The LEO Test

Glimpses of Application: Relative Centers and the Lens Effect

With renewed energy, now that I have largely reorganized my office to conform with the overall structure of my courses and other responsibilities at the college, I focus my attention on the LEO model.  This particular entry will seem unpolished, but it is evident to this author that the idea cannot, and must not, wait. 

Measuring the overall ideological preferences of a given community, and describing that community in terms of political ideological space, grants several intriguing opportunities for application.  In particular, one "problem" that emerged from efforts to collect a purposive sample of subjects to submit to the LEO Inventory was that of the disconnect between perceived ideological preference and measured ideological preference.  This disconnect appeared to extend to the test subject's impression of adherents to actors who held to other ideological preferences.   This suggests a distortion of perception, not only of others observed by the subject, but of the subject himself.   In order to explain why this might be so, a critical sample will need to be collected of individuals within a given community, so that a 'relative center' may be identified.  This relative center will likely deviate from the 'true' center within LEO space itself, and one may be able to measure whether those within the community who tend towards that center will be viewed and will consider themselves 'centrists', unless the members of the community generally are conscious of the ideological preference of the community as a whole.  Those on the periphery may perceive themselves as radicals or extremists oriented in the direction of the nearest quadrant of LEO space, or else across the threshold of one of the major descriptive axes. 

I consider this hypothesis after reflecting on a student volunteer who described himself as an 'anarchist.' He was brought up within a strongly 'conservative' household and made a conscious decision to rebel once he decided that his family adhered to order for order's sake.  His contrarian opinions were met with hostility and members of his own household labeled him an anarchist, and so be believed himself to be.  Once he took the LEO Inventory, however, his preference was revealed to be that of a moderate left libertarian, rather than an anarchist, and he needed an explanation.  A subsequent interview suggested that his household was heavily oriented around order, and that most of its members believed their position was in the "center" of the "mainstream" of the American body politic.  If this be so, then the community that represented that particular household may affect the perception of ideological preference space in much the same manner as a lens affects one's optical perception of other objects in space.  This particular simile has this author wondering if this lens effect is characteristic of political communities in LEO space. 

In order to test this hypothesis, a major project would have to be undertaken.  First, one must describe the lens.  To do this, at least three variables must be measured:  1. a community's aggregate political ideological preference in LEO space. 2. the self-perception of a sample within that community regarding their own political preferences. 3. the perception of that sample of respondents by the community itself.

Second, one must observe a political actor through different positions within that community, and compare the community's perceptions with that actor's actual measured position in LEO space.  One must find either the test community's aggregate perception or mean perception of a particular political actor whose LEO preference is already established.  Then, individuals within the test community must report their perceptions of the ideological preference of the political actor under scrutiny, without the members of the community being aware of the LEO score itself.   The differences between the individuals' perception of the external political actor and the community perception of that same actor should vary with the difference of preference between the individuals self-perceived preference and the mean preference score of the community as a whole.  If it does, then a community's average LEO preference may be viewed as a lens in LEO space, and we may actually describe differences between one's perception of a political actor's ideological preference and their actual measured ideological preference in LEO space as ideological myopia.

Nevertheless,  at this very early stage, in nearly every case of test subjects who took the inventory, a cursory comparison of the subject's inventory scores with the scores gleaned from the campaign literature of presidential candidates accurately predicted the inventory-taker's voting preference in the election under scrutiny, which suggeests that the members of the sample acted in accordance with their actual ideological preferences, and they generally decided to vote or not vote based on their awareness of particular candidate choices and whether they conformed to their own measured ideological preferences or not.  Some voted for their preference, others voted against their opponent, if no match was perceived to be available.  Not voting also appeared to be related to failure to find a LEO match.  One is left to wonder if ideological myopia, if it exists, has any effect on voter choice.

Why It Is Imperative to Continue Work on this Project

This publication completely escaped my attention last year.  Apparently the Pew Research Center is using a variation of the Political Compass/Nolan Test.

http://pewresearch.org/pubs/17/in-search-of-ideologues-in-america

A Long Hiatus Indeed

I have difficulty believing that just over a year has passed since last I updated this project.  My duties at Mountain View College have prevented me from adding new content, but this does not mean that I have not continued work on the project. 

For my Introductory American Government course, I have developed a presentation that deals with many of the issues which were explored here on this project.  For those who, despite the year-long absence, are still returning to this project to see if any new work has been done, I shall include a link here to the presentation.  Placed in the context of political cultures and ideologies, this presentation includes material on traditionalism, individualism, and moralism (the political cultures associated with order, liberty, and equality)  as well as material on comparative ideological frameworks, including the Spectrum, the Compass, and LEO space.  A visual representation of LEO space is included, and it appears to be much more complicated than I had first conceived. 

With the fullness of time, I may be able to include a discussion of relative centers within political communities or constitutional frameworks.  A major hypothesis of the LEO model is, after all, that perceptions of ideological preferences vary with the relative ideological center of any given community.  But testing this hypothesis will have to wait for now.

Please see the link below:

Political Cultures and Ideologies (PowerPoint Presentation)

A Note on Method

It has occurred to me, after perusing some of the commentary advanced elsewhere (particularly Jeremy Jose's insightful comments on his weblog and elsewhere) that I have been remiss in a formal description of the principal methods used in this project.  In truth there are at least two distinct projects act work involving the interpretive framework of the LEO test.  Some results of the first project are available, while the results of the second, an inventory that gathers individual-level data, has not yet been described.  It would seem that the best movement forward may be to explore a few of the methodological points of this inquiry, since time has not permitted the continuing inquiry of individual political actors for which I had hoped.  Hopefully, more will follow...

Work Continues

I only have time to post briefly as I am consistently overwhelmed with the quotidian responsibilities of faculty life.  However, I wish to call attention, for those reader who are still following this project, to a recent application of the LEO test to the partisan landscape in Australia.  Jeremy Jose, whom I credit with exposing the limits of this analytical tool vis a vis judicial decisions, recently decided to revisit the LEO model to see what it misht reveal about Australian political parties.  His results may be found here.

I hope to return to this project soon that I may include some of the more intriguing results from an application of the LEO model to individual respondents.  Hopefully, more should follow soon.

A Brief Note

It is likely that those few readers who had been keeping up with this project have since left, frustrated that no new content has appeared in a long time.  Given the priority with which the author's regular employment has required, not much time could have been devoted to the LEO Test project without compromising the overall tenor of the project itself.  The analysis provided on this project are original to the project itself, and since data gathering is still not automated, the time investment required exceeds my feeble supply.

That said, however, the recent defeat in the Democratic Party primary in Connecticut of incumbent Senator Joe Lieberman, just in time for the end of the second Summer session at my home campus, coupled with commentary on the Donklephant site, has forced this author out of his apparent slumber and demanded that Lieberman be considered.  The visuals have not been prepared yet, but the results are worth noting.  Using five separate measurements, testing the incumbent Senator from Connecticut's "Issues" pages has yielded a very clear establishmentarian tendency.  Further analysis will be required, but provisionally it appears that Mr. Lieberman's overall preference is basically conservative... More later (hopefully)

A Peculiar Search Engine

My thanks go to Justin Gardner at Donklephant for calling attention to this latest development.  Apparently a search engine, Kosmix, sorts results of user searches on political issues according to their ideological slant, either liberal, conservative or libertarian.  Since ideological preference has been the central research question of the LEO Test, I must wonder how Kosmix sorts their results. I find the search engine making the distinction among these three ideological preferences particularly intriguing, since that distinction appears to coincide precisely with this project.

A Few Obervations about the LEO Test

First:  even with a two-month hiatus from adding content to this page, daily traffic has been higher than I expected. 

Second: Richardson's State of the State address is finished. His speech has two distinct peaks: one in his 'Year of the Child' section, and another in 'Moving New Mexico Forward'.  On the whole, though, Richardson appears to have given a generally communitarian address. Details to follow soon.

I also decided to test (Current Governor of Texas) Rick Perry's last State of the State address. I was surprised: Perry's speech yields an egalitarian with a strong libertarian leanings in three out of four measures. The fourth measure shows a libertarian with some egalitarian tendencies.

Third: I have two preliminary scores for President Bush's Fifth State of the Union address.  One is a handcount gathered as he delivered the address, while the other is an automated frequency measurement of specific indicators predefined in four distinct test tables.

They conflict.

This difficulty reveals a likely limitation in the dictionary file.  The handcount yielded a moderate conservative with strong libertarian leanings, which is what I expected.  However, the transcript of the address found a completely different score.  Three out of four measures gave me an egalitarian score, while the fourth gave me a left libertarian score.

The disadvantages of a dictionary file or automated test table are clear: dictionary files are blind to context as well as to grammatical transformations not included in the list. Without a systematic means of electing terms to the dictionary file, many items are therefore left out. 

Manual counts, however, have other disadvantages. Although context can be captured with a manual count, the process is far more cumbersome--one might even brand such demands a great deal more time and effort, and the observer's personal bias will taint the sample.

Here is a possible means to correct it: Constructing any test table for the LEO test should begin with a frequency distribution of substantive words and phrases from known ideologues.  Those words and phrases with the highest frequency become test indicators of that particular ideological preference. The terms I am currently using in the dictionary files are based on their proximity to particular ideas advanced by known ideologues.  I believe several terms are missing from the test tables I am currently using.

That will take quite some time.

The Task Ahead

This week promises to have numerous sources available for scrutiny.  I currently have plans to examine New Mexico Governor Bill Richardson's recent State of the State address.  A manual test suggested an interesting strategy--that Gov. Richardson's address is tailored such that he begins squarely in one ideological camp, coasts through another in the middle of the speech, and finishes  with a third preference.  I have yet to run it through the instrument, however, because I still need to convert the document from .pdf to .txt .

Based on the perceived ideological movements in Richardson's speech an examination of the State of the Union address tomorrow might need to include a higher resolution for its sampling frame, i.e. a record of LEO signatures per paragraph or major section.  I suspect that a closer look at segments of speech should reveal something about the preferences the speaker wants audiences to remember vs. the overall tenor of the speech. (Richard Cheney's RNC 2004 speech was particularly instructive about appearances and the 'effective truth'.)   However, since I have not expressly used this technique before, any comparison to past addresses by the President should only involve the total score.

After a Long Hiatus

For those who are interested, I found that the end of 2005 proved to be a time of significant delays and distrasctions. This week should mark a return to both the project and to contributions to Donklephant.  Suffice it to say, there is still much to be done, and I just learned two weeks ago that a service is available on my home campus, Mountain View College in Dallas TX, to assist with the grant process.  I believe I am near the point where gathering data for analysis may require formal assistance. 

My continued thanks still go out to Zathras of The Projection Blog, Jim of Lemurland, and Jeremy of Dispatches from the Moderate Left for their active interest in the project, and my sincere apologies go out to Justin Gardner at Donklephant for not having kept up.  I'm slowly waking up from an apparent and unintended hibernation, but I still need to catch up.  Much has happened, and much is about to happen.

Delays

II must apologize for the delays in content; almost a month has passed since new material has appeared here.  Much has been afoot which has required my attention elsewhere.  My contributions to the Donklephant weblog have suffered as well.  Suffice it to say that more should be forthcoming.  I have recently been asked to comment on Sen. Lieberman vis a vis perceived positions and ideological myopia, but it seems that Justin Gardner's post on Obama and Lugar also deserve some attention.  These should be forthcoming, either on Donklephant or the LEO Test.

The next LEO article Linked

For those who are interested, another LEO article has been submitted to Donklephant.  You may find it here.  The subject concerns the current Supreme Court nominee, Samuel Alito, and suggests that the Administration's current marketing strategy on the surface looks like a return to ideological appeasement, but underneath suggests a slightly different attempt to cross ideological lines to ensure support and prevent a filibuster attempt by potential opponents in the Senate.

On Senator Chuck Hagel

It has recently come to my attention, thanks to a brief post by Russ Mitchell at Boxing Alcibiades,  that the Senator from Nebraska, Chuck Hagel, has been labeled both a presidential hopeful and a 'centrist' by David Yepsen of the Des Moines Register.  Russ posits that identifying Senator Hagel thus is a sign that "American Politics is messed up", and also a sign that the Democratic Party in the US has moved far to the 'left', assuming of course that Yepsen is himself a Democrat, despite not having overtly announced his partisanship in the Register article. 

Granted, Carl at High Energy in Nebraska is doubtless more familiar with Senator Hagel than I am; yet Russ' dispute of Yepsen's claim proved for me too much too resist putting the Senator to the LEO test.

I initially sought to test Senator Hagel's Issues pages in a manner similar to my tests of Senators Santorum and Clinton  back in August.  However, Hagel's web presence proved to be a challenge, since the manner in which the Nebraska Senator chose to present his Issues pages was by linking to a Thomas server result for the 108th Congress. This choice suggests one of at least two possible interpretations:

1) that the Nebraska senator is interested in facilitating full disclosure of his activities in the US Senate, or at least his sponsorship record, or

2) that the Nebraska senator is interested in obscuring his ideological preference from those who are neither inclined to search through the text of bills nor familiar with the language of legislation.   

A quick perusal of his sponsorship record during the 108th Congress shows that Sen. Hagel's name tends to appear alongside both Joseph Biden of Vermont and Richard Lugar of Indiana with roughly equal frequency.  This same perusal also reveals that unless Hagel is the principal sponsor of the legislation, he also tends not to put his name to a bill until after some revisions have been made.

But what of his putative centrist position?  To test this claim, I have decided to take a sample of Sen. Hagel's 108th Congress sponsorship record and put it to the LEO test using the same method I  have used to test other political actors before.  Here are the results:

Raw Scores:

Adjusted Scores:

Aggregate, Mean, and Median Scores:

The sample itself is composed of the text of the first fifty bills returned by the Thomas server's search engine for Senate bills featuring Chuck Hagel as a sponsor or cosponsor at their introduction to the Senate.  Additional bills were included avter having removed those bills which only gained the Senator's support after it had been revised.

The first graph above shows the raw results of the four test tables.  It is curious that the first table yielded a generally establishmentarian (conservative) signature while the other three produce solidly --and similarly-- egalitarian (liberal) signatures.  It is likely that a key ideological indicator is not captured by the first table that is captured by the other three.The second graph shows percentage scores for each of the four tables.  One may note that in terms of the LEO model, from the three tables that yield a generally liberal score, the other two sets of indicators appear to be more or less evenly divided.  One may also note that for every table, libertarian references trailed.The third graph is a relatively new product within this project.  It shows the proportions of the aggregate, average, and median LEO scores from the four test tables.  Each one indicates a clearly liberal signature.

Given this signature, one may reasonably assert that ideological libertarians are not likely to view Chuck Hagel as a centrist.  Instead, they will see him as a liberal, and perhaps as a liberal ideologue.  Ideological conservatives are perhaps less likely to see Sen. Hagel as a liberal ideologue, but they will still be disinclined to call him a centrist. 

Why, then, does David Yepsen insist that Sen. Hagel is a centrist?  And why does Russ Mitchell suggest that Yepsen's claim reflects an inherent "leftism"?  There are two possible explanations. 

First, it is possible that both Yepsen and Mitchell are observing Chuck Hagel through the figurative lenses of their own distinct ideological preferences. 

In the first case, Yepsen's own ideological preference may be extended somewhat father along the egalitarian line, and because Hagel has demonstrated some conservative attenuation of his egalitarian preference Yepsen therefore sees the Senator more moderate than he sees himself.  Because David Yepsen has not been tested, the ideological myopia interpretation remains a hypothesis.

Russ Mitchell, on the other hand, has been tested using an ideological preference inventory based on the LEO model, and is shown to maintain a different ideological preference from the Nebraska Senator. See below:

Mitchell's LEO score (used with permission)

In this case, Russ Mitchell's ideological preference is extended farther along the libertarian line than Sen. Hagel.  Also, since Mitchell's egalitarian score trails the other two much like Sen. Hagel's libertarian score, it is likely that he would recognize liberal rhetoric more readily than one whose egalitarian score was higher.

Second, and more likely, Yepsen is observing the Nebraska Senator using a variation of the Political Spectrum model as a yardstick, while Mitchell is clearly not.  The sum of libertarian and establishmentarian scores for each test is nearly equivalent to Hagel's egalitarian score, and if one were to assume, albeit incorrectly, that libertarians are a species of conservative, only then could one reasonably assert that Chuck Hagel is a centrist.

Both interpretations above may explain why the Des Moines Register and Boxing Alcibiades differ in their assessment of Sen. Chuck Hagel.  However, the second may explain why the author of Boxing Alcibiades, an ideological libertarian, openly challenged Yepsen's claim.

A Note for Interested Readers

The next substantive article on Donklephant has posted.  Those who are interested may find a link below.

Presidential Damage Control

A Response Observed: The Miers Appointment

I must apologize for the delay.  I have a completed article currently sitting on hold over at Donklephant because of formatting issues with the tables.  Tables are available on this page, but apparently the images are pixellating badly on the other site. Much of this is due to multiple configurations which are not necessarily fully compatible with one another. 

I have heard it said that the more complex a system is, the greater the probablility it will fail.  While the LEO model itself is not particularly complicated, the medium of communicating findings under the LEO model is. Hence the delays and frustrations associated with delivering this content.

The second substantive article for Donklephant discusses the President's radio address on October 8th as a direct response to the conservative backlash against the appointment of Harriet Miers to the US Supreme Court. This conclusion emerged over a direct observation and comparison of LEO test scores for both the Appointment speech on the 3rd and the radio address itself.  Here are the results:

Test 1 Results

Test 2 Results

Test 3 Results

Test 4 Results

As stated in the Donkelphant article,

For every measurement using the LEO test, the egalitarian signature drops,  indicating that the President made an effort in the radio address on the 8th to compensate for the excessively egalitarian announcement on the 3rd, likely in response to the conservative backlash in the intervening days. 

Based upon this brief comparative analysis using the LEO model, the best way to describe Mr. Bush’s radio address on October 8th is “damage control.”  No matter what happened with the other ideological indicators under the LEO test, it is abundantly clear that the President’s statements on the 8th of October were meant at the very least to appear less liberal, perhaps to appease the conservative establishment in the Republican party, especially in the US Senate.

Yesterday Senator Schumer intimated that support for Miers' confirmation simply was not available on the Senate floor, and that the appointment is likely to fail.  While this statement may simply be a reflection of Schumer's partisan opposition, it may also be a sign that the balancing act the president has attempted to perform over the last two weeks is failing as well.

Another venue

For those who are interested, I have started a series of articles on Donklephant.  I'm trying to make sure the articles there are at least weekly, but this week derailed much of that effort.  Rest assured the project continues, and some of the data are included here as well.
Here's a link to the first substantive LEO article.

Some thoughts on Forthcoming Projects

Up to this point the focus of this project has been on measuring individual political actors who have published their opinions, either through campaign literature, bill sponsorship, speeches, or "issues pages" on their official webpages.

1. The Ideology and Partisanship Project.

To be sure this effort shall continue, especially since I have recently received a putative 'survey' similar to the RNC solicitation late last year, but this time from the DNC. I should like to see what preference is revealed, and whether its rhetorical volume is comparable to the RNC's That should be up soon as a second installment of the "Ideology and Partisanship" project, but with a clear view to the Democrats rather than the Republicans, who have been treated earlier.

2. A LEO test of pundits and electoral hopefuls.

This is actually two projects. The first project is a continuation of the test first conducted on the articles of Andrew Sullivan.  It appears to me that if Andrew Sullivan bears testing, so do others.  The second is a continuation of the test conducted on Senators Clinton and Santorum.  My original intent was to have a LEO test for every sitting Senator, but that project will simply have to wait until the method of building target files is less tedious.

3. A LEO test of American Founding documents or constitutions.

Inspired by the method of submitting the text of legislative proposals to the LEO test, I submitted the US Constitution and its amendments In a separate experiment.  This offered an interesting, though unsurprising analysis.  I must be sure to submit a write-up these findings. I should also look either to other major documents from the first generation of the United States in order to develop a clearer picture of the ideological character of the founding era, or else to the constitutions of other bodies politic to form a basis for comparison.

However, there is much more to the LEO test than this. 

4. The Inventory of Individual Ideological Preference.

One application of the LEO model which has borne the most fruit has been an individual inventory based on the LEO test table.  Until now it has only received passing reference in the discussion areas of this project, but the development of this tool has contributed substantially to refining the model itself.

The purpose for the inventory is clear:  since not all individuals have a published record from which to discern ideological preferences, another tool is needed.  The inventory fills that niche.  Respondents to the inventory may at first be unsure of their own preferences, or at least where they fit in the space described by the LEO model.  Respondents may also have taken other "political quizzes"(see the category stripe left at bottom) and have been dissatisfied by the results returned.  Others may wish to compare or contrast their own results with the results gathered from known political actors, especially presidential candidates.  This last option may even serve as a predictive model of voting behavior.   

The inventory begins with an expanded version of the LEO test table, and uses a method of observation similar to personality profiles. Unlike other political tests, the inventory itself makes no reference to specific policy measures, nor does it identify partisan positions.  Instead, this project uses simple word association involving predefined ideological indicators arranged in specific word groupings across three separate instruments.  Three preference scores, an opposition score, and an ambivalence score (if available) are captured, as well as the general orientation either towards ideology or moderation. 

These scores may then be compared with the results of standard LEO tests on the written records of political actors.  In the original experiment, the actors compared with individual respondents were six presidential candidates in the 2004 campaign season.

5. A list of theoretical findings.

Because of these projects, it has become apparent that a summary of general findings directly connected witht the LEO model will be required.   Certain terms which I have come to associate with the LEO model over the last few months must be defined in a denotative glossary, such as 'rhetorical volume', 'ideological preference', 'antiposition', and others--but not until they present themselves in the write-up.

6. The LEO test methodology.

In the interest of full disclosure and the need for replication of these experiments (for that is indeed what they are) the methodological portion of this project must eventually be disclosed. 

On a personal note, I am beginning to feel as if this project is becoming somewhat large for one researcher to handle, especially now that the academic year is in full swing.

A Brief Note

To those who have been anxiously awaiting a further exploration of subjects using this model and have been frustrated I apologize.  Now that the new academic year is in full swing, it has become somewhat difficult to advance this research project.  After all, I am but one, while the areas of application for the LEO test are staggering.  Furthermore, the foray into judicial decisions demonstrates the difficulties inherent in attempting to work within the context of individual cases.  Thus far the commentary on this subject has been fruitful, and has afforded all of us an opportunity to appreciate that complexity.

A future post should deliver a little more concerning the methodological nature of the LEO test, what future refinements of the test should include, and also a section devoted to the inventory of individual ideological preference based on the original LEO test. 

Keep in mind, of course, that the LEO test is a long-term project, and is by no means near completion. While some posts will come quickly, others will arrive with greater deliberation, for verily it has been noted that this page is less a weblog than an academic exercise in political science.

On Judicial Decisions, Part 2

The first part of this inquiry into what insights the LEO test may offer when applied to judicial decisions suggested that an ideological preference in the opinions proferred by members of the Federal Court might be discernable, at least when appled to such landmark cases as Brown v. Board of Education of Topeka KS and Miranda v. Arizona.  The scores that resulted from an analysis of the text of the opinions not only appeared to describe the preferences of the majority of the Supreme Court, but also the preferences of concurrent and dissenting opinions.  Taken together, these scores suggested a clear ideological difference between the majority opinion and the dissent in Miranda v. Arizona, especially, along predictable lines outlined by the LEO model. 
    Furthermore, if the method of observation as applied to judicial decisions is taken to be substantively valid, a moderate should be disturbed by the nomination of John Roberts to the Supreme Court to fill the seat vacated by Sandra Day O'Connor.  If this analysis be valid, moderates should be doubly concerned since today the president nominated Roberts again, this time to fill the late William Rehnquist's position as Chief Justice, despite the paucity of bench experience on Roberts' part.   The results of the application of a direct LEO analysis suggested in the Ramaprakash case (at least) that Roberts' ideological position was so far from 'center' or even moderation that were one to base his assessment of Judge Roberts on the Ramaprakash case, one should cast aside any hope for moderate discretion on Roberts' part, should he be confirmed.

One must remember, however, that in science all conclusions are tentative, and political science is of course no different.  At this time I have not gathered enough evidence to arrive at any general conclusion that Roberts is an extreme establishmentarian.  Nor have I conclusively ascertained that the current application of the LEO test to judicial decisions is substantively valid. 

In fact, the measurement of the next case, Roe v. Wade (1973), suggests either that a critical analytical component for judicial decisions may be missing, or else that the issues involved are somehow extraordinary.   As I have stated earlier, until this component is found I cannot be truly confident about any conclusions derived from the curent application of the LEO test.

Roe v. Wade (1973)

First of all, allow me to point out although I have read nowhere near as many court opinions as any practicing attorney, I found the  Roe decision to be perhaps one of the oddest sets of legal opinions I have yet encountered.  I may be mistaken about this conclusion based on my limited experience (and both Mark and Jeremy may feel free to correct me if I am), but I doubt that many other opinions go to the lengths as the Roe decision in order to explain  the opinion of the court.   

    To wit, the Roe decision includes references to classical sources such as the Persian Empire, Hippocrates, the Pythagorean School, Plato, and Galen.  It also cites English sources Bracton and Coke in the "common law" section, English statutory law, British parliamentary decisions from the early 20th Century, and finally American Law.  Indeed, the degree to which the Supreme Court relies on philosophical as well as classical and medieval sources to establish the opinion of the court in this case appears, at least to this humble layman, to be extraordinary.

    Indeed, Mark Zuniga's comments on an earlier post suggests that the language of Roe is in fact so unusual that legal professionals typically rely on opinions subsequent to it, rather than Roe itself.  "The first thing to look at," he told me,  "is Planned Parenthood of Southeastern Pa. v. Casey, 505 U.S. 833 (1992).  A future article may yet put this case to the test.

    That said, how do the opinions measure under the scrutiny of the LEO test as it is currently applied?

    Roe v. Wade contains three distinct opinions: the prevailing opinion, Justice Stewart's concurrent opinion, and Rehnquist's dissent.  The test to which these opinions were subjected is precisely the same as described earlier for the Brown decision, the Miranda decision, and others.  The hypothesis regarding this set of opinions is this: Since Roe v. Wade is (to paraphrase Jeremy Jose) a 'notoriously liberal' decision, or else (more likely given the peramaters of the LEO model) a libertarian position, with emphasis on the part of abortion supporters on the individual's 'right to choose' whether or not ot terminate a pregnancy, it should contain either a high instance of egalitarian (or 'liberal') indicators or a high number of libertarian indicators.  Rehnquist's dissent, on the other hand, should contain a high number of establishmentarian (or 'conservative') indicators, given his reputation as the standard-bearer for active conservatism in the Supreme Court.

For the record, this is not in fact what was observed, which perhaps makes the Roe decision even more perplexing.  See the graphs below (Click to enlarge):

Roe v. Wade Majority

      

Justice Stewart's Concurrence

Justice Rehnquist's Dissent

Except for the second test table which measures a higher percentage of libertarian indicators in the Roe majority sample, the test generally yields a higher percentage of establishmentarian indicators than any other.  Was the language of the majority opinion based upon conservative rhetoric?  Perhaps; perhaps not.  In his commentary on an earlier article, Jeremy Jose mused that    

"Legal opinions are probably a bad thing to analyse because judges consciously (in rhetoric if not always in practice) try to make it look as if they're doing nothing but applying precedent. You'll only rarely find broad assertations of philosophical principles, most of the time judges will be trying to obscure what they're doing by making it look like they're doing nothing."

This may be precisely what happened with the majority opinion in Roe v. Wade. 
Even so, all is not lost.  Interestingly, Justice Stewart's concurrent opinion is entirely consistent with the original hypothesis, for the test of this opinion returns a decidedly libertarian signature with every test table. 

The most peculiar result was that of the dissent by Justice Rehnquist.  Unfortunately Justice Rehnquist died two days before I could post this result, and so we can no longer hope to ask him directly about the language of his dissent, and shall have to settle for a close reading of the text he leaves behind.

    The first measure presented a strong 'right-libertarian' score, with over 55% of the total number of indicators appearing in the sample identified as references to 'liberty'. 
The second and fourth measures are much more peculiar, for they suggest that the dissent focuses on 'equality' rather than 'order' for its ideological direction.  Nevertheless, the fourth measure does produce a higher percentage of references to 'order' than any of the others,  and it is perhaps more fitting to compare it with the third measure, which shows a nearly centrist position, but more closely resembling a 'centrist communitarian', if such an ideological creature exists.

    A closer examination of the language of Rehnquist's dissent reveals something that the LEO test as it is currently applied is ill-equipped to handle.  While Rehnquist does in fact make several references to the principle of equality, and in particular to the "Equal Protection Clause" of the Fourteenth Amendment, he does so to explain his belief that it has been applied erroneously by the majority of the Court. This language is the likely reason for the high relative 'E' score. 

However, it must also be noted that the total number of ideological indicators in Rehnquist's dissent was quite low (the highest total was only 23 references) while the total rhetorical volume for the majority opinion appears sufficient to have drowned out Rehnquist's dissent.  See below (click to enlarge):

In the above graph, the y-axis indicates the raw number of references to liberty, equality or order as defined by the test tables indicated within the given samples.  Thus, "Majority 1" refers to the results of a search of the majority opinion for ideological indicators (keywords) defined by the first test table.  "Stewart 1" refers to similar results for Justice Stewart's concurrent opinion, and "Rehnquist 1" refers to a similar test of Justice Rehnquist's dissent.

When these raw numbers are compared, Rehnquist's dissent remains far below the other opinions in the number of ideological indicators defined by the four test tables used.  Such a low number of indicators in the sample suggests that Justice Rehnquist may have consciously attempted to avoid grounding his dissent on an ideological argument, while the majority sought to find an established argument supporting their position.

But does the above result mean that the language of the majority in Roe spoke from a fundamentally 'conservative' position?  Is such an argument even possible?

At this point it is instructive to consider the basic character of the abortion issue.  Although it has emerged in the last 32 years as perhaps the single most divisive domestic political issue in the United States, it does not conform to any single ideological preference.  Theoretical arguments both for and against abortion can be made from the standpoint of all three ideological axes.  For example:

A libertarian argument for abortion might focus on the freedom to choose what to do with one's body and one's cells, and with this freedom comes the freedom neither to conceive bear a child.  According to this argument, until the child leaves the mother's womb (and maybe even after, if the putative libertarian is an anarchist) the child's remains the property of the mother, and depending on the extent of the argument, the father as well. 

A libertarian argument against abortion might emphasize the responsibilty of each individual who chooses to have sex, and might be inclined to advocate a limited abortion right in cases of rape or endagerment to the life of either mother or child. (Interestingly the Texas law challenged by Roe appears to have made this exception.)

An egalitarian argument for abortion might focus on the perceived inequality in the role of mothers and fathers in gestation, that pregnancy relegates the pregnant to the status of second-class citizens, and that women are effectively slaves both to fathers and children.  Under this argument, the unborn is an unjust burden, and active steps must be taken to ensure either that fathers bear the same burden, or that women bear no burden.   

An egalitarian argument against abortion would emphasize the equal right of the child to live, that the unborn are human beings, and therefore posess the same fundamental equality with their parents.  According to this argument, elective abortion is a kind of murder, for it unjustly deprives a fellow human being the equal enjoyment in the right to life.

An establishmentarian argument for abortion might emphasize the need to regulate public health so that women do not endanger themselves by seeking illegal abortions in potentially septic environments. Or, more disturbingly it may focus on the need to regulate population growth, to weed out unhealthy or undesirable traits or populations, or to minimize the burden of state support for indigent populations--in short to manage a part of an industrialized population.

An establishmentarian argument against abortion would focus on the practice as a threat to the established order, a transgression against either a moral order or a religious edict, or a violation of traditional practices and codes of ethical conduct.  This argument would especially arise if innovative techniques of abortion were introduced. 

For these reasons, any regime --liberal, conservative, or libertarian; communist, fascist (or anarchist?)-- must at some point confront the issue itself. 

Tempus Fugit

For those who have been following this series of articles, I thank you for your patience.  The last two weeks have been filled with an extraordinary amount of other obligations and concerns, including the adaptation of course syllabi to a new basic text for the courses I am teaching this semester, preliminary plans for a future semester involving fieldwork, my first visit to the District of Columbia for the Annual Conference of the American Political Science Association,  and concerns about friends on the Gulf Coast. 

Although today is now Labor Day, rest assured the second part of the Judicial Decisions Article is forthcoming, and hopefully shall be available before the end of the day.