic: not because of the techniques they involve, whether they 'determine cognition' or whatever the phrase is, but because we cannot assess how good a qualitative study is in any reasonable, generally accepted way.

This is both absolute (we cannot say whether a study is 'good' or 'bad') and relative (we cannot say if one study is better or worse than another).

The difficulty is not with qualitative methods, per se.
The "problem" is with reality, and the reality of causation underlying human behavior. As has been argued earlier, the best we can do is offer competing analysis of the motivations underlying human behavior. Is there an objective way to decide between these competing analysis? Unfortunately, there is not.

A prime critique made by the Perestroika movement of quant analytical methods is that these methods are not reflective of actual social/political phenomena, and the finite quality of conclusions derived from these methods are baseless -- and patently so. Alternatively, the conclusions are trivial. This is what leads some to conclude that for too many quantitative scholars the methods take priority over actually analyzing empirical reality.

If you believe that we are generally trying to accumulate knowledge about cognition (or whatever) then this should strike you as troubling: if we have no criteria to compare studies, then we certainly have no way to accumulate knowledge in the discipline.

As written earlier, the competing explanations themselves become "knowledge" and their accumulation gives us "insight" into other social/political phenomena. Thus, the claim is that competing analysis of the formulation of the Social Security Act of 1935 (for example) has lent meaningful "insight" into the formulation of other major pieces of U.S. legislation.

4/11/2008 8:59 AM  
Anonymous Anonymous said...

To be honest, I've noticed that debate here is starting to dry up: I think that is because several posters have reached my situation: they ask questions, they give examples, and they just can't get straight answers so...what's the point.

I have a different interpretation of the waxing and waning of this blog. Firstly, it has been going on for about a year -- more if you count how this discussion was initiated on the job rumors blogs. So over the the course of this long time participants are going to drop off. Secondly, this conversation has been rather intense -- with close to 1500 posts over the last three iterations of the methods/theory blog (including this one). With this many posts, issues have been exhausted and even recycled, and this is going to lead to drop off.

I will also add that I believe this discussion has lasted as long as it has precisely the qual side has sought to offer thoughtful and full-faithed answers and arguments. Otherwise, this conversation would have died virtually at inception.

4/11/2008 9:24 AM  
Anonymous Anonymous said...

I believe this discussion has lasted as long as it has precisely the qual side has sought to offer thoughtful and full-faithed answers and arguments

No matter how many times you repeat it, "HUMAN COGNITION IS UNPREDICTABLE AND I AM DIRECTLY MEASURING HUMAN COGNITION AND YOU ARE NOT" doesn't constitute thoughtful or full-faithed answers or arguments.

4/11/2008 12:59 PM  
Anonymous Anonymous said...

the finite quality of conclusions derived from these methods are baseless -- and patently so.

What do you mean by 'finite quality of conclusions'? I'm a quantoid and have never heard of this term. Maybe you can help enlighten me because I have seriously no clue what you're talking about.

I hope we're not going back to the uncertainty vs. certainty of quantitative results debate we had months ago. As a reminder: there is no definite conclusions that can be made based on quantitative studies. Any econometrics textbook will tell you that.

So again, what do you mean by finite quality? Is that a term of art that is used which I'm unaware of?

4/11/2008 2:50 PM  
Anonymous Anonymous said...

RE: 2:50 p.m.

What I am speaking to is the idea that we can "reject" a specific analysis of social/political phenomena through quantitative analytical methods.

4/11/2008 7:12 PM  
Anonymous Anonymous said...

HUMAN COGNITION IS UNPREDICTABLE AND I AM DIRECTLY MEASURING HUMAN COGNITION AND YOU ARE NOT

What is the disagreement with this formulation?

4/12/2008 7:19 AM  
Anonymous Anonymous said...

the idea that we can "reject" a specific analysis of social/political phenomena through quantitative analytical methods.

Argh! Why do you such odd and pretentious verbiage? It doesn't make people take you any more seriously, or make you appear more intelligent. If you want to get technical, the quants can start talking about "second order efficiency" and the like, and soon everyone else will be lost.

What is a "specific analysis"? A case study? An interview? Qual methods in general? A hypothesis? Who is talking about 'rejecting' a case study?

I ask because 'reject' means something very specific (connected to Neyman hypothesis testing) in social science.

I would actually like to understand what you are arguing, but your bizarre and lazy choice of expression makes this impossible.

4/12/2008 7:26 AM  
Anonymous Anonymous said...

HUMAN COGNITION IS UNPREDICTABLE AND I AM DIRECTLY MEASURING HUMAN COGNITION AND YOU ARE NOT

What is the disagreement with this formulation?


I suspect some of the following might elucidate people's irritation:

1) 'cognition' is a silly, pretentious term in this context, which lacks the rigorous definition we see in neuroscience/psychology. It gives a sheen of scientific legitimacy to the qual project which many believe is lacking

2) 'unpredictable', as defined on this board by quals essentially connotes some lack of a modens ponens type relationship. That is, it used in the strict sense that it is used in physics. That would be fine, if it wasn't for the inconsistency: 'cognition' is used in a completely unscientific way.

3) outside of neuroscience, cognition cannot be measured. Perhaps 'measured' is meant in some non-scientific sense (where we typically require a common metric and a notion of magnitude) in which case we have now shifted paradigms again away from science (yet 'unpredictable' is apparantly being used in the scientific sense)

4) because of (3), any non-scientific claim that cognition is being 'directly measured' by qual studies will be a fortiori true for quant studies

5) there is absolutely no literature or findings to back the qual claim up. You may as well claim that only linear regression can reveal the truth in men's souls: it is a statement of faith, nothing else.

4/12/2008 7:37 AM  
Anonymous Anonymous said...

You are just blowing smoke. The basic premise that thinking (i.e., cognition) drives human behavior is self-evident. (What do you believe determines your behavior?) It is amazing that you demand to see literature that makes such a statement!

Additionally, the idea that people can and do change their minds, and thus pursue a completely different course of action is also self-evident. This by itself renders human behavior unpredictable.

In the end I have to conclude that your protestation is a transparent and weak effort to throw dust in our eyes and disorient the readers of this blog. Because no one can logically deny the obvious facts that thinking determines human behavior, and thinking (i.e., cognition), and thus behavior, can and does change for no apparent reason -- the result is human behavior is unpredictable.

4/12/2008 7:50 AM  
Anonymous Anonymous said...

You are just blowing smoke.

No, you are! Seriously, grow up, pay attention and try to understand the debate here. If it's above and beyond you, just say so.

your protestation is a transparent and weak effort to throw dust in our eyes and disorient the readers of this blog.

ditto

Because no one can logically deny the obvious facts that thinking determines human behavior, and thinking (i.e., cognition), and thus behavior

Yep: it's a vacuous truth, and no one denies it.

They deny it can be 'directly measured' by qualitative work. Let's see the literature that claims it can.

Of course, you can't provide this literature (you would have done so months ago if you could), so I don't really know why I request it.

Instead, I imagine you will point to the "Western canon" that says human thought causes behavior (which, again, no one denies), or you will point to an article that says understanding cognition might be important (which is quite separate from whether it can be directly measured).

The funny thing is, I actually teach quant and qual classes. I happily school my students (undergrad and grad) about the benefits and drawbacks of different methods and I have many good things to say about qual approaches. This stuff about 'directly measuring cognition' is just nonsense though.

My students know it. I know it. Everyone on this board knows it. The journal editors know it. Political science departments know it.

4/12/2008 8:08 AM  
Anonymous Anonymous said...

the idea that people can and do change their minds, and thus pursue a completely different course of action is also self-evident. This by itself renders human behavior unpredictable.

If only there were ways to deal with events that at an individual level had a large random component... some sort of grouping or something. Well, a man can dream.

It would be fun to watch the ILP try to read Foundation.

4/12/2008 8:09 AM  
Anonymous Anonymous said...

You are just blowing smoke. The basic premise that thinking (i.e., cognition) drives human behavior is self-evident. (What do you believe determines your behavior?) It is amazing that you demand to see literature that makes such a statement!

I am not the poster from 4/12/2008 7:37 AM, so s/he can correct me if I am wrong. But I think your criticism of what s/he was saying is totally off base. S/he is not denying that cognition is important and/or demanding to see citations to this effect. Rather, the point is:

1. that you make baseless claims (i.e., ones not substantiated by the literature) that the only way to get at cognition is through qualitative analyses.

and

2. you use the term "cognition" without defining it, or defining it in such a broad sense that it really has no meaning.

Again, I am not the original poster that you are responding to, but please try to stick to the criticism at hand when replying, and do not put words into someone else's mouth when they obviously weren't making that point.

4/12/2008 8:11 AM  
Anonymous Anonymous said...

the only way to get at cognition is through qualitative analyses.

That has never been asserted.

you use the term "cognition" without defining it, or defining it in such a broad sense that it really has no meaning.

This is just a red herring. The obvious and clear point is that thought processes (i.e., human cognition) determines human behavior. It is not incumbent upon the political scientist to define human cognition any more precisely.

4/12/2008 8:34 AM  
Anonymous Anonymous said...

They deny it can be 'directly measured' by qualitative work. Let's see the literature that claims it can.

It is true that the qualtoid scholar seeks to directly "measure/gauge" human cognition (i.e., thinking). I doubt a journal article stating as much would convince you or other skeptics.

4/12/2008 8:37 AM  
Anonymous Anonymous said...

The obvious and clear point is that thought processes (i.e., human cognition) determines human behavior. It is not incumbent upon the political scientist to define human cognition any more precisely.

If you say so, then it must be true. Personally, my idea of good research involves precise definitions of key concepts. But, the next time I get criticized for this in a review, I'll make sure in my rebuttal to the reviewer to provide a citation to your argument on this blog. Something like this maybe:

"I really think the reviewer is wrong that I need to define [insert concept here]. It has been repeatedly 'shown' by ILP (American and Comparative Blogs, 4/12/2008 8:34 AM) that it is not incumbent upon the political scientist to precisely define concepts. Do you dare defy the wisdom of ILP?"

If that person is anywhere near as intelligent as you are, then they will see the error of their ways, throw up their hands in defeat, bow to the supreme intelligence of the almighty ILP, and tell the editor to publish my manuscript. Of course, the editor will also be aware that s/he is a mere mortal that cannot approach your infinite wisdom and will publish my manuscript in the very next issue without further delay.

4/12/2008 8:47 AM  
Anonymous Anonymous said...

RE: 8:47 a.m.

The issue is why does any qualtoid have to define/identify cognition to the level that a neuro-scientist would? The analogy would be for a Congress scholar to define an MC like a physician would?

4/12/2008 8:55 AM  
Anonymous Anonymous said...

The issue is why does any qualtoid have to define/identify cognition to the level that a neuro-scientist would?

Once again, you either misunderstand the argument or you willfully misinterpret. I will be charitable and suppose the former, in which case the debate on this blog is above and beyond you.

'Cognition' is a key concept for you. Indeed, it is the key concept, because you believe that 'measuring' it is the political science enterprise.

Key concepts require tight, rigorous definitions: that is how a science communicates. You have not provided one for 'cognition'. Why not simply write 'thinking'? Of course, we know why: it would undercut your scientific/intellectual pretentions and expose your claims. I hope 'cognition' impresses your undergrads: it wouldn't impress mine.

Similarly, you misuse 'measuring'. Why not write 'give a subjective interpretation of a person's thinking'? This is what you are doing. The reason is because such language would be accurate, and would expose your claims for what they are.

Intriguingly, you are absolutely insistent on the definition of 'prediction'---and you have several times complained that quants are using the term 'unscientifically'.

Try to be consistent: you never know, someone in the discipline might take you seriously one day.

4/12/2008 9:43 AM  
Anonymous Anonymous said...

The issue is why does any qualtoid have to define/identify cognition to the level that a neuro-scientist would? The analogy would be for a Congress scholar to define an MC like a physician would?

Once again, I did not say that the "qualtoid" has to define it the same way that a neuro-scientist would. All that I am saying is that YOU NEED TO DEFINE IT precisely. Please please please do not misrepresent what I am saying.

4/12/2008 9:50 AM  
Anonymous Anonymous said...

Once again, I did not say that the "qualtoid" has to define it the same way that a neuro-scientist would. All that I am saying is that YOU NEED TO DEFINE IT precisely.

Why?

4/12/2008 9:59 AM  
Anonymous Anonymous said...

RE: 9:43 a.m.

So you agree with the arguments put forward by the qualtoid, but your criticisms of his/her position completely center on verbiage and semantics?

4/12/2008 10:02 AM  
Anonymous Anonymous said...

Why?

WHAT???????

4/12/2008 10:26 AM  
Anonymous Anonymous said...

Once again, I did not say that the "qualtoid" has to define it the same way that a neuro-scientist would. All that I am saying is that YOU NEED TO DEFINE IT precisely.

Why?

Let me ask you this. If I said that I was going to study "influence" (say, presidential influence) but I didn't define "influence" (in either a quantitative or qualitative study) would you have a problem with this? Why or why not?

4/12/2008 10:29 AM  
Anonymous Anonymous said...

RE: 10:26 a.m.

You do not explain why the qualtoid has to provide a more precise definition of human cognition.

4/12/2008 10:30 AM  
Anonymous Anonymous said...

Let me ask you this. If I said that I was going to study "influence" (say, presidential influence) but I didn't define "influence" (in either a quantitative or qualitative study) would you have a problem with this? Why or why not?

I would not demand a definition of influence. Influence is a standard word and it requires no further elaboration. Now, whether I accept your broader argument depends wholly on the coherency of your position and the evidence you marshal in support of it.

4/12/2008 10:33 AM  
Anonymous Anonymous said...

The quantoid asks for precision because he/she thinks social/political phenomena is essentially the same as physical phenomena. So whereas precision and specificity is often possible in the physical sciences, it is not in the social sciences. Amorphorousness and imprecision are the hallmark of the social/political world, and any legitimate social science needs to reflect that amorphorousness and imprecision -- otherwise it is pseudo-science.

4/12/2008 10:53 AM  
Anonymous Anonymous said...

Amorphorousness and imprecision are the hallmark of the social/political world, and any legitimate social science needs to reflect that amorphorousness and imprecision -- otherwise it is pseudo-science.

And that's all there is to see folks.

The argument/debate/whatever you want ends here.

We actually have a poster making the case for less clarity and more vagueness in our conceptions, operations and conclusions in the discipline.

(As an aside, I hope you weren't the same poster who wanted me and many others to 'clarify' our points earlier.)

Time to wrap up the thread, me thinks...

4/12/2008 11:13 AM  
Anonymous Anonymous said...

We actually have a poster making the case for less clarity and more vagueness in our conceptions, operations and conclusions in the discipline.

What is your argument for specificity and precision in the social sciences? For instance, how would define "influence" in the example above?

4/12/2008 11:19 AM  
Anonymous Anonymous said...

Political influence certainly exists, but it manifests in myriad ways in myriad circumstances. Thus, any predetermined, rigid conception of influence would miss/exaggerate/minimize instances of political influence. This is why case studies are required to accurately detect and describe political influence when it occurs.

4/12/2008 11:27 AM  
Anonymous Anonymous said...

So now we have two posts from quallies arguing that:

-we shouldn't define concepts since amorphousness and imprecision are the hallmark of social world; and

-having less knowledge in quantitative methods might be preferable to having more knowledge.

Keep 'em coming!

4/12/2008 7:21 PM  
Anonymous Anonymous said...

we shouldn't define concepts since amorphousness and imprecision are the hallmark of social world

Concepts in the social sciences should be defined, but not always with the precision and specificity suggested by some. Our analytical concepts must have the flexibility to be useful in the multitudinous contexts and virtually infinite scenarios that political/social behavior occurs.

having less knowledge in quantitative methods might be preferable to having more knowledge

It is true that when individuals spend years mastering particular statistical techniques that they may become incapable critiquing these techniques, and accepting their limitations. So for too many of these individuals, their specific area of expertise, i.e., quantitative methods, are the answer to all empirical questions.

4/13/2008 8:09 AM  
Anonymous Anonymous said...

The debate here is done: this is not an issue of qual vs quant or rational choice vs Perestroika or any such thing.

It is, in fact, the whole of modern political science (quants + quals + rat choicers + perestroikans) vs a couple of people on this board who want political scientists to aspire to less precision in their work.

Notice that I do not claim that quant or qual or whatever is more precise, simply that everyone (and I do mean everyone) I've ever met in the discipline believes being precise matters: whether they are talking about the appropriateness of a case-study or the interpretation of a t-test.

I think this will be my last comment on here, unless the debate moves on substantially.

One last point though:

Our analytical concepts must have the flexibility to be useful in the multitudinous contexts and virtually infinite scenarios that political/social behavior occurs.

do you really think this type of pretentious phrasing helps your cause? Perhaps this is what it means to be an 'academic' to you. Do you think it impresses people? Makes them think you are a truly deep thinker?

It doesn't: we see through it---we've all got the same GRE scores, we all understand what the words mean. It's just so much verbiage, and makes it abundantly clear how weak your position is.

4/13/2008 8:35 AM  
Anonymous Anonymous said...

It is, in fact, the whole of modern political science (quants + quals + rat choicers + perestroikans) vs a couple of people on this board who want political scientists to aspire to less precision in their work.

I would think that all political scientists aspire for analytically powerful work. If that requires less precision in our ideas and more flexibility, than so be it.

do you really think this type of pretentious phrasing helps your cause? Perhaps this is what it means to be an 'academic' to you. Do you think it impresses people? Makes them think you are a truly deep thinker?

For someone aggressively arguing for precision in concepts, it is odd that you would stridently denounce someone for seeking to appropriately calibrate their language. In addition to the contradictory aspects of your position, it is also ostensibly anti-intellectual.

4/13/2008 8:57 AM  
Anonymous Anonymous said...

In addition to the contradictory aspects of your position, it is also ostensibly anti-intellectual.

The poster was trying to make his position appear anti-intellectual, but it might not really be?

You keep using that word. I do not think that word means what you think it means.

4/13/2008 9:09 AM  
Anonymous Anonymous said...

to appropriately calibrate their language. In addition to the contradictory aspects of your position, it is also ostensibly anti-intellectual.

Just can't help yourself, can you? Well, proved my point I guess.

4/13/2008 9:10 AM  
Anonymous Anonymous said...

You keep using that word. I do not think that word means what you think it means.

Ha! I noticed that too: I actually think it is quite funny and endearing. Like when a child learns a new word and wants to use it at every possible opportunity, whether it is a appropriate or not. (My kid brother used to say he felt 'sporadic'.)

4/13/2008 9:13 AM  
Anonymous Anonymous said...

Put differently, why the hostility to language? Often anti-intellectualism manifests itself as resentment toward an aptly varied vocabulary. Again, why the hostility to language?

4/13/2008 9:25 AM  
Anonymous Anonymous said...

Often anti-intellectualism manifests itself as resentment toward an aptly varied vocabulary.

That's the spirit: when you are in a hole...just keep digging...

BTW, I think the issue is that you tend to misuse words and seem to believe that someone here is impressed by your ability to do so.

4/13/2008 9:39 AM  
Anonymous Anonymous said...

I think the issue is that you tend to misuse words and seem to believe that someone here is impressed by your ability to do so.

Firstly, I challenge/reject the suggestion that I misuse words. Secondly, this is presumably a blog of Ph.D.s, so I use college level vocabulary. I would not expect my vocabulary to impress (nor intimidate) anyone on this blog.

4/13/2008 11:40 AM  
Anonymous Anonymous said...

I challenge/reject the suggestion that I misuse words

What do you think "ostensibly" means?

4/13/2008 1:01 PM  
Anonymous Anonymous said...

ostensibly: outwardly appearing as such

4/13/2008 1:09 PM  
Anonymous Anonymous said...

ostensibly: outwardly appearing as such

So when you say someone's position is "ostensibly anti-intellectual", what does that mean?

That it could very well be (pro-)intellectual but appears to you as anti-intellectual?

4/13/2008 1:14 PM  
Anonymous Anonymous said...

It is true that when individuals spend years mastering particular statistical techniques that they may become incapable critiquing these techniques, and accepting their limitations. So for too many of these individuals, their specific area of expertise, i.e., quantitative methods, are the answer to all empirical questions.

And shouldn't this be applicable to qualitative methods/area studies as well? I've seen many quallies so closely wedded to their methodology (e.g., elite interviews) or area (e.g., Latin America) that they can't fathom knowledge being derived from other methods.

I still, however, wouldn't argue that one ought to know less about Latin America than more in order to have an informed discussion about the area.

4/13/2008 1:38 PM  
Anonymous Anonymous said...

So when you say someone's position is "ostensibly anti-intellectual", what does that mean?

That it could very well be (pro-)intellectual but appears to you as anti-intellectual?


Zing! Oh man, GAG must be flaming red right now. Webster.com, dictionary.com, scrabulous... must... find... some... website... that... I... can... cite... to... support... my... ostensibly... valid... position...

4/13/2008 1:42 PM  
Anonymous Anonymous said...

good point from 4/13/2008 1:38 PM...

...have a flick through your Kaplan GRE word list to see if you can provide a stinging rebuke.

4/13/2008 1:53 PM  
Anonymous Anonymous said...

So when you say someone's position is "ostensibly anti-intellectual", what does that mean?

It appears (looks to be) anti-intellectual. I invoke the qualifier "ostensibly" to allow for the possibility that I am misinterpreting/misunderstanding the message of the poster.

Moreover, someone could intend one thing, and write/say something that has a different meaning. Thus, once that person is alerted to the apparent meaning of their statement they may correct the record. In this way the use of the word "ostensibly" is to tell someone what they are seemingly averring. Of course, this person could affirm that the apparent meaning of their statement was the intended meaning.

4/13/2008 1:56 PM  
Anonymous Anonymous said...

Moreover, someone could intend one thing, and write/say something that has a different meaning.

A bit like you with 'ostensibly' then?

4/13/2008 2:07 PM  
Anonymous Anonymous said...

You're misusing it.

For it to be "ostensibly anti-intellectual," it would have to be the case that the author clearly intended it to be anti-intellectual -- this could not be deniable -- but might have been mistaken. As in, the ostensible purpose of something is a stated purpose which can differ from the true purpose.

As "cognition" is not a synonym for "mental activity," "ostensibly" is not a synonym for "apparently."

Disingenuous mountebanks with their subliminal chicanery. A pox on them!

4/13/2008 2:11 PM  
Anonymous Anonymous said...

As in, the ostensible purpose of something is a stated purpose which can differ from the true purpose.

The operative word in your formulation is "can", but often times the apparent or ostensible is also the actual. So someone can appear to be "racist", and in the end actually be one. There is no denotation in the word ostensible that appearances must diverge from reality.

4/13/2008 2:21 PM  
Anonymous Anonymous said...

Again, the part that's relevant, and that demonstrates your poor command over literate American English, is that for something to be "ostensible," it must be clear beyond argument that the person meant to convey the message that they are "ostensibly" conveying.

Which isn't true in this case, of course. The poster wasn't trying to make his position appear anti-intellectual (whether it is or isn't). Rather, it is clear from the context that the poster meant for his position to appear anti-pseudointellectual.

Again, the point isn't that the post may have been anti-intellectual. It's that you're a dolt.

4/13/2008 4:07 PM  
Anonymous Anonymous said...

it must be clear beyond argument that the person meant to convey the message that they are "ostensibly" conveying.

What?!

4/13/2008 6:18 PM  
Anonymous Anonymous said...

And shouldn't this be applicable to qualitative methods/area studies as well? I've seen many quallies so closely wedded to their methodology (e.g., elite interviews) or area (e.g., Latin America) that they can't fathom knowledge being derived from other methods.

I am not certain why you are invoking area studies (i.e., Latin America). In any event, I have never known a qualtoid adopt such a strident/intolerant attitude. I have only know quantoids hold that all questions can be addressed via quantitative methods. This attitude has been manifest on this blog.

4/13/2008 6:22 PM  
Anonymous Anonymous said...

What?!

It's really quite simple. I'll use small words.

If X is ostensibly Y, then whoever said that meant to convey that X is Y. The person who said that intended to give the appearance that X is Y, even if X is not Y. The goals of the speaker can't be a subject of reasonable debate. Usually, "ostensible" is a near-synonym to "stated." As in "stated goal" versus "ostensible goal." This is even evident in the etymology, for God's sake. Ostensible from Latin "ostendere" to show or display, because the shown or displayed thing may not be the true thing.

If I say that your comment "What?!" is ostensibly Spanish, I would be wrong because no reasonable person would agree that you intended your comment to appear to be Spanish.

This is relevant to you because no reasonable person would agree that the relevant poster had the intent of appearing to be anti-intellectual, even if he or she is not actually so. This means that it is not ostensibly anti-intellectual, because that is not how the words "ostensible" or "ostensibly" are used by broadly literate speakers of American English.

4/13/2008 7:13 PM  
Anonymous Anonymous said...

This is relevant to you because no reasonable person would agree that the relevant poster had the intent of appearing to be anti-intellectual

I disagree.

4/14/2008 5:36 AM  
Anonymous Anonymous said...

While we are on misuse of words by ILP, can I ask about

As a point of logical,
(4/01/2008 7:13 AM)

and

That is the serious flaw in your logical. (4/07/2008 4:49 PM)

Have we started using 'logical' as a noun now? Or is this (ostensibly) nonsense?

4/14/2008 5:52 AM  
Anonymous Anonymous said...

RE: 5:52 a.m.

There is a difference between a misuse of a word and typos.

4/14/2008 6:34 AM  
Anonymous Anonymous said...

There is a difference between a misuse of a word and typos.

Ostensibly, I guess.

4/14/2008 6:42 AM  
Anonymous Anonymous said...

Just finished some qual research reading this mornings paper. Now on to some quant models using the insights I've gained from the qual research. I'm into the multimethod approach.

4/14/2008 7:25 AM  
Anonymous Anonymous said...

Q: How many quals does it take to change a lightbulb?

A: can you really put a number on that?

-----
Thank you, thank you, I'm here all week -- do try the buffet.

4/14/2008 7:41 AM  
Anonymous Anonymous said...

I disagree.

You're not reasonable, so you don't count.

4/14/2008 9:22 AM  

<< Home