Methodology/Theory Debates
posted by American and Comparative Jobs at 9/10/2008 06:10:00 AM
American and Comparative Jobs
Hello bloggers! As requested, I have turned off comment moderation. But, if it becomes a personal free-for-all I will ask someone else to manage the blog because I do not have time to devote to continuously deleting comments.
323 Comments:
Who wrote that cognition causes cognition? I do not even know what this means?
9/08/2008 9:46 AM: "Accordingly, if cognition is caused by something else in the social world, the root cause of this is cognition as well.
It's cognition all the way down."
9/08/2008 10:16 AM: "It's really very clear to me. You want to find out what causes cognition. All human behavior is caused by cognition. Do you hold that something other than cognition causes cognition?"
9/08/2008 12:31 PM: "Very well, you concede. Your earlier question of what causes cognition was misguided. In the social sciences, cognition causes everything, including cognition."
You're getting away from political science, aren't you?
Why?
Because you're now interested in cognition. Just like the folks working on the effects of redistricting are not political scientists (according to your own words), you're now veering away from political science. You're in the realm of cognitive psychology. Don't you see it?
9/08/2008 9:46 AM: "Accordingly, if cognition is caused by something else in the social world, the root cause of this is cognition as well.
It's cognition all the way down."
9/08/2008 10:16 AM: "It's really very clear to me. You want to find out what causes cognition. All human behavior is caused by cognition. Do you hold that something other than cognition causes cognition?"
9/08/2008 12:31 PM: "Very well, you concede. Your earlier question of what causes cognition was misguided. In the social sciences, cognition causes everything, including cognition."
You are either quoting yourself, or someone who is hostile to the qualitative argument. As such it is not an affirmative argument in favor of focusing on cognition for political analysis, but someone who either does not understand this position or is consciously seeking to malign it.
Because you're now interested in cognition. Just like the folks working on the effects of redistricting are not political scientists (according to your own words), you're now veering away from political science. You're in the realm of cognitive psychology. Don't you see it?
The effects of redistricting do fall into the realm of political science. What does not is the demographic components of drawing districts.
As for cognitive psychology, to the extent that this discipline can give us insight into political decision-making, then, yes, it can overlap with political science.
So you've retracted yourself. That's good to hear. I didn't think you were capable of doing so.
Now if you could only explain why you thought that 40 degrees F was twice as hot as 20 degrees F, (was it a simple misunderstanding of the question; was the question worded poorly, etc.)...
So you've retracted yourself.
???????
A reading assignment for the qually:
http://gking.harvard.edu/files/nolie.pdf
Cognition all the way down.
Brilliant!
Although I doubt cognition soup is as good as turtle soup. As Ministry said, the mind is a terrible thing to taste.
A reading assignment for the qually:
http://gking.harvard.edu/files/nolie.pdf
Don't forget that you're trying to engage someone who said earlier that no theory was necessary when coding variables (i.e., for measurement). Good luck.
I just started reading the King and Powell piece recommended earlier, and on page 3 an explicit claim is made that "laws" serve as basis for the use of quantitative techniques. They write:
"The idea is that in statistics and other areas, we often need to be able to draw random samples of two variables, say x and y . . . (which indicates the law [emphasis added] governing how particular values of x and y occur together) . . .
Precisely because the authors erroneously believe human social/political behavior is driven by "laws", their commentary on qualitative methods contains little insight. So, for instance, they make the claim that there is no qualitative position on inference (p. 7). Of course, as has been argued at length on this blog the qualitative basis of inference is based inferring human cognition.
Additionally, the authors hold a clear bias toward quantitative technique. They hold in "situations where you are able to quantify the essential information, gear up to do a proper statistical analysis, and commit the time and resources necessary, it is worth going ahead because the increased precision will likely yield for more accurate results" (11-12).
What is most interesting, and telling, is that the prime empirical evidence the authors put forward to support this claim are instances where statistical analysis has out-predicted qualitative thinkers (11). Of course, the weakest component of social science is predictive, and statistical analysis has only marginally improved on this deficiency. Understanding the autonomy inherent in human reasoning qualitative scholars are loath to make specific predictions. In contrast, quantitative social science scholars have created something of a cottage industry out of making predictions, and, perhaps, this is most evident in political science -- most glaringly in the realm of presidential elections. Of course, U.S. presidential elections are dyad outcomes. It would be much more impressive and persuasive if quantitative predictions about presidential elections were highly successful if multiple outcomes were possible.
I just started reading the King and Powell piece recommended earlier, and on page 3 an explicit claim is made that "laws" serve as basis for the use of quantitative techniques. They write:
"The idea is that in statistics and other areas, we often need to be able to draw random samples of two variables, say x and y . . . (which indicates the law [emphasis added] governing how particular values of x and y occur together) . . .
I guess the analogy they were making to a joint probability distribution between X and Y was lost on you...
In that passage, x and y aren't variables like income and vote. Rather, they're talking about theory and evidence. The paper is making the analogy to MCMC with iterative processes to indicate how theory and data can be combined in an iterative process as well.
You don't have to agree with their conclusion. But you should at least reocognize that the "laws" they talk about aren't laws of social/political behavior.
Actually, now that I read it again, I'm appalled that you interpret their sentence as a claim that laws serve as the basis for the use of quantitative methods. Is that how you, a qualitative researcher, conducts content analysis, or even interpretive analysis of texts?
Read the passage again, for heaven's sake! I'll admit that it's not easy--one has to know what MCMC is, what Gibbs sampling is, what a bivariate probability distribution is, and what a conditional probability is. I understand if this is lost on you--but don't ascribe your own thoughts to their work. You're completely off base here, and the saddest thing is you don't even know it...
the qualitative basis of inference is based inferring human cognition.
What does this mean?
Do you know what a theory of inference is? From your post, I don't think you do. Human cognition is not a theory of inference.
Additionally, the authors hold a clear bias toward quantitative technique. They hold in "situations where you are able to quantify the essential information, gear up to do a proper statistical analysis, and commit the time and resources necessary, it is worth going ahead because the increased precision will likely yield for more accurate results" (11-12).
Stop the presses! Gary King favors quantitative methods! Oh the humanity!
By the way, did you notice that even they write this is applies to a "small number of situations"? So don't freak out!
But why are you so adverse to this? I, for one, would be willing to say the following instead: "in situations where you are able to [read the material in its original language--say Spanish], gear up to do a proper [content analysis of the original sources], and commit the time and resources necessary, it is worth going ahead because the increased precision will likely yield for more accurate results"
Would you agree with the preceding statement? It's not a slight against quants to say that one should learn Spanish if one has the time/resources to do so and one is able to then use it properly. Just as it isn't a slight against anyone to say
"hey, in those rare instances where you have time, the data can be quantified, and you can do a proper quantitative analysis, why not do it?"
Don't panic, they're just offering you tips. But to hide behind the "bias" the authors have is to miss the substantive point they are making. As a qual researcher, don't you cite people who are "biased" toward qual research? Of course you do--and it's okay.
Of course, U.S. presidential elections are dyad outcomes. It would be much more impressive and persuasive if quantitative predictions about presidential elections were highly successful if multiple outcomes were possible.
Fair point. If I flip a coin, I'll be right 50% of the time, so it's not very meaningful.
Just keep in mind, however, that the cottage industry you're referring to doesn't simply attempt to predict which party wins the White House. Rather, they attempt to predict the percentage of the vote each candidate will receive (for example). So there are an infinite number of outcomes to choose from.
Still, we're dealing with a two-candidate race, so you're right that it's not as hard as predicting the share of the vote that each Israeli party will get.
But your claim that there are only two outcomes to choose from in those predictive models is actually incorrect (in those models that predict the vote share). Just so we're clear on what these models are actually trying to do.
"Actually, now that I read it again, I'm appalled that you interpret their sentence as a claim that laws serve as the basis for the use of quantitative methods. Is that how you, a qualitative researcher, conducts content analysis, or even interpretive analysis of texts?"
Yup. Once again we are faced with the poster making bold claims that reveal them to be either ignorant or willfully misrepresenting the views of others. It has happened time and again.
But you should at least recognize that the "laws" they talk about aren't laws of social/political behavior.
It is very important to stress that King and Powell write the "law governing how particular values of x and y occur together) . . . ."
Therefore, your denial that the authors are referring to laws as the basis of statistical analysis makes little sense. You write "they're talking about theory and evidence."
But your claim that there are only two outcomes to choose from in those predictive models is actually incorrect (in those models that predict the vote share). Just so we're clear on what these models are actually trying to do.
The relevant point is that King and Powell do not point to the ability of quantitative scholars to precisely predict presidential vote totals, but solely the winner. Moreover, there is no consensus among presidential election scholars on how to specify a model to reliably predict popular vote totals of U.S. presidential elections. So this field of study has not meaningfully advanced our ability to predict the future.
the qualitative basis of inference is based inferring human cognition.
What does this mean?
It means that qualitative theories are predicated on inferring the content of human cognition/thinking.
One point raised by King and Powell does point to the scientific superiority of qualitative methods. They write:
"Qualitative scholars understand that the real question social scientists have in many cases is what explains an event: Why did the Cuban missile crisis happen? Why, in free and fair elections in Weimar Germany, did voters choose to elect Hitler? What was the cause of the French Revolution? Why did the terrorists blow up those buildings on 9/11?"
The authors add:
"Quantitative scholars can proceed to define and evaluate causal effects via potential outcomes for each candidate explanatory factor -- which is their most helpful suggestion in cases like these -- but they have not properly defined or formalized the definition of an explanation or how to estimate it. They need to understand that explanations are a legitimate target of social scientific inquiry."
While King and Powell do not explain why qualitative research "leads" on this issue, the answer is obvious. Qualitative research proceeds from a clearly defined causal chain. In other words, all causation of human behavior proceeds from human cognition (albeit shaped by various factors). Thus, when qualitative researchers ask "Why did the Cuban missile happen?", what they are asking and answering is what was going through people's (i.e., decision-makers') minds that lead to these actions/behaviors (i.e., the Cuban missile crisis). In contrast, quantitative research is not so clearly defined, and as a result is a much more murky (less scientific) endeavor.
t is very important to stress that King and Powell write the "law governing how particular values of x and y occur together) . . . ."
Therefore, your denial that the authors are referring to laws as the basis of statistical analysis makes little sense. You write "they're talking about theory and evidence."
read the paragraph again, please. Don't you see they treat x as theory and y as evidence? Don't you see that?
Sorry, x=data, y=theory.
Do you disagree?
Do you know which "laws" they are referring to with respect to the joint bivariate distribution p(x,y)? Do you not understand that p(x,y) describes the law (they use "law", I would have used "process") that governs how x and y co-vary?
They're making an analogy to a joint probability distribution. I don't know what else to say, other than you're obviously unaware of the subtleties of their argument. They should probably change the language to help those without statistical knowledge understand the passage.
It means that qualitative theories are predicated on inferring the content of human cognition/thinking.
That is NOT what K&P are talking about when they discuss a theory of inference. They're talking about how you can infer something to a universe of cases based on a sample of cases. That's what they are talking about. Can you follow any argument???
The relevant point is that King and Powell do not point to the ability of quantitative scholars to precisely predict presidential vote totals, but solely the winner.
WRONG! You are either being willfully deceitful or simply ignorant. I don't know which one is worse. I suggest you look at the studies they cite. For example, they cite Campbell's article discussing presidential vote forecasts (not winners!). Really, there is a huge literature on vote forecasts model. Not much on "predicting the winner" models. Note that this doesn't mean there isn't any (e.g., the literature on the 1936 Literary Digest poll).
Moreover, there is no consensus among presidential election scholars on how to specify a model to reliably predict popular vote totals of U.S. presidential elections. So this field of study has not meaningfully advanced our ability to predict the future.
You're right, there is a disagreement in the literature about how to specify vote forecasting models. According to you, then, disagreement about model specification doesn't meaningfully advance our understanding. Really? I thought you had said earlier that disagreements in the qualitative literature were good because they would allow the reader to assess the evidence and decide for him/herself. I'd say that you're right--and that the same logic applies to forecasting models. I'll give you an example. There was a model long ago that showed that you could forecast how the stock market would perform based on the team that won the Super Bowl. That model was really strong in terms of prediction. Do you think that as a reader, you're unable to assess for yourself whether that is a "good" model or a bad one? Are you saying that you do not possess the tools to assess the validity of such a model?
I am completely confident that the qually does not possess the tools to assess the validity of such a model?
The effects of redistricting do fall into the realm of political science. What does not is the demographic components of drawing districts.
You've retraced yourself here. You used to say that studies that examined the partisan bias of redistricting were not political science. Now you're saying that examining the effects of redistricting IS political science. Partisan bias is an effect of redistricting. Thus, you have retracted your previous comment.
I'm happy to see that, under some conditions, you are open-minded enough to change your tune.
Question for the quants on this blog:
which quantitative model of armed conflict predicted the success of the surge in Iraq? I will surmise that no predictive quantitative model is up to that task.
Here is a model: More troops equals higher probability of winning.
You used to say that studies that examined the partisan bias of redistricting were not political science. Now you're saying that examining the effects of redistricting IS political science.
Determining that a district has been drawn in a bias manner is not political science. It is demography. The effects of partisan bias in drawing districts -- i.e., who gets elected; how it affects are legislative dynamics -- is political science.
Do you know which "laws" they are referring to with respect to the joint bivariate distribution p(x,y)? Do you not understand that p(x,y) describes the law (they use "law", I would have used "process") that governs how x and y co-vary?
They're making an analogy to a joint probability distribution. I don't know what else to say, other than you're obviously unaware of the subtleties of their argument. They should probably change the language to help those without statistical knowledge understand the passage.
You do not like K&P's argument, so you try to change it. Interest . . .
That is NOT what K&P are talking about when they discuss a theory of inference. They're talking about how you can infer something to a universe of cases based on a sample of cases. That's what they are talking about. Can you follow any argument???
The point is they are wrong in their statements about inferences in qualitative research. That is the issue at hand.
WRONG! You are either being willfully deceitful or simply ignorant. I don't know which one is worse. I suggest you look at the studies they cite. For example, they cite Campbell's article discussing presidential vote forecasts (not winners!). Really, there is a huge literature on vote forecasts model.
Irrespective of Campbell et al.'s content, K&P do seem to point to picking presidential winners, not vote totals. In any event, they should have been clearer on the matter.
You're right, there is a disagreement in the literature about how to specify vote forecasting models. According to you, then, disagreement about model specification doesn't meaningfully advance our understanding. Really? I thought you had said earlier that disagreements in the qualitative literature were good because they would allow the reader to assess the evidence and decide for him/herself.
You are comparing apples and oranges. There is vast difference between analyzing and explaining something that happened in the past, and predicting the future. Until, and unless, there is confidence in the specification of a model predicting the future then it has little to no analytical utility.
The effects of partisan bias in drawing districts -- i.e., who gets elected; how it affects are legislative dynamics -- is political science.
And that is what the literature on partisan bias of redistricting actually does. Again, I'm not using the term "partisan bias" willy-nilly. There's a literature that examines how redistricting schemes can systematically favor one party in terms of how votes are translated into seats. That's the partisan bias I'm referring to.
So we're in agreement: examining partisan bias of redistricting plans IS political science. Why do you still refuse to admit it???
You do not like K&P's argument, so you try to change it. Interest . . .
From p.3 of their paper: "Finally, we shore up the mathematical foundations of this procedure by noting that iterating in this way is a qualitative version of Gibbs sampling in MCMC statistical algorithms.... Under the right conditions, we can prove mathematically that the iterative process will converge to draws from the desired joint distribution--which under the analogy to our case, should give the right theory and the right evidence for the theory.... convergence to the joint distribution under Gibbs sampling requires that we iterate enough times, collect enough data implied by the theories, and explore enough theories consistent with the data.... In practice, the only real check on whether we have reached convergence is to see whether in a long string of iterations we find the same theory along with consistent observations..."
They are making an ANALOGY! Can't you see it?
The point is they are wrong in their statements about inferences in qualitative research. That is the issue at hand.
That's a fair point. It's not the same as saying that qualitative research is based on a theory of inference called 'human cognition.' That was the point I was making--human cognition is NOT a theory of inference.
Irrespective of Campbell et al.'s content, K&P do seem to point to picking presidential winners, not vote totals. In any event, they should have been clearer on the matter.
Ah, irrespective of the work cited... I think that's a cop-out.... nevertheless, they talk about forecasting presidential elections and refer the reader to models of vote forecasts. I think a fair-minded reading of the passage would conclude that they are talking about vote forecasting models. If not, can you point me to recent forecasting models whose main forecast (i.e., dependent variable) is which of the two candidates will win the election? I cannot think of one--all the ones I know are vote forecasting models (or seat forecasting models for the House).
There is vast difference between analyzing and explaining something that happened in the past, and predicting the future.
Oh, I didn't know that! LOL. Of course those are different. But tell me, what data do you think one uses to predict the future? Past data? Future data?
Until, and unless, there is confidence in the specification of a model predicting the future then it has little to no analytical utility.
And how much confidence would you need in a model to conclude that it has utility? 95%? What's your benchmark? I'm serious: how can you judge if predictive models have utility (given that they are published before the future occurs)?
All in all, you're really digging yourself in a deeper hole every time you post.
Re-read the King piece. Familiarize yourself with the literature on partisan bias (it has a very specific meaning in the literature on redistricting and electoral systems more generally). Read one or two predictive vote studies. I'm telling you, my guess is you will learn a lot from all that.
Our qually friend once again discredits himself with an erroneous reading of a scholarly piece. I'm starting to understand why s/he constantly cites the New York Times in his posts.
Oh man, I'm going to have fun this semester with his book in my grad research design class... I'll let you know how the discussion goes. I use it as an example of how bad qualitative research looks like...:)
They are making an ANALOGY! Can't you see it?
No
That was the point I was making--human cognition is NOT a theory of inference.
Why?
I'm serious: how can you judge if predictive models have utility?
If they can reliably predict the future, then they have utility. If they cannot, then they have none.
Determining that a district has been drawn in a bias manner is not political science.
A bias manner?
English, motherf----r! Do you speak it?
They are making an ANALOGY! Can't you see it?
No
Really? That's sad.
That was the point I was making--human cognition is NOT a theory of inference.
Why?
Well, do I need to spell out what a theory of inference is? See 11:14.
If they can reliably predict the future, then they have utility. If they cannot, then they have none.
You're evading the question, Mrs. Palin. Earlier, you used the word "confidently" now you use "reliably." My question to you: how much confidence/reliability is needed for the forecast to be valid? I'm trying to engage you in a substantive point, but you keep evading it. If I say: Obama wins 53-47 and he wins 51-49, is my model "good"? What if he wins 57-43? 58-42? 60-40? Where do you draw the line and why?
Also, you say that a model has utility if it can reliably predict the future. Let's say we say that this means if it can predict at least 9 of the next 10 elections. Does this mean that a model's utility will only be determined in 40 years? By the same token, we cannot claim that a model has no utility before then (or at least before we get two wrong), no? You're establishing a very weird benchmark. Perhaps you should think about publishing this--it would change the way social scientists make predictions. It's quite revolutionary.
Well, do I need to spell out what a theory of inference is? See 11:14.
I understand that you have offered one theory of inference. I have offered a different one. Is it your position that only one theory of inference can be valid?
You're evading the question, Mrs. Palin. Earlier, you used the word "confidently" now you use "reliably." My question to you: how much confidence/reliability is needed for the forecast to be valid? I'm trying to engage you in a substantive point, but you keep evading it. If I say: Obama wins 53-47 and he wins 51-49, is my model "good"? What if he wins 57-43? 58-42? 60-40? Where do you draw the line and why?
Also, you say that a model has utility if it can reliably predict the future. Let's say we say that this means if it can predict at least 9 of the next 10 elections. Does this mean that a model's utility will only be determined in 40 years? By the same token, we cannot claim that a model has no utility before then (or at least before we get two wrong), no? You're establishing a very weird benchmark. Perhaps you should think about publishing this--it would change the way social scientists make predictions. It's quite revolutionary.
The more precisely that a model can consistently predict the presidential vote totals the more confidence people will have in that model. Yes, it will take time for any model to build up a track record before will begin to have confidence in it and view it as reliable.
All right people, we're gonna try something a little different today. We are going to write an essay of no less than a thousand words describing to me what you think a theory of inference is. And when I say essay, I mean essay. I do not mean "human cognition" repeated a thousand times. Is that clear Mr. Qually?
What I would be interested to know is why you reject an approach that would be directly geared toward inferring human cognition (i.e., thinking)?
I understand that you have offered one theory of inference. I have offered a different one. Is it your position that only one theory of inference can be valid?
I haven't offered a theory, only a definition. I still don't understand how "human cognition" is a theory of inference. It may be a theory of how humans act, but it's not a theory of how we can make inferences from cases we observe, is it?
The more precisely that a model can consistently predict the presidential vote totals the more confidence people will have in that model. Yes, it will take time for any model to build up a track record before will begin to have confidence in it and view it as reliable.
OK, you had three chances, and you've evaded each time. This must mean you have no answer. What does it mean to "precisely," "consistently," "confidently," "reliably" make predictions? You know, adding more adjectives doesn't help me understand your use of the other adjectives.
Again, what's your benchmark?
What I would be interested to know is why you reject an approach that would be directly geared toward inferring human cognition (i.e., thinking)?
Wow, isn't it now obvious that the qually doesn't understand the most basic question of inference?
Qually, rational choice theory. Theory of inference, yes or no? I'm not asking whether you agree with the theory--does it fall under a theory of inference, yes or no?
What I would be interested to know is why you reject an approach that would be directly geared toward inferring human cognition (i.e., thinking)?
"Human cognition" is no more a theory of inference than "rational choice" is.
Again, what's your benchmark?
In my view, any such model would have to predict presidential vote totals within 1 percent of their actual outcomes. Anything else, in my estimation, would amount to little more than guesswork.
It may be a theory of how humans act, but it's not a theory of how we can make inferences from cases we observe, is it?
It goes back to earlier discussions. Namely, that people will disagree whether inferences about human cognition are valid/credible. In the end, there is no pre-set formula.
Let us take current presidential politics as an example. Is Obama's recent lurch to the right indicative of how he would govern, or is it a posture adopted solely for electoral purposes? Different people are going to infer what his state of mind based on different criteria. There is no adjudicating such a dispute. The idea that a statistical technique can resolve such a matter is based on ignorance and hubris.
Different people are going to infer what his state of mind based on different criteria.
So, you're saying that the fact that quallies like you are sloppy and imprecise is a good thing?
Sorry, but the mainstream of political science endeavors to make clear coding choices, consider all evidence available, and develop solid inferences using whatever techniques are appropriate for a problem. That's where I stand.
Sorry, but the mainstream of political science endeavors to make clear coding choices, consider all evidence available, and develop solid inferences using whatever techniques are appropriate for a problem. That's where I stand.
Apart from taking a stand, you have to explain how your position is empirically valid?
"Human cognition" is no more a theory of inference than "rational choice" is.
Exactly! Finally someone gets it right.
In my view, any such model would have to predict presidential vote totals within 1 percent of their actual outcomes.
Finally. Thank you for providing a benchmark.
I will note that such a benchmark, of course, is arbitrary (i.e., why 1% why not 1.4% 2% 3% 3.666667%). But at least you have a clear benchmark.
But let's be clear: you would reject a model that would have consistently predicted the Republican vote with a 1.1% avg. error?
Let us take current presidential politics as an example. Is Obama's recent lurch to the right indicative of how he would govern, or is it a posture adopted solely for electoral purposes? Different people are going to infer what his state of mind based on different criteria. There is no adjudicating such a dispute. The idea that a statistical technique can resolve such a matter is based on ignorance and hubris.
I don't know any quantitative model that could achieve that--so you're putting up a red herring here. Unless you can point me to some quant. work that does.
You like putting up straw men, don't you?
I don't know any quantitative model that could achieve that
That is my point exactly!
I will note that such a benchmark, of course, is arbitrary
I agree completely. That is why I was reticent to put forward a specific/quantitative definition of confidence, because people will legitimately disagree over how effective a model has to be in order to inspire "confidence."
The idea that a statistical technique can resolve such a matter is based on ignorance and hubris.
So why put up this red herring? You put up this sentence as "an idea"... but it's your idea... then you shoot it down as ignorance and hubris on the part of quants... but no one on the quant side made that claim...
I agree completely. That is why I was reticent to put forward a specific/quantitative definition of confidence, because people will legitimately disagree over how effective a model has to be in order to inspire "confidence."
OK, we're in agreement. So why adopt such a black/white tone ("models that cannot confidently predict the future have little to no utility")?uses
By making that statement, you invite the question: what constitutes confidence?
Let's put it another way. Let's say that I have a predictive model that uses "experience" measured in number of years in government as the independent variable. Then "winner" or "% vote" on the left-hand side. How would such a model perform? Probably poorly (e.g., Kerry and Gore had more years in government than Bush; Ford had more years than Carter; etc.). So that model probably can't explain (or predict? I don't know b/c I don't see the future) results. According to your logic, we haven't learned anything. I would venture that we did learn something: that experience measured as nb. of years in govt. doesn't correlate with electoral success.
So, what's the next step? Someone may say: "we can therefore repudiate the claim that experience helps win elections." Someone else might say: "hold on, perhaps your measure didn't capture the concept of experience appropriately. Let me suggest a different measure." And so on.
I'd say that we learned that experience measured in years in govt doesn't predict success. Whether this means that experience doesn't help AT ALL, well, it would be up to the next researcher to measure experience differently.
But to say that there is little or no utility in the model is akin to saying that there is little or no utility in learning that something may not matter as much as we thought it did.
I would venture that we did learn something: that experience measured as nb. of years in govt. doesn't correlate with electoral success.
In the first instance, sometimes these factors may correlate and sometimes they may not. Do we need an involved statistical analysis to tell us this? No!
Most importantly, the statistical analysis you propose does not get at why a relationship exists, or does not exist. As a result, your exercise would not generate any knowledge.
Most importantly, the statistical analysis you propose does not get at why a relationship exists, or does not exist. As a result, your exercise would not generate any knowledge.
Would your qualitative analysis tell us why a relationship exists or does not exist? No. The theory tells us why something matters or does not matter, irrespective or whether we are dealing with qualitative or quantitative methods. Get it?
So once again, your point is not about qualitative versus quantitative analysis, but rather if the analysis is based on a sound theoretical argument or not.
Would your qualitative analysis tell us why a relationship exists or does not exist?
Yes it would. At least that would be the goal of the qualitative analysis -- not establishing simple correlation.
The theory tells us why something matters or does not matter, irrespective or whether we are dealing with qualitative or quantitative methods. Get it?
I do not understand what you are arguing here.
So once again, your point is not about qualitative versus quantitative analysis, but rather if the analysis is based on a sound theoretical argument or not.
What does this mean?
Let's stop responding to ILP and see how long it takes before he posts again.
What would be the over/under on that? I say 8 days.
Qually,
Don't let snide comments stop you from posting. I have learned a lot!
me too! more so than in so-called methods courses they make you ake in grad school.
Thank you for your show of support.
A quant wrote the following: The theory tells us why something matters or does not matter
I think what this person is seeking to communicate is that he does not critically analyze theorizes/evidence for himself. Instead, he follows whatever protocol is stipulated by quantitative methods (BTW, which determine his theories). In the end, it does not matter what is the "truth", but what the methods tells us. Thus, there is no critical distance between researcher/thinker and method.
RE: 11:02 a.m.
It could be that this quant is hiding behind his/her methods/theories, as quants are apt to do.
At what point does "New" political science become old? What's the shelf life of something being called "new"?
From the Chronicle of Higher Education:
The researchers' report, published in today's issue of the journal Science, suggests that genetic differences may help explain why some people favor capital punishment and the Iraq War, while others support gun control and foreign aid. It's part of a growing field, called "genopolitics," that is threatening to rewrite the rules of political science, which hold that political beliefs are shaped by people's environment and experiences. To work in the new field, political scientists are scrambling to learn genetics, neuroscience, and other aspects of biology.
Does this line of research have any validity, or is this quant research run amoke -- with undertones of eugenics?
Source: http://chronicle.com/daily/2008/09/4669n.htm?utm_source=at&utm_medium=en
Does this line of research have any validity, or is this quant research run amoke -- with undertones of eugenics?
So that's the choice set, huh?
Will Americans elect Obama or will they prove to be a racist people?
Think how wonderful it would be for quants if it was conclusively determined that political/social behavior is determined by genes (put differently, the laws of physics). Human cognition would no longer be the cause of human behavior. Then we could focus on the inanimate forces of genetics -- where quantitative methods have supreme utility.
if it was conclusively determined that political/social behavior is determined by genes
When will you learn that quant methods cannot conclusively determine anything?
When will you learn that quant methods cannot conclusively determine anything?
Where in 7:15 a.m. is it asserted that quant methods conclusively determine anything, or can conclusively determine something?
I guess I assume that when he wrote "conclusively determined" he meant conclusively determined. My bad.
But then again, I remembered an exchange on this thread several months ago where ILP argued that quants would "determine" things, something he vehemently disagreed with. Of course, it was a straw man as no quant would ever claim to conclusively determine anything.
This led to a discussion of p-values, which are relevant to the discussion. Then ILP dismissed that as being "mumbo jumbo" as if somehow p-values were irrelevant to the discussion of deterministic vs. probabilistic conclusions.
Again, my bad.
Think how wonderful it would be for quants if it was conclusively determined that political/social behavior is determined by genes (put differently, the laws of physics).
Another straw man put up by ILP. Don't bite, folks
Another straw man put up by ILP.
How is what the ILP wrote a straw man?
Will Americans elect Obama or will they prove to be a racist people?
We can blame the public for the public’s lack of enthusiasm for Obama, as Democratic apologists are apt to do. Nevertheless, the Obama campaign has staked out a neo-liberal and neo-militarist position, not to mention embraced the religious/social right. Thus, Obama in his advertising alleges that McCain will continue Bush’s policies. What people, need to hear, however, is how Obama will differ from Bush’s policies. Obama has not sought to substantively distance himself (and the Democratic Party) from Bush. (Of course, the Democrats refused to distance themselves from Bush and Co. when Congressional Democrats did not initiate impeachment proceedings.)
Dumb qually!!! Quantoids are not trained to study politics! So stop discussing politics (a straw man), and learn to count -- because non-political political science is the future.
You (sarcastically) mention how wonderful it would be for quants to conclusively determine behavior. You then go on to claim that this is akin to eugenics.
Quants will be the first to tell you that they cannot conclusively determine anything.
So your entire post is a straw man.
Sorry, you've been caught red handed. How embarrassing! LOL
What is a neo-militarist position?
What is a neo-militarist position?
The use of military power to dominate/control Eurasia.
The use of military power to dominate/control Eurasia.
Silly qually, we're at war with Eastasia. Always been that way.
Silly qually, we're at war with Eastasia.
You may not realize it, but the point was made about Eurasia (the entirety of the European and Asian landmass), not East Asia.
You may not realize it, but someone is making fun of you using a reference to 1984 (a book by George Orwell).
How come the methods/theory debates accumulate much much faster than methods/theory job rumors??
Because GAG isn't on the job market.
In 1984 there was Eurasia (as well as Eastasia).
RE: 1:46 PM
On this blog there is usually a specific line of argument/debate, whereas on the other blogs it is normally idle banter.
In 1984 there was Eurasia (as well as Eastasia).
You may not know this, but in 1984 characters believe that a past war with Eurasia was actually a war with Eurasia.
You may not know this, but you can donate one or more of your vital organs to the Aperture Science Self-Esteem Fund for Girls.
Obama has not sought to substantively distance himself (and the Democratic Party) from Bush.
Someone hasn't been following the campaign.
But this is besides the point. Bring your discussion of the 2008 campaign to a different thread. 9/19/2008 11:18 AM was giving you an example of a false dichotomy--and you respond with this diatribe?
No wonder folks don't take you seriously.
Someone hasn't been following the campaign.
But this is besides the point. Bring your discussion of the 2008 campaign to a different thread. 9/19/2008 11:18 AM was giving you an example of a false dichotomy--and you respond with this diatribe?
No wonder folks don't take you seriously.
This is an excessively defensive and rancorous response. The earlier point raised about the Obama campaign is perfectly appropriate for a blog committed to an analysis of politics. If you followed this thread, you would know that.
The earlier point raised about the Obama campaign is perfectly appropriate for a blog committed to an analysis of politics
The earlier point (which I had made) was about a false dichotomy, not about the 2008 campaign. I picked the Obama analogy.
But hey, if you want to go off topic, you're welcome to do so. I just don't think your post belongs on a "methodology/theory" debate. But I guess the line between scholarship and blind advocay has been blurred a long time ago.
Great "analysis" of the 2008 campaign, by the way.
RE: 1:11 P.M.
Do you deny that the Obama campaign has staked out a neo-liberal and neo-militarist position, not to mention embraced the religious/social right.?
RE: 7:31 P.M.
Do you deny that you can not follow a conversation, and that you mistake a rhetorical device in one conversation for a separate conversation?
How come the methods/theory debates accumulate much much faster than methods/theory job rumors??
The median score on the autism scale in this thread is a standard deviation higher than it is in the other threads.
Do you deny that you can not follow a conversation, and that you mistake a rhetorical device in one conversation for a separate conversation?
This is incoherent.
RE: 4:45 AM
Do you deny that you can not follow a conversation, and that you mistake a rhetorical device in one conversation for a separate conversation?
I think the point here is that Obama will say one thing to one audience and something else to a different audience. This does raise the question, however, of which of the two messages should people believe? (If only we could gain scientific insight into his cognition!)
It is worth noting that under the leadership of Bill Clinton and the Democratic Leadership Council the Democratic Party has significantly shifted to the right, and embraced the ideology of the free market and distanced itself from notions of social justice and economic regulation. The historical record also demonstrates that the Democratic Party is just as committed as the Republican Party to using military power to impose U.S. geopolitical goals.
This leads back to the first matter raised -- why has the public not embraced Obama (especially given the strong public opposition to Iraq, the faltering economy, and the unpopularity of the Republican incumbent)? The answer is simple: the public, for good reason, does not think that Obama (and the Democratic Party) represent a break with the political agenda that currently predominates in Washington.
"I think the point here is..."
No. the point is that someone used an example of a false dichotomy to point out one in your comment that "Does this line of research have any validity, or is this quant research run amoke -- with undertones of eugenics?", and you failed to understand it. Instead you read into the comment what you wanted and assumed it was someone making and argument.
You do this sort of thing all the time. Whats the deal?
So that's the choice set, huh?
Will Americans elect Obama or will they prove to be a racist people?
This was the comment that started the line on Obama's campaign. It appears reasonable to take up the question of whether a public vote against Obama is the result of racism.
So that's the choice set, huh?
Will Americans elect Obama or will they prove to be a racist people?
"This was the comment that started the line on Obama's campaign. It appears reasonable to take up the question of whether a public vote against Obama is the result of racism."
The person was making fun of your false dichotomy. By presenting you with an equally as absurd one. They were not trying to raise the question of race in the campaign. They were trying to make a point in the other conversation.
Can you really not follow that?
They were not trying to raise the question of race in the campaign.
While this person's intent was presumably not to raise race and the presidential election, he/she, nonetheless, did.
7:02 am
My mistake is thinking that quants want to, or can, discuss politics. Never forget that quantoids advocate non-political political science.
It's the Perestroika-Quantoid Cage Match. Rock on!!!
Let's picture it Senoir Prof X caged with Senior Prof Y in a down and dirty fly of quantoid-interpretive method word rage. FUN!
I think a cage match would take care of many of these problems.
It's the Perestroika-Quantoid Cage Match. Rock on!!!
Let's picture it Senoir Prof X caged with Senior Prof Y in a down and dirty fly of quantoid-interpretive method word rage. FUN!
I think a cage match would take care of many of these problems.
There is but one man who should represent the quants in such a battle... Mike Munger!
Hot! I'll bring beer!
But who would dare to represent the Perestroikans?
Do you deny that you can not follow a conversation, and that you mistake a rhetorical device in one conversation for a separate conversation?
Zing!
While this person's intent was presumably not to raise race and the presidential election, he/she, nonetheless, did.
Folks, don't let the ILP redirect the conversation to avoid the issue at hand--i.e., the undeniable fact that he came up with yet another false dichotomy.
Of course he realizes he got caught. So what does he do? He makes this a debate about Obama--circumventing the intent of the very post that introduced Obama to the discussion.
Remember, ILP is the one who couldn't get himself to admit that he just didn't understand the logic inherent in the measurement of temperature.
He will do anything to avoid facing his own shortcomings...
He will do anything to avoid facing his own shortcomings...
Yes, because the ILP is clearly so intellectually superior to anyone else on this blog that we need to force him to face his own shortcomings. Otherwise, his genius will swell his head. (ha . . . ha . . . ha . . .)
Question on endogeniety (no job rumors I’m just looking for free statistical advice). When analyzing infant mortality rates (the dependant variable) I include independent variables: low birth weight, prevalence of acute respiratory infection (ARI), female literacy, economic development, governmental health care expenditures, foreign aid, etc. But, governmental health care expenditures not only influence infant mortality but also low birth weight and ARI. So, the answer is to find a instrumental variable that highly correlates with government health care expenditures, has not affect on infant mortality rates and is uncorrelated with the error term. Seriously is such a thing possible—if I know how to reduce the error term by adding another variable I would? I’ve seen articles where the author simply lags the independent variables (oh yes, I have panel data) in order to get them ahead of the dependent variable.
To move the problem further, I am sure economic development and foreign aid actually affect the level of government health care expenditures—so do I need to instrument government health care expenditures also?
and if you have the STATA 10 commands that would be even more helpful!
You are not describing a problem of endogeneity, you're describing a complex causal relationship in which a whole bunch of Xs affect each other in addition to affecting Y. Instrumental variables are not the appropriate strategy here. An instrument will correct things like true endogeneity (where infant mortality rates affect your independent variables) and measurement error (if your independent variables are measured poorly).
Now, I would also suggest that you probably DO have a problem of endogeneity, but that's not the question that you asked.
1:58's question is a perfect example of the quants' over-reliance on obscure statistical techniques. Instead of asking big questions that pertain to democracy, justice, and fairness, quantoids are obsessed with perfecting their quantitative analytical tools that tend to obfuscate--rather than illuminate--important questions.
This makes ILP's point even more cogent.
Instumental variables are obscure quantoid techniques.
re: 3:33
"An instrument will correct things like true endogeneity (where infant mortality rates affect your independent variables)"
So true endogeneity only includes reciprocal causaltiy?
How does one determine what would make a good instrament?
When analyzing infant mortality rates (the dependant variable) I include independent variables: low birth weight, prevalence of acute respiratory infection (ARI), female literacy, economic development, governmental health care expenditures, foreign aid, etc.
This is not a political science question. It is a public health matter.
1:58's question is a perfect example of the quants' over-reliance on obscure statistical techniques. Instead of asking big questions that pertain to democracy, justice, and fairness, quantoids are obsessed with perfecting their quantitative analytical tools that tend to obfuscate--rather than illuminate--important questions.
This makes ILP's point even more cogent.
Quiet, George. Adults are talking.
This is not a political science question. It is a public health matter.
Sorry you feel that way. But this question was not directed at you, so why are you responding?
When analyzing infant mortality rates (the dependant variable) I include independent variables: low birth weight, prevalence of acute respiratory infection (ARI), female literacy, economic development, governmental health care expenditures, foreign aid, etc.
This is not a political science question. It is a public health matter.
When you don't have anything constructive to contribute, the best task is to whine about it. Good jerb, qually!
Sorry you feel that way. But this question was not directed at you, so why are you responding?
It is reasonable to openly wonder why someone would bring a public health matter to a political science blog?
It is reasonable to openly wonder why someone would bring a public health matter to a political science blog?
This is because many/most quants do not practice political science, but are instead practicians of advanced statistics.
It is reasonable to openly wonder why someone would bring a public health matter to a political science blog?
The author wants to understand the link between government health care expenditures and infant mortality. I can think of no nobler endeavor for a political scientist than trying to understand how political institutions can improve the human condition. Too bad you feel otherwise.
I can think of no nobler endeavor for a political scientist than trying to understand how political institutions can improve the human condition.
The issue raised is one for public health experts, not political scientists. If you think this is a noble endeavor, then you should become a public health specialist.
Would it help if I add that infant mortality rates is a common proxy for poverty? Is poverty and economic variables (FDI, trade, foreign aid)of a political interest?
How else can one examine government/international policy outcomes affect human development without the use of statistics?
How else can one examine government/international policy outcomes affect human development without the use of statistics?
What public policies are you proposing to study? On the question of infant mortality, again there are public health experts who specifically study public health policies and infant mortality. Additioanlly, they are more appropriately trained to study such matters than political scientists.
The issue raised is one for public health experts, not political scientists. If you think this is a noble endeavor, then you should become a public health specialist.
I'm a political scientist who cares about the human condition. So I will do research that allows me to address these issues. I will make no apologies.
Political scientists who refuse to engage with policy relevant research will be consigned to the dustbin of history. I know where I stand. Where do you stand, qually?
Instead of asking big questions that pertain to democracy, justice, and fairness
Good to know that in your world, infant mortality rates aren't related to justice or fairness, or that the question of "Which babies do we not care about if they're dying" is not a big one.
tend to obfuscate--rather than illuminate--important questions
So you're saying that infant mortality isn't an important question?
I did not write 9:24 a.m., so I am not going to defend it.
So you're saying that infant mortality isn't an important question?
Of course it is. Nevertheless, a study of how the level of public health expenditures effects infant mortality is not germane to political science. The fact that the dependent variable of the study is infant mortality means that the proposed model is outside the purview of political science.
Part of it is the question of training. Graduate training in political science does not normally include courses on infant related pathologies, and relevant treatments.
The matter of the relationship between public health expenditures and infant mortality is not seemingly a matter of politics. So, for instance, the politics of a community could be to maximize public health spending (however that might be determined). This would be a political decision that warrants study by political scientists. The matter of to what extent this level of spending actually impacts infant morality does not warrant study by a political scientist. Again, this is a matter for experts in public health to contemplate; and they do.
The matter of the relationship between public health expenditures and infant mortality is not seemingly a matter of politics.
This claim is simply ludicrous. Politics shapes public health expenditures. Do you deny this?
It's true that qualitative methods do not contribute much to such research, but that just further impugns their utility for answer the big political science questions of justice and equality.
Politics shapes public health expenditures. Do you deny this?
You are making my point for me. How does the study proposed enhance our understanding of the politics that shapes public health expenditures?
ILP is correct. Political science should be interested in the causes of policies. The effects of said policies should be examined by experts in the relevant policy areas. For example, unless you have been trained in environmental sciences (or a connected field), a political scientist does not possess the appropriate tools to examine the effects of environmental policy on air quality.
You are making my point for me. How does the study proposed enhance our understanding of the politics that shapes public health expenditures?
You're confused. The study examines the effect of policy on health outcomes.
You're confused. The study examines the effect of policy on health outcomes.
This is exactly the point. It examines the effects, which does not contribute to the understanding of the causes. It therefore falls outside the purview of political science.
This is exactly the point. It examines the effects, which does not contribute to the understanding of the causes. It therefore falls outside the purview of political science.
Your apparent definition fails to recognize that political science must be concerned with the major problems confronting society: questions of democracy, justice, and equality.
I am forced to wonder if "political scientists" who refuse to tackle these questions can be called political scientists at all.
I am forced to wonder if "political scientists" who refuse to tackle these questions can be called political scientists at all.
Sure they can. They just cannot be called "environmental scientists," "public health specialists," or "demographers." Which, for most political scientists, is accurate.
How does the study proposed enhance our understanding of the politics that shapes public health expenditures?
We don't know anything about the study or the theories it posits.
We only know that one of the two following is true:
(1) The study has a theory such that one observable implication of that theory has something to do with the connection between spending and infant mortality, and the researcher hopes to confirm that observable implication.
(2) The study is set up in such a way that a reasonable alternative theory has an observable implication that says something about the relationship between infant mortality and spending, and the researcher hopes to disconfirm that observable implication.
You do not know what the study actually hopes to find. You do not know how the study attempts to go about demonstrating whatever its main findings are. You only know that one piece of the puzzle has something to do with an empirical model of infant mortality.
Sure they can.
First you wrote that this study does not concern itself with "big questions that pertain to democracy, justice, and fairness."
Then you wrote that this is OK.
I see where you stand, qually. And I'm disappointed in you.
You do not know what the study actually hopes to find. You do not know how the study attempts to go about demonstrating whatever its main findings are. You only know that one piece of the puzzle has something to do with an empirical model of infant mortality.
The question remains as how a study that purports to establish a finding between public health expenditures and infant morality can be conceived as political science? If the matter is what factors drive infant morality, then on its face this is a public health issue, not a political science one.
RE: 2:27 p.m.
I do not know why you think a qual wrote that. It sounds the reasoning of someone that wants to mock the qual critique of quantitative work.
If the matter is what factors drive infant morality, then on its face this is a public health issue, not a political science one.
If the matter is the role of government expenditures, then it is clearly a political science question.
If the matter is the role of government expenditures, then it is clearly a political science question.
It appears that you do not understand what science is. It is about determining why. Why is infant mortality at the rate it is? That is a scientific inquiry, and one that is answered by public health experts, not political scientists. For the political scientist a central matter is why does the state do what it does?
Political scientists answer many questions. What are the effect of politics is clearly one of them. These effects may include many things: health is one of them, but also the economy, society, technology, and so forth. This is the essence of political science: what are the consequences of politics?
Here is a prominent article in the APSR that uses health outcomes as a dependent variable. http://www.jstor.org/pss/3118203. It appears that I am correct, and that you are not.
Here is a prominent article in the APSR that uses health outcomes as a dependent variable.
You make a good point with this article. It does significantly stretch the boundaries of political science. The only justification I can see for this piece appearing in APSR is because it deals with war, traditionally a subject treated by political science/history.
Political scientists answer many questions. What are the effect of politics is clearly one of them. These effects may include many things: health is one of them, but also the economy, society, technology, and so forth.
I can see how health, the economy, technology, etc. could be the subject of political science. The model proposed earlier, however, does not qualify as political science. More specifically, where are the politics in arguing that infant morality is (or is not) a function of public health expenditures?
The only justification I can see for this piece appearing in APSR is because it deals with war, traditionally a subject treated by political science/history.
So if a political science article uses a political phenomenon as an independent variable, it's political science.
So a paper using public health expenditures as an independent variable is political science. Done. Thanks, George!
The question remains as how a study that purports to establish a finding between public health expenditures and infant morality can be conceived as political science?
In your terms, by having a theory that is itself "political science," but that has at least one observable implication about infant mortality. I don't know what such a theory might be; I'm just a caveman and your modern things frighten and confuse me.
But there's one thing I do know: making a leap from "I'm looking for information about empirically modeling X" to "X IS THE FOCUS OF YOUR STUDY AND IT IS THEREFORE NOT POLITICAL SCIENCE" is simply unwarranted.
Do I really need to explain to you that you can look for observable implications of your theory that are essentially spinoffs, and that don't have much to do with your core causal story itself, and that confirming these things can make your causal story more believable, especially if the things you're looking for are counterintuitive?
You are overlooking the specific part of 2:30 p.m.:
If the matter is what factors drive infant morality, then on its face this is a public health issue, not a political science one.
Also, Re: 7:25 a.m.
My name is not George!!!
So if a political science article uses a political phenomenon as an independent variable, it's political science.
If we accept this supposition, then there is no limit to the remit of political science. If the discipline explains everything, it runs the risk of explaining nothing.
If we accept this supposition, then there is no limit to the remit of political science. If the discipline explains everything, it runs the risk of explaining nothing.
Certainly wrong. Political science is the study of political phenomena as either causes or effects.
You insist on studying only the latter, which is peculiar, closeminded, and wrong.
Qually, you really took a ride on the failboat this time.
The presidential address at the 2007 APSA Annual Meeting has four entire paragraphs about how political science should contribute to public health questions!
So, Robert Axelrod, one of the most famous political scientists ever and then president of the leading political science professional organization, isn't advocating for political scientists to do political science?
I think careful comparison can establish why the effects of war on public health can be deemed political science, while solely analyzing the effect of public health expenditure levels on infant morality is not. In the case of war (i.e., politics through a different means) it is fairly patent how public health is adversely affected: through the destruction/disruption of food and water related infrastructures, medical services, etc.
In the instance of public health expenditures and infant morality the relationship is much more murky. Do public health expenditures include spending on food and clean water? Are particular medical procedures (e.g., abortion or HIV treatments) excluded from public expenditures? Moreover, which of these tacks have the most beneficial effect on infant morality?
I would argue that the first two questions are clearly in the purview (and training) of political scientists. Decisions on these matters are clearly the result of politics. Whereas the third question falls outside the realm of political science, and into the area of medical science and public health research.
Put differently, public health experts are in the best position to determine which types of public health expenditures (and at what levels) improve infant morality, whereas political scientists bring no particular expertise to this matter.
Therefore, the ILP’s position is that the result of war (i.e., destruction, death) is the result of politics (because war is a type of politics). In contrast, the results of public health expenditures on infant morality is not necessarily the result of politics (e.g., insufficient medical treatments to deal with particular ailments). Alternatively, it is up to medical science and public health research to determine which forms and what levels of public health expenditures improve infant health. Interesting . . .
"Politics is who gets what, when, and how." Harold Dwight Lasswell
Surely dead babies qualifies as those who did not get...so its a political science question.
"Politics is who gets what, when, and how."
You are proving my point. Let us take the early days of the AIDS pandemic. Suppose billions of dollars was spent on treating the ailment. But because there was no effective treatment for the disease, these expenditures were not reducing infant morality. Thus, infants (or their representatives) and other victims of AIDS where determining "what, when, and how", but this politics is not reflected in infant morality.
You are proving my point. Let us take the early days of the AIDS pandemic. Suppose billions of dollars was spent on treating the ailment. But because there was no effective treatment for the disease, these expenditures were not reducing infant morality. Thus, infants (or their representatives) and other victims of AIDS where determining "what, when, and how", but this politics is not reflected in infant morality.
Suppose the opposite: that there is a cure for maternal malnutrition, which reduces infant mortality, and government health expenditures provide support for this. Now politics IS affecting infant mortality.
Suppose the opposite: that there is a cure for maternal malnutrition, which reduces infant mortality, and government health expenditures provide support for this. Now politics IS affecting infant mortality.
The broader point is the study as originally proposed does not capture any of this nuance. Instead, it is a ham-fisted and misguided effort to posit infant morality as a function of public health expenditures writ large.
Actually the original study used public health expenditures as one of many variables. I think it included foreign aid, FDI etc (international variables of globalization). Also, social indicators like female literacy. Perhaps a bit naive but hardly ham-fisted and misguided.
And, might I add, no one actually answered his/her question about instrumental variables.
And, might I add, no one actually answered his/her question about instrumental variables.
That's the sad part, really.
As a result, you can be sure that no one will ever ask a methods question again.
Does the qually *need* to weigh in on every single methodological issue of our time?
As a result, you can be sure that no one will ever ask a methods question again.
So no quant will ever posit methodological questions on this blog again because their work/ideas will come under scrutiny. Coming from a defender of quant research, this is a particularly interesting and telling claim.
"As a result, you can be sure that no one will ever ask a methods question again."
Well if they want their question answered there is a great place to ask it. The PolMeth list is open to anyone. You can subscribe to the list and post questions, even anonymously. Anyone doing quant work can learn a thing or two just by reading the questions and answers.
http://polmeth.wustl.edu/polmeth.php
No George, you misunderstand. People will refrain from posting methods questions on this board because you claim that their work isn't political science rather than answering their questions.
An analogy: I wouldn't ask you about quantitative methods; just as I wouldn't ask Hitler about democracy.
7:08 a.m.,
You are confirming the worst stereotypes of quants. "As long as the methods are 'right', it does not matter if the study tells us anything." Alternatively, "quants are all [quant] methods, and no substance."
P.S.: I am not George.
You are confirming the worst stereotypes of quants. "As long as the methods are 'right', it does not matter if the study tells us anything." Alternatively, "quants are all [quant] methods, and no substance."
You never once claimed that the study didn't tell us anything. Or that the methods were inappropriate. You only claimed that it's not political science.
Fail.
George has a real problem here. He can't stand quantitative methods and the advancement of science, so his last hope is to try to claim that any study that he doesn't like "isn't political science."
It's a weak attempt to rescue something out of nothing, which is exactly what this blog shows his argument to have boiled down to.
Thank you for the PolMeth web address
He can't stand quantitative methods and the advancement of science
Quals take issue with quants because what they practice for the most part is not science. How can a study that is "ham-fisted" or "naive" (as a sympathetic commentator put it) be good science (political or otherwise), or tell us anything about the real world.
If I were a quant, I would be the first to critique a study that gives quant methods a bad name. But because there is no scientific core to most quant research, quants are solely drawn together in political solidarity (not scientific solidarity).
What you are ignoring, qually, is that the very first response was a constructive criticism of the proposed research design by a quant.
In other words, you can't even read. Fail.
You are not describing a problem of endogeneity, you're describing a complex causal relationship in which a whole bunch of Xs affect each other in addition to affecting Y. Instrumental variables are not the appropriate strategy here. An instrument will correct things like true endogeneity (where infant mortality rates affect your independent variables) and measurement error (if your independent variables are measured poorly).
Now, I would also suggest that you probably DO have a problem of endogeneity, but that's not the question that you asked.
This does not began to get at the profound problem embedded in the proposed study. Instead, this comments ignores the substantive and significant flaws of the model, and focuses on the technical components of the study. It is a true (superficial) quant observation.
The poster responded to the original question by showing how the question was inappropriate. What's wrong with that?
Someome simply wanted advice on how to deal with his/her endogenous variables. I'm just saying that as a practical matter, folks will refrain from posting here in the hope that they get useful responses.
I'm not saying that you (or anyone else) isn't allowed to question a study. I'm just a little disappointed that due to the exchange that ensued, folks will refrain from posting methods question on this thread.
No need to get into one of your rants about quants and stereotypes, really, I'm just making a small point. I guess that I should have expected that my post would be met with a defensive post by our insecure qually...
ILP, question for you.
Yesterday, we saw a really important political debate surrounding an important piece of legislation. Ultimately, for various reasons (which, I know, you will say that we need cognition to understand them, blah, blah) the bill failed. If you juxtapose the DOW (or S&P) to the timeline of the bill, you will see that it started falling just about the time that it became clear that the bill wouldn't pass (i.e., not after the final tally, but during the rollcall when it became clear that the Noes would probably have it).
Anyway, would you consider the question of how the passage (or failure) of legislation in Congress affects the economy (in this case measured by the performance of various stock indices) to fall outside the purview of political science?
Before responding, keep in mind that politics is interested in who gets what, when, how (at least that's one plausible definition). A lot of people took hits, 401k, stock portfolios, mutual funds, etc. So lots of people got what=less; when=yesterday early afternoon; how=devaluation of stocks. So do you think that understanding whether this bill (or any other bill) causes these sorts fluctuations isn't political science?
I guess that I should have expected that my post would be met with a defensive post by our insecure qually...
So your concern is that quants will not post their methodology questions on this blog because their research will fall under scrutiny. And the qual is called insecure!!!
Amazing reasoning that could only be arrived at by an quant -- who has every reason to be insecure (since his/her work probably has no scientific basis). For the record I always strongly welcome/encourage a critical treatment of/attitude toward my work.
Anyway, would you consider the question of how the passage (or failure) of legislation in Congress affects the economy (in this case measured by the performance of various stock indices) to fall outside the purview of political science?
It depends on the study being proposed.
So your concern is that quants will not post their methodology questions on this blog because their research will fall under scrutiny. And the qual is called insecure!!!
No, the concern is that quants will not post their methodology quetsions on this blog because they won't be answered (though I guess someone offered a response that the model didn't necessarily suffer from endogeneity).
Can't you even follow an argument (even when you disagree with it)?
Let's put it another way. Imagine that there's someone on this blog who repudiates every paper if it's written in English. Someone posts a question, which is immediately followed by two dozen posts discussing whether English is the appropriate language for the paper. Now, the discussion in and of itself might be interesting. I suspect, however, that it wouldn't be too helpful for the poster.
It depends on the study being proposed.
Thank you for enunciating your position. At least you've shown a willingness to modify it when faced with strong counter-arguments. You see, in the past, you said that the effects of politics isn't political science (it's demography, it's health policy). Now, at least, you've open the door. I commend you on being open-minded
You see, in the past, you said that the effects of politics isn't political science (it's demography, it's health policy).
You are being disingenuous, or your position is disingenuous. No qual has has ever been so broad and ham-fisted in their approach.
Can't you even follow an argument (even when you disagree with it)?
Whose fault is it if you cannot clearly articulate your position.
Imagine that there's someone on this blog who repudiates every paper if it's written in English.
This analogy is so "off the wall" that it does not even rise to the level of a straw man. The critique was germane, appropriate, and important.
The qually is on the run now. These two posts demonstrate his inability to follow simple logic.
This analogy is so "off the wall" that it does not even rise to the level of a straw man.
Funny on how that didn't stop you from engaging in a debate about the Obama vote earlier--I guess the claim that the the only reason Obama would lose is due to racism wasn't off the wall for you...
The critique was germane, appropriate, and important.
Tell me how germane and appropriate the critique was in terms of dealing with potential endogeneity issues in the model.
Whose fault is it if you cannot clearly articulate your position.
Now it's just sad... According to your own logic, you admit you couldn't follow the argument, since you now fault me for not being clear.
Fair enough. But why not say earlier that you couldn't follow the argument instead of trying to figure out what my (unclear according to you) post argued?
Tell me how germane and appropriate the critique was in terms of dealing with potential endogeneity issues in the model.
You take an interesting/usual approach to analyzing work. You limit your observations/comments to the areas the author dictates.
You take an interesting/usual approach to analyzing work. You limit your observations/comments to the areas the author dictates.
Not at all. But if I ask someone "where is the nearest SUV dealership," and the response is "you shouldn't buy a SUV," it's not very helpful in helping me locate a dealership, isn't?
If I read a paper, I can be a critic. If someone asks a very specific question, I'm assuming that person is looking for a specific response. If the question had been "should I model it this way, or do you have a better approach", then it's a different story.
So we differ on this one. It's okay, we'll both live long and productive lives.
But if I ask someone "where is the nearest SUV dealership," and the response is "you shouldn't buy a SUV," it's not very helpful in helping me locate a dealership, isn't?
You analogies are way, way off and not at all helpful/appropriate.
A more germane analogy is if someone goes to the doctor complaining about one ailment, and the doctor discovers a different, unrelated affliction, but does not mention it to the patient because that is not what was initially asked about.
A more germane analogy is if someone goes to the doctor complaining about one ailment, and the doctor discovers a different, unrelated affliction, but does not mention it to the patient because that is not what was initially asked about.
Except that in this case, the analogy would be "someone goes to the doctor for expert opinion on an ailment, and the janitor overhears the patient and tells him that he shouldn't bother with the ailment, that he should be treating something else instead."
You see, the problem is that you've demonstrated in the past that you have absolutely no knowledge of statistics. Thus, if someone comes to this blog with a problem related to statistics, you will automatically chime in with an opinion that is not helpful and is based on a lack of knowledge about the statistical issue at hand.
It's not because you have an opinon that it's informative and helpful. You might think otherwise, and that's okay (we often cannot see our own shortcomings, and in your case that's patently true). I wonder whether the original poster thinks the discussion was informative/helpful. Shouldn't s/he make that determination? As a quant, I would think that it's not. But I'm happy to let others tell me why I'm wrong.
I wonder whether the original poster thinks the discussion was informative/helpful. Shouldn't s/he make that determination?
Why?
you will automatically chime in with an opinion that is not helpful
I think the opinion I offered is extremely helpful. If the author cannot see that, doesn't the fault lie with him/her?
As a quant, I would think that it's not.
Isn't that the real problem. That quants tend to overly emphasize quant methods over substance.
I think the opinion I offered is extremely helpful. If the author cannot see that, doesn't the fault lie with him/her?
How can it be the author's fault that you offered an irrelevant opinion?
How can it be the author's fault that you offered an irrelevant opinion?
The discussion that followed the author's initial positing of his/her model demonstrates the relevance of the comments that I offered.
The discussion that followed the author's initial positing of his/her model demonstrates the relevance of the comments that I offered.
No, it showed how posters had to explain to you once again how political science works. And in the end, you were proven wrong, again. How, then, is that relevant?
No, it showed how posters had to explain to you once again how political science works. And in the end, you were proven wrong, again.
You and I drew very different conclusions from that discussion.
Well, like any good qually, evidence doesn't matter much for you.
The following nicely sums up the substantive conclusion of the discussion:
The result of war (i.e., destruction, death) is the result of politics (because war is a type of politics). In contrast, the results of public health expenditures on infant morality is not necessarily the result of politics (e.g., insufficient medical treatments to deal with particular ailments). Alternatively, it is up to medical science and public health research to determine which forms and what levels of public health expenditures improve infant health.
like any good qually, evidence doesn't matter much for you.
Given that this post is not accompanied by any evidence, I find this assertion incredible!
That's because you stopped reading, George. The next post sums it up nicely:
"Politics is who gets what, when, and how." Harold Dwight Lasswell. Surely dead babies qualifies as those who did not get...so its a political science question."
The lesson here: never trust a qually to represent evidence faithfully.
In fact, qually, I think it's time to call you out. J'accuse! I accuse you of willfully ignoring evidence that does not support your arguments. If this were a published paper, it would amount to research fraud.
It is you that is quoting from this blog in self-serving and deceptive manner. This, as you well know, was the response to the post you draw upon:
You are proving my point. Let us take the early days of the AIDS pandemic. Suppose billions of dollars was spent on treating the ailment. But because there was no effective treatment for the disease, these expenditures were not reducing infant morality. Thus, infants (or their representatives) and other victims of AIDS where determining "what, when, and how", but this politics is not reflected in infant morality.
This is fraudulent. You refused to read further because you knew that your point was then absolutely demolished.
So, what have we here? A qually being fraudulent in his research. Nothing new, perhaps, but disappointing for those of us who care about political science.
A qually being fraudulent in his research.
Do you have any evidence to support this claim? Sadly, I have found too many quants that have no interest in the truth, or the world as it actually operates.
RE: 6:29 a.m.
What you also wantonly overlooked was how a quant proponent acknowledged at the end of the discussion how the proposed model was "naive"?
The evidence lies above. You present selective readings of evidence that is plainly available for all to see.
Come on George, tell me now that this evidence is immaterial because I didn't examine cognition.
tell me now that this evidence is immaterial
What evidence are you referring to?
Your cite 9/29/2008 8:51 AM, but fail to cite the next post, 9/29/2008 11:49 AM. The reason is because 9/29/2008 11:49 AM proves your point wrong. This is willfully deceitful, and if this were published, it would amount to fraud.
Your cite 9/29/2008 8:51 AM, but fail to cite the next post
How is this evidence of fraud? Did I claim that 9/29/2008 8:51 AM was the last post on the subject at hand? No! Like many quant researchers, you are not careful or thoughtful in arriving at your conclusions.
The following nicely sums up the substantive conclusion of the discussion:
The result of war (i.e., destruction, death) is the result of politics (because war is a type of politics). In contrast, the results of public health expenditures on infant morality is not necessarily the result of politics (e.g., insufficient medical treatments to deal with particular ailments). Alternatively, it is up to medical science and public health research to determine which forms and what levels of public health expenditures improve infant health.
Well, um, I would actually argue that you are applying different standards here.
You claim that the result of was is the result of politics, whereas infant mortality might be due to many non-political factors (such as insufficient treatments).
Leaving aside for now the question of whether insufficient treatment could also be the result of politics, I'd argue that the result of war can be due to many similarly non-political factors: weather, the availability of state-of-the-art weapons, competing soldiers' willingness to fight and die, the state of the economy (which you don't automatically think is a result of politics--see your earlier answer to the question about the stock market), etc.
So there, the result of war can be due to non-political factors as well.
But because you were stomped by the poster who brought up the APSR article, you made up this flimsy criterion on the fly.
Suppose billions of dollars was spent on treating the ailment. But because there was no effective treatment for the disease, these expenditures were not reducing infant morality. Thus, infants (or their representatives) and other victims of AIDS where determining "what, when, and how", but this politics is not reflected in infant morality.
Wouldn't this finding be of importance to politics? If billions are being spent and there is no reduction in mortality, wouldn't that research be useful when debating whether to continue this funding? I can tell you that such research would be cited on Capitol Hill by opponents of funding. Isn't that politics? Knowing that a policy isn't prodcuing the intended effects?
So according to you, ILP, your own posts are informative/helpful? To whom? You?
Now, you're being a little self-centered here. Of course you think your posts are useful. I think mine are as well. So there.
But how do others see it? It's only helpful if others think it is. Otherwise, it's the tree falling in a forest...
If John McCain says "my speach was great." Does that provide any information as to whether his speech was, indeed, great? Given that he has the incentive to say so?
Wouldn't this finding be of importance to politics? If billions are being spent and there is no reduction in mortality, wouldn't that research be useful when debating whether to continue this funding? I can tell you that such research would be cited on Capitol Hill by opponents of funding. Isn't that politics? Knowing that a policy isn't prodcuing the intended effects?
This would be important to the politics of expenditures on public health. Nevertheless, it demonstrates that the relationship between public health expenditures and infant morality does not provide a reliable indicator of the politics of expenditures on public health.
I'd argue that the result of war can be due to many similarly non-political factors: weather, the availability of state-of-the-art weapons, competing soldiers' willingness to fight and die, the state of the economy
These factors become intermediate factors within the context of the politics of war. Whereas issues of sufficient food, clean water, adequate treatments, the competence of health care professionals, violence, can (normally do) stand outside of (apart from) public health expenditures.
<< Home