The Volokh Conspiracy - Dan Klein Responds to VC Comments:

A few weeks ago I posted on the new paper by Dan Klein and Charlotta Stern, "Narrow-Tent Democrats and Fringe Others: The Policy Views of Social Science Professors." The post spawned a vigorous and interesting discussion in the Comments (available here). It is also relevant to the Leagal Affairs debate that Orin flagged earlier today. I invited Dan to provide a brief response to some of your Comments if he wanted to, and I am delighted that he agreed to do so. Here is his response:

Todd Zywicki posted an entry on Charlotta Stern’s and my paper “Narrow-Tent Democrats and Fringe Others: The Policy Views of Social Science Professors,” to appear in Critical Review: An Interdisciplinary Journal of Politics and Society. The post was followed by 34 comments.

My 2003 survey of members of six leading scholarly associations has gotten a fair amount of attention. The “Narrow-Tent” paper is the big paper coming out of the survey. If you want to read just one paper on the survey, this is the one to read.

This memo is a follow-up to the set of Volokh comments.

Stern and I are open about our libertarian sensibilities. We agree with Gunnar Myrdal that such openness is the best way to avoid bias and advance discourse that involves deep-seated judgments and interpretations. For a review of Myrdal and a justification of disclosure of ideological sensibilities (and of exposé), please see my paper “Sense and Sensibilities.”

But many commenters say “A-ha!, these researchers are libertarians, and that undermines the whole study.” Sorry, but the conclusion doesn’t follow.

Many revert to l-v-c convention (that is, liberal v. conservative) in putting libertarians as conservatives. The paper shows not only that Dems and Greens are highly statist, but also that Repubs are much more statist than some think, and on some issues more statist than the Dems (e.g., immigration, military action, drugs and prostitution restrictions). My agenda is to criticize l-v-c. I think the terms “liberal” and “conservative” are simply degenerative. Other than code (in the US context) for Dem and Repub, they are terribly fuzzy. Our research tacitly criticizes Republicans and conservatism (in a narrow sense that is different from libertarianism), and I think that is one reason that conservatives ignore many of our findings. For example, the cluster analysis shows that conservatives (in the narrow sense made clear in the paper) are closer to the establishment left and even to the progressives than they are to the libertarians. This finding is remarkable. The first division in the cluster analysis is between the tiny, tiny number of libertarians (of which 40% vote Republican and 11% vote Democratic) and the vast mass of all others who are mostly significantly more statist.

For libertarians, the voluntary/coercive distinction is not only meaningful, but gives rise to important analytic structure in categories and formulations. The distinction is not black or white, but is a matter of degree. My goal is to make people on both the left and the right face up to the fact that they are much more agreeable to coercive policy than they are usually prepared to admit.

Responses to specific points:

1. The Dem’s tent is narrow. The study finds that over the 18 policy questions, the Dems have a narrower tent than the Repubs. (Note, one is a “Dem” based on saying he mostly votes for Dem candidates, not based on being registered Democratic, and likewise for Repubs.) Specifically, the sum of 18 Dem-group standard deviations is 17.1, while the sum for the Repub-group is 23.1. Thus, the sum is 35% larger for the Repubs. The greater diversity of the Repub tent is also shown by plotting the policy index scores for each individual in each group. The Republican group shows significantly more spread (pp. 17-19). Commenters including Gabriel Rossman, Bruce Wilder, and Justin challenge our finding, saying that the result comes from the libertarian selection and formulation of the 18 policy questions. They do not make the case, however. The questions are not remarkable, and the scoring is granular, based on a continuous scale, not a binary categorization into “statist” and “libertarian”. I really don’t see anything about the 18 questions that would deflate the diversity of the Dems or puff it up for the Repubs. Think about it, very few Dems really favor free enterprise and all favor redist/welfare state/protect the weak type policies. I suspect that in the mind of those on the left, the differences between progressives and the establishment left seem larger than they really are. Meanwhile Repubs have both libertarian types and (narrow) conservative types. Realize that the libertarians are often far-out in their opinions, so they stretch the tent more than their number would make one think. Thus, is it really surprising that the Dems have a significantly narrower tent? In the comments the only substantive challenge comes from Gabriel Rossman: “if they had asked about, say, gay marriage they might have found more uniformity among Republicans.” My guess is that we would find academic Dems to be quite uniformly supportive of gay marriage (as Stern and I are), while academic Repubs being more mixed (again, one must recognize that those with libertarian tendencies tend to vote Republican). My guess is that Rossman’s suggested question would have reinforced the narrow-tent finding.

2. Are Dems and Repubs much different? A commenter named “alkali” looked carefully at the 18 issues and notes: “the poll asked respondents to rate themselves on a 5 point scale (from ‘strongly support’ to ‘strongly oppose’) with respect to 18 particular policies. On 8 of the 18 issues, the average Dem is less than 1 point away from the average Rep. On no issue are the average Dem and Rep more than 2 points apart.” Two points. First, the “5 point scale” has a measure of only 4 units, so a difference of 1 is a difference equaling 25% of the scale, not 20%. The differences are somewhat bigger than it might seem. But, second, “alkali” is pointing out something that we highlight in the paper, even in the abstract, namely, that the Republicans are generally also quite supportive of status-quo interventions. Very few are serious supporters of individual liberty. Again, the cluster analysis shows that the (narrowly defined) conservatives are closer to the establishment left and progressives than they are to the libertarian cluster.

3. Regarding the libertarian formulation of the policy questions. Each policy question posits some form of government intervention and then asks degree of support/opposition. It is deliberately constructed to provide a uniform format suitable for creating an overall policy index of degree of statism v. libertarianism. Is there something wrong with that? I think that is the analytic contribution of the study. Many on the left object to it, but frankly I think that is because this method exposes their degree of statism. Gabriel Rossman writes: “the questions are horribly worded and reflect an extreme libertarian bias (for better or worse, things like legalizing heroin and eliminating progressive income tax simply aren't on the table politically). the proper thing to do would have been to adopt questions from the GSS, both because these questions are well tested and also because this gives a solid baseline of the general population.” While I agree that getting comparisons to the general population would be a great benefit of using GSS or other commonly used questions, the problem is that those questions do not get at the core analytic issue of statism v. libertarianism (coercive v. voluntary action). Most surveys are mired in the l-v-c swamp. The big difference between my research and practically all other work on political culture is that I avoid the l-v-c swamp.

4. Do the policy questions favor the Republicans? I do not think that the set of 18 questions are slanted so as to overstate the statism of the Dems and understate the statism of the Repubs. The whole realm of redistribution/welfare state, which the Dems are more statist on, is limited to just one question. Meanwhile, there are three separate questions for drugs, prostitution, and gambling, which are personal-liberties questions I had hoped to see Democrats be significantly more libertarian than the Republicans on (on the whole, they were only somewhat more, alas). Admittedly, the whole realm of military action is limited to just one question, but, hey, that issue is not really straight forwardly statist v. libertarian anyway (similarly with one other question, namely monetary policy), as there is a libertarian flavor to, say, toppling a Saddam Hussein (although Stern and I tend to lean against the Iraq invasion). On the whole, I feel that the findings about relative statism of Dems v. Repub voters are reasonable. Notice that this is about voters’ views, not regimes’ deeds. That the Republicans now in power in Washington have in fact been inimical to individual liberty is not relevant to what we are discussing.

5. Is Democratic really leftist? Justin has a point here, but I think it is pretty clear that we are focusing on the US context and that everything is dominated by the status quo.

6. Why is the academy dominated by the left? This question is tossed around by the commenters, often insinuating that we took a stand on the question. But Stern and I say at the start of the paper that we do not attempt to explain it. I do hope, however, that people observe our contribution to the explanandum: We provide knock-down evidence that Republican-voting PhDs who are members of the leading scholarly association in their field (American Historical Association, etc.) are significantly more likely to land outside of academia in the fields of sociology and history, and same with weaker effect in economics, political science, and legal and political philosophy. Also, across all six disciplines studied, in the multivariate regressions, the correlation between voting R and landing outside academia consistently holds up at 0.01 (see pp. 22 and 34).

I thank Todd Zywicki and The Volokh Conspiracy for hosting this discussion, and those who commented on paper for taking the time.

Daniel Klein, Professor of Economics, George Mason University

	[Todd Zywicki, January 23, 2006 at 5:31pm] Trackbacks Dan Klein Responds to VC Comments: A few weeks ago I posted on the new paper by Dan Klein and Charlotta Stern, "Narrow-Tent Democrats and Fringe Others: The Policy Views of Social Science Professors." The post spawned a vigorous and interesting discussion in the Comments (available here). It is also relevant to the Leagal Affairs debate that Orin flagged earlier today. I invited Dan to provide a brief response to some of your Comments if he wanted to, and I am delighted that he agreed to do so. Here is his response: Todd Zywicki posted an entry on Charlotta Stern’s and my paper “Narrow-Tent Democrats and Fringe Others: The Policy Views of Social Science Professors,” to appear in Critical Review: An Interdisciplinary Journal of Politics and Society. The post was followed by 34 comments. My 2003 survey of members of six leading scholarly associations has gotten a fair amount of attention. The “Narrow-Tent” paper is the big paper coming out of the survey. If you want to read just one paper on the survey, this is the one to read. This memo is a follow-up to the set of Volokh comments. Stern and I are open about our libertarian sensibilities. We agree with Gunnar Myrdal that such openness is the best way to avoid bias and advance discourse that involves deep-seated judgments and interpretations. For a review of Myrdal and a justification of disclosure of ideological sensibilities (and of exposé), please see my paper “Sense and Sensibilities.” But many commenters say “A-ha!, these researchers are libertarians, and that undermines the whole study.” Sorry, but the conclusion doesn’t follow. Many revert to l-v-c convention (that is, liberal v. conservative) in putting libertarians as conservatives. The paper shows not only that Dems and Greens are highly statist, but also that Repubs are much more statist than some think, and on some issues more statist than the Dems (e.g., immigration, military action, drugs and prostitution restrictions). My agenda is to criticize l-v-c. I think the terms “liberal” and “conservative” are simply degenerative. Other than code (in the US context) for Dem and Repub, they are terribly fuzzy. Our research tacitly criticizes Republicans and conservatism (in a narrow sense that is different from libertarianism), and I think that is one reason that conservatives ignore many of our findings. For example, the cluster analysis shows that conservatives (in the narrow sense made clear in the paper) are closer to the establishment left and even to the progressives than they are to the libertarians. This finding is remarkable. The first division in the cluster analysis is between the tiny, tiny number of libertarians (of which 40% vote Republican and 11% vote Democratic) and the vast mass of all others who are mostly significantly more statist. For libertarians, the voluntary/coercive distinction is not only meaningful, but gives rise to important analytic structure in categories and formulations. The distinction is not black or white, but is a matter of degree. My goal is to make people on both the left and the right face up to the fact that they are much more agreeable to coercive policy than they are usually prepared to admit. Responses to specific points: 1. The Dem’s tent is narrow. The study finds that over the 18 policy questions, the Dems have a narrower tent than the Repubs. (Note, one is a “Dem” based on saying he mostly votes for Dem candidates, not based on being registered Democratic, and likewise for Repubs.) Specifically, the sum of 18 Dem-group standard deviations is 17.1, while the sum for the Repub-group is 23.1. Thus, the sum is 35% larger for the Repubs. The greater diversity of the Repub tent is also shown by plotting the policy index scores for each individual in each group. The Republican group shows significantly more spread (pp. 17-19). Commenters including Gabriel Rossman, Bruce Wilder, and Justin challenge our finding, saying that the result comes from the libertarian selection and formulation of the 18 policy questions. They do not make the case, however. The questions are not remarkable, and the scoring is granular, based on a continuous scale, not a binary categorization into “statist” and “libertarian”. I really don’t see anything about the 18 questions that would deflate the diversity of the Dems or puff it up for the Repubs. Think about it, very few Dems really favor free enterprise and all favor redist/welfare state/protect the weak type policies. I suspect that in the mind of those on the left, the differences between progressives and the establishment left seem larger than they really are. Meanwhile Repubs have both libertarian types and (narrow) conservative types. Realize that the libertarians are often far-out in their opinions, so they stretch the tent more than their number would make one think. Thus, is it really surprising that the Dems have a significantly narrower tent? In the comments the only substantive challenge comes from Gabriel Rossman: “if they had asked about, say, gay marriage they might have found more uniformity among Republicans.” My guess is that we would find academic Dems to be quite uniformly supportive of gay marriage (as Stern and I are), while academic Repubs being more mixed (again, one must recognize that those with libertarian tendencies tend to vote Republican). My guess is that Rossman’s suggested question would have reinforced the narrow-tent finding. 2. Are Dems and Repubs much different? A commenter named “alkali” looked carefully at the 18 issues and notes: “the poll asked respondents to rate themselves on a 5 point scale (from ‘strongly support’ to ‘strongly oppose’) with respect to 18 particular policies. On 8 of the 18 issues, the average Dem is less than 1 point away from the average Rep. On no issue are the average Dem and Rep more than 2 points apart.” Two points. First, the “5 point scale” has a measure of only 4 units, so a difference of 1 is a difference equaling 25% of the scale, not 20%. The differences are somewhat bigger than it might seem. But, second, “alkali” is pointing out something that we highlight in the paper, even in the abstract, namely, that the Republicans are generally also quite supportive of status-quo interventions. Very few are serious supporters of individual liberty. Again, the cluster analysis shows that the (narrowly defined) conservatives are closer to the establishment left and progressives than they are to the libertarian cluster. 3. Regarding the libertarian formulation of the policy questions. Each policy question posits some form of government intervention and then asks degree of support/opposition. It is deliberately constructed to provide a uniform format suitable for creating an overall policy index of degree of statism v. libertarianism. Is there something wrong with that? I think that is the analytic contribution of the study. Many on the left object to it, but frankly I think that is because this method exposes their degree of statism. Gabriel Rossman writes: “the questions are horribly worded and reflect an extreme libertarian bias (for better or worse, things like legalizing heroin and eliminating progressive income tax simply aren't on the table politically). the proper thing to do would have been to adopt questions from the GSS, both because these questions are well tested and also because this gives a solid baseline of the general population.” While I agree that getting comparisons to the general population would be a great benefit of using GSS or other commonly used questions, the problem is that those questions do not get at the core analytic issue of statism v. libertarianism (coercive v. voluntary action). Most surveys are mired in the l-v-c swamp. The big difference between my research and practically all other work on political culture is that I avoid the l-v-c swamp. 4. Do the policy questions favor the Republicans? I do not think that the set of 18 questions are slanted so as to overstate the statism of the Dems and understate the statism of the Repubs. The whole realm of redistribution/welfare state, which the Dems are more statist on, is limited to just one question. Meanwhile, there are three separate questions for drugs, prostitution, and gambling, which are personal-liberties questions I had hoped to see Democrats be significantly more libertarian than the Republicans on (on the whole, they were only somewhat more, alas). Admittedly, the whole realm of military action is limited to just one question, but, hey, that issue is not really straight forwardly statist v. libertarian anyway (similarly with one other question, namely monetary policy), as there is a libertarian flavor to, say, toppling a Saddam Hussein (although Stern and I tend to lean against the Iraq invasion). On the whole, I feel that the findings about relative statism of Dems v. Repub voters are reasonable. Notice that this is about voters’ views, not regimes’ deeds. That the Republicans now in power in Washington have in fact been inimical to individual liberty is not relevant to what we are discussing. 5. Is Democratic really leftist? Justin has a point here, but I think it is pretty clear that we are focusing on the US context and that everything is dominated by the status quo. 6. Why is the academy dominated by the left? This question is tossed around by the commenters, often insinuating that we took a stand on the question. But Stern and I say at the start of the paper that we do not attempt to explain it. I do hope, however, that people observe our contribution to the explanandum: We provide knock-down evidence that Republican-voting PhDs who are members of the leading scholarly association in their field (American Historical Association, etc.) are significantly more likely to land outside of academia in the fields of sociology and history, and same with weaker effect in economics, political science, and legal and political philosophy. Also, across all six disciplines studied, in the multivariate regressions, the correlation between voting R and landing outside academia consistently holds up at 0.01 (see pp. 22 and 34). I thank Todd Zywicki and The Volokh Conspiracy for hosting this discussion, and those who commented on paper for taking the time. Daniel Klein, Professor of Economics, George Mason University