2. Modern-Day Proxies for Ideology
Scholars studying the effects of ideology on Supreme Court decisionmaking
initially used approaches similar to those used by Professor Pritchett. Scholars found
patterns of voting behavior, and interpreted them to mean that ideology affected
decisionmaking.1
Yet the very thing those scholars wanted to explain—votes on the
merits—were used to create the measure of ideology that was used to “explain” the votes.
Do you see the problem with this? It’s circular!
The independent variable in studies like these is ideology, which is measured
using the votes themselves. The dependent variable is the vote as well. It’s not surprising
at all that these things are related to one another, since they’re precisely the same thing.
Indeed, a finding such as this is just as “profound” as finding a perfect relationship
between temperatures measured in degrees Fahrenheit and degrees Celsius; it just doesn’t
make any sense to call one the dependent variable, one the independent variable, and then
get excited when they are strongly related. Obviously they are. Unfortunately, much of
the early work on the attitudinal model suffered from this tautology, using votes to
measure ideology (or attitudes), which were then used to “explain” the votes themselves.
Doing so provides no explanation at all.
a. Appointing President as Proxy
Rather than drawing assumptions about judges’ ideologies based on their votes,
more recent studies often use the party of the appointing President as a proxy for judicial
ideology. For example, as we’ve seen, that is how Professor Revesz measured judicial
ideology in his study of the D.C. Circuit.
While reliance on the party of the appointing President avoids the circularity
problem that haunts studies like Pritchett’s, it has its own drawbacks. Indeed, one of
Judge Edwards’ specific complaints about Professor Revesz’s study relates to the
decision to code judges’ “ideology” based on the party of the appointing president. Judge
Edwards found this to be extremely problematic. (We’ve broken up Judge Edwards’
argument for ease in considering it.)
1. “As Revesz’s title suggests, ‘ideology’ is a major component of his
study. Here choice of language matters tremendously. The loaded word
‘ideology’ makes headlines and attracts attention; it is therefore very
striking that Revesz never bothers to proffer a definition of this allimportant
term. Rather, he explains that he will use as a proxy for
ideology ‘the views generally held by the party of the appointing
President.’ This approach raises a number of serious problems.”2
1
See David W. Rohde and Harold J. Spaeth, SUPREME COURT DECISION MAKING (1976); Glendon A.
Schubert, THE JUDICIAL MIND: THE ATTITUDES AND IDEOLOGIES OF SUPREME COURT JUSTICES,
1946-1963 (1954).
2
Id. at 1347.
2
2. “The threshold problem is that Revesz does not provide an
explanation of the usefulness of the proxy of the party of the
appointing President as a basis for judicial ideology…. Rather
obviously, the categories ‘Democrat’ and ‘Republican’ do not
necessarily match sympathy towards greater or lesser degrees of
environmental protection. The policies of individual Presidents—to
say nothing of judges—may diverge greatly on such a complex topic.
Without providing some statistical measure of the significance of the
asserted correlation between political party and environmental policy,
it is suspect for Revesz to assume that his ideological proxy is
meaningful.”3
3. “But more importantly, Revesz never explains why the supposed
high political profile of the D.C. Circuit, even if it were established
fact, would ensure a close connection between the ideology of an
appointing President and the ideology of an individual judge….
[N]umerous Supreme Court Justices, past and present, have famously
disappointed the parties of the Presidents who appointed them by their
independence and unpredictability….”4
Judge Edwards’ complaint goes to the heart of using what is called a proxy. It is a
simple fact of empirical study that often you cannot code the variables (dependent and
independent) for precisely what you want to study. Instead, you oftentimes have to rely
on one variable that is easy to measure that stands in for, or represents, another. The
variable that stands in is called a proxy variable. Here, Revesz used political party of the
appointing President is a proxy for ideology. What is the argument for its use as a proxy?
In what ways is it limited?
In his own study of decisionmaking in the circuit courts, which we will examine
below, Professor Cass Sunstein provided this rationale for using the political party of the
appointing President to measure judges’ ideology:
Presidents are frequently interested in ensuring that judicial appointees
are of a certain stripe. A Democratic president is unlikely to want to
appoint judges who will seek to overrule Roe v. Wade and strike down
affirmative action programs. A Republican president is unlikely to
want to appoint judges who will interpret the Constitution to require
states to recognize same-sex marriages. It is reasonable to hypothesize
that as a statistical regularity, judges appointed by Republican
presidents…will be more conservative than judges appointed by
Democratic presidents.5
3
Id.
4
Id. at 1348-1349.
5
Sunstein, Schkade & Ellman, supra, at 303.
3
Look back at the tables in the previous section. Does the party of the appointing
President help explain the voting blocs of Justices we see? On the more modern Court,
five Republican appointees vote together most of the time. But, of the remaining four
Justices (Breyer, Souter, Stevens, and Ginsburg), two are Democratic appointees and two
are Republican appointees. Using party to as a proxy for ideology simply does not work
very well for some Justices, like Justices Blackmun, Stevens, and Warren—all
Republican appointees who were very prominent liberals during their time on Court.
President Eisenhower once quipped that appointing Justice Warren was “the biggest
damned-fool mistake I ever made.”6
What do you think causes such “mistakes”? Are
there motivations other than ideology that might account for Presidents’ choices for the
Supreme Court?7
b. Segal-Cover Scores
To break out of some of the perceived the problems associated with using
Justices’ votes or the party of the appointing President to measure ideology, Professor
Jeffery A. Segal and his collaborator Professor Albert D. Cover invented a novel
approach that looks at newspaper editorials about Supreme Court nominees.8
Editorial
boards oftentimes write opinions about nominees to the Court and their suitability for the
job. Not surprisingly, more liberal boards like the New York Times tend to favor more
liberal nominees; more conservative boards like the Wall Street Journal favor more
conservative nominees. Segal and Cover collected all editorials written about all
nominees to the Court—including those who were never confirmed—by four leading
national newspapers. They did this from the nomination of Chief Justice Earl Warren
forward.
Here is how Professors Segal and Cover describe the construction of their
measure:
We started by conducting a content analysis of a source that contains
comparable information on each justice since Earl Warren: statements
in newspaper editorials from the nomination by the president until the
confirmation vote by the Senate. Newspaper editorials are one of the
few independent sources of information that contain comparable data
on each justice. Only editorials prior to confirmation by the Senate
were examined, for editorials after confirmation will undoubtedly be
influenced by votes a justice casts and thus are not independent of
those votes. We selected four of the nation's leading papers, two with a
liberal stance (the New York Times and Washington Post) and two with
a more conservative outlook (the Chicago Tribune and Los Angeles
Times).
6
CONG. QUARTERLY, THE SUPREME COURT: JUSTICE AND THE LAW 175 (3rd ed. 1983).
7
See Edwards/Livermore at 1919 (citing personal relationships and party building as alternative
motivations);
8
Jeffrey A. Segal & Albert D. Cover, Ideological Values and the Votes of U.S. Supreme Court Justices,
83 AM. POL. SCI. REV. 557 (1989).
4
To conduct the content analysis, we trained three students to code each
paragraph for political ideology. Paragraphs were coded as liberal,
moderate, conservative, or not applicable. Liberal statements include
(but are not limited to) those ascribing support for the rights of
defendants in criminal cases, women and racial minorities in equality
cases, and the individual against the government in privacy and First
Amendment cases. Conservative statements are those with an opposite
direction. Moderate statements include those that explicitly ascribe
moderation to the nominees or those that ascribe both liberal and
conservative values.9
Segal and Cover then constructed a measure of liberalism by taking the difference
in the number of liberal paragraphs and conservative paragraphs, dividing by the total
number of applicable paragraphs, and rescaling to an interval where 0 is unanimously
conservative, and 1 is unanimously liberal. This avoids the circularity problem because
the editorials are written before the Justice decides any cases. Figure XX shows the
resulting Segal-Cover scores from Chief Justice Warren to Justice Sotomayor.
9
Id. at 559.
5
Figure XX. Segal-Cover measure of judicial liberalism for justices confirmed 1953-
2009. Jeffrey A. Segal and Albert D. Cover, Ideological Values and the Votes of U.S.
Supreme Court Justices, 83 AM. POL. SCI. REV. 557 (1989). Data obtained from
http://www.stonybrook.edu/polsci/jsegal/qualtable.pdf (accessed July 14, 2010).
Republican nominees marked in red; Democratic nominees marked in blue. [We should
change one of the dots to an x or something similar that’s visible in black and white.]
Eyeball the chart. What do you think? How did they do? You can see that the
measure does fairly well differentiating within political party, especially among
Republicans. The Republicans nominated by President Eisenhower are notably more
liberal than those appointed by Presidents Nixon, Reagan, and Bush I and II.
Taken at face value, the scores make a lot of sense. Justice Scalia is measured as
the most conservative justice nominated to the court; Justices Fortas, Brennan, and
Marshall are the most liberal. There are some misses, too, especially Justice Blackmun.
And Justices Stevens and Souter are coded as more conservative than Justices O’Connor
and Kennedy, although—as the tables in the previous section showed—Stevens and
Souter tended to vote with Democratic appointees Breyer and Ginsburg?
What do you see as the utility of this measure? Do you see any problems?
6
c. Giles-Hettinger-Peppers Scores
As clever as the Segal-Cover scores are, they are going to present a challenge for
any study of the lower courts, because there may not be much in the way of editorials for
most lower court judges.10
But as we’ve seen, there are shortcomings with using the party
of the appointing President. Not all Republican appointees are the same; neither are all
Democratic appointees. The question is what else could be used to distinguish among
different nominees selected by the same President.
Professors Michael Giles, Virginia Hettinger, and Todd Peppers have developed a
measure that seeks to differentiate among appointees by exploiting information about
how the appointment and confirmation process actually works.11
Article II, Section 2 of
the Constitution stipulates that the president nominates judges with the “Advice and
Consent of the Senate.” By custom, the White House oftentimes confers with home-state
senators—particularly those from the President’s own party—when there are vacancies in
the Courts of Appeals and the District Courts. The Senate Judiciary Committee also
provides “blue slips” to both home-state Senators to solicit their input into the nomination
process. The home-state Senators either approve or disapprove of prospective nominees,
or sometimes just don’t return these pieces of paper to the committee. This is the formal
way that the norm of “senatorial courtesy” works in the Senate. Some would go as far as
to claim that home-state Senators of the president’s party essentially pick nominees for
the lower federal courts.12
Others argue that a nomination will not go forward without
support from the home-state Senator.13
Because of the tradition of senatorial courtesy,
there are good reasons to think that Republican from Texas would look very different
than a Republican from Maine, or that a California Democrat is different from a South
Carolina Democrat.
Professor Giles and his colleagues decided they needed a measure of ideology that
took account of not only the appointing President, but the home-state Senators as well.
10
Corey Rayburn Yung argues that “a similar approach at the lower federal courts is simply
impossible because of the limited, and event-driven coverage of other federal judges.” Corey
Rayburn Yung, Judged by the Company you Keep: An Empirical Study of the Ideologies of Judges on
the United States Courts of Appeals, 51 B.C. L. REV. 1133, 1149 (2010). At the state level, Professors
Paul Brace, Laura Langer, and Melinda Gann Hall note that a Segal-Cover type approach at the
state level would be exceedingly difficult to implement. Paul Brace, et al., Measuring the Preferences
of State Supreme Court Judges, 62 J. POL. 387 (2000). However, a Segal-Cover type approach has
been used for at least one state. In a study of the California Supreme Court, Emmert and Traut
capitalize on the comparatively high salience surrounding the California Supreme Court to use a
Segal-Cover approach to measure ideology. Craig F. Emmert & Carol Ann Traut, The California
Supreme Court and the Death Penalty, 22 AM. POL. Q. 41 (1994).
11
Michael W. Giles, et al., Picking Federal Judges: A Note on Policy and Partisan Selection Agendas,
54 POL. RES. Q. 623 (2001).
12
Asked about his role in the judicial nomination process during the Reagan administration, Texas
senator Phil Gramm stated, “I’m given the power to make the appointment. . . . The people elected
me to do that.” Cragg Hines, Dispensing Legal Plums, HOUSTON CHRON., Apr. 21, 1985, at 1.
13
Harold Chase, FEDERAL JUDGES: THE APPOINTING PROCESS 7 (1972).
7
To construct their measure, they needed a good measure of ideology in Congress. While
there are a handful of viable options, they chose the state-of-the-art and widely used
ideological scores of members of Congress and presidents developed from by roll call
record by Professors Keith Poole and Howard Rosenthal.14
These scores differentiate
presidents and Senators within the same political party.
Professor Giles and his colleagues assume that presidents will appoint members
close to themselves in this ideological space, and that same-party Senators will as well.
When there is no same-party Senator, the nominee gets the president’s score. When there
is a same-party Senator (or two), the nominee gets the Senator’s score (or the average of
the two Senators scores). This measure captures some of the heterogeneity within the
Democratic and Republican parties. Their method can be applied to the study of district
court judges as well, or in the few states with similar selection systems.15
Is the choice of whether to use the party of the appointing President or the GHP
measure consequential? In practice, no. Indeed, these measures are so strongly correlated
that we reach almost exactly the same inference regardless of the measure used. The
results of nearly every study that uses party of the appointing president would be exactly
the same if the GHP measures were used instead.16
One reason this is the case is that
today our political parties are very polarized. While there are differences within the
parties, the differences across parties are far more stark. The GHP measure differentiates
within party, but those differences are dwarfed by the differences between Democrats and
Republicans, which is what is picked up by the political-party-of-President proxy.
d. Scaling and Martin-Quinn Scores
Justices can serve on the Supreme Court for a very long time; Justice Stevens, for
example, served almost 35 years before stepping down in 2010.17
During long careers we
might expect some Justices to change their minds about important policy questions. As
we have noted, Harry Blackmun, who was nominated by President Nixon, was one of the
most liberal justices on the Court by the end of his career. But, by all accounts, Justice
Blackmun started out rather conservative.18
14
Keith T. Poole & Howard Rosenthal, IDEOLOGY AND CONGRESS (2008).
15
Christina L. Boyd, The Impact of Courts of Appeals on Substantive and Procedural Success in the
Federal District Courts (2010) (working paper).
16
For example, Professors Gregory C. Sisk and Michael Heise find that the choice of measure doesn’t
matter much in their study of religious freedom cases in the circuit courts. Gregory C. Sisk &
Michael Heise, Judges and Ideology: Public and Academic Debates About Statistical Measures, 99 NW.
U. L. REV. 743 (2005). And so, too, for a myriad of other studies we discuss below. The substantive
conclusions are strikingly similar regardless of the strategy employed.
17
Sheryl Gay Stolberg and Charlie Savage, Stevens’ Retirement is Political Test for Obama, N.Y.
TIMES, Apr. 9, 2010, at A1.
18
Linda Greenhouse, BECOMING JUSTICE BLACKMUN: HARRY BLACKMUN’S SUPREME COURT
JOURNEY (2005); Lee Epstein, et al, Ideological Drift Among Supreme Court Justices: Who, When, and
How Important? 101 NW. U. L. REV. 1483 (2007).
8
None of the measures we have discussed thus far can account for this sort of
change, because they all measure ideology at the time of confirmation. Look at Figure
XX; does Justice Blackmun’s score seem accurate throughout his career? Another
problem, as mentioned before, is that some of the measures don’t work very well outside
of the domain of civil rights and civil liberties cases.19
An alternative measurement strategy was developed by Professors Andrew D.
Martin and Kevin M. Quinn.20
Their goal was three-fold: to develop a measure of
ideology that was applicable across a number of issue areas, that allowed for the
comparison over time even of justices who had not served together on the Supreme
Court, and that would allow the locations of justices to shift during their tenure on the
Court.
The Martin-Quinn model is a “spatial” model, meaning that it identifies the
Justices’ legal positions as what are called “ideal points” along a left-right ideological
spectrum. In order to understand the concept, think about a case in which the Justices
could take legal positions that ranged across a scale from the left to the right. One good
example is the legal standard to apply in a case of claimed sex discrimination. In
Frontiero v. Richardson, 411 U.S. 677 (1974), the Supreme Court was asked to decide
whether the medical and dental benefits provided by the military to families of service
members should differ based on the gender of the service member. Under existing law,
the wives of service members were simply presumed to be dependent and therefore
entitled to benefits. But if the husband of a female service member wanted benefits, the
couple had to prove that he actually depended upon her for over one-half of his support.
The question was whether that discrimination was constitutional under the Equal
Protection Clause. Not only did the Justices differ on the outcome, but they also differed
on the legal standard that should apply. Some of the Justices (Brennan, Douglas,
Marshall, and White) believed that the law should be evaluated under a “strict scrutiny”
standard under which very few laws discriminating on the basis of sex would survive.
Consider this a position toward the left of the scale. Other Justices (Powell, Burger, and
Blackmun) believed that discrimination on the basis of sex should be reviewed under a
“rational basis” standard, which was much more forgiving of governmental
discrimination. This is a position to the right end of the scale. But among those Justices
who would apply rational basis scrutiny, all but one voted to strike down the law in
Frontiero. The sole exception was Justice Rehnquist, who applied rational basis scrutiny
to uphold the law, placing his legal position even further to the right. And although in
Frontiero itself there was no majority for any clear position, a few years later, in Craig v.
Boren, a majority of the Justices agreed on an “intermediate scrutiny” position that was
somewhere in the middle. Figure XX depicts these positions. Do you see how each of
these options can be thought of as points on a “left-right” scale?
19
Epstein & Mershon, supra.
20
Andrew D. Martin & Kevin M. Quinn, Dynamic Ideal Point Estimation via Markov Chain Monte
Carlo for the U.S. Supreme Court, 1953-1999, 10 POL. ANALYSIS 134 (2002).
9
Figure XX. The policy dimension in Frontiero v. Richardson, 411 U.S. 677 (1974).
So how do you figure out what these ideal points are for a particular justice over
the course of a number of cases, and how the justices’ ideal points relate to one another?
The Martin-Quinn model uses a lot of computing power to do this, “estimating” two
things at one time. First, it essentially does what Professor Pritchett did, but for all cases
decided since 1937. The model looks at all the votes in all the cases and figures out the
agreement pattern among the Justices. This alone provides a great deal of information
about where the Justices stand in relation to one another. But then the model does a
second thing. Given the information about where the justices are located, the computer is
able to find out in each case where the “cutpoint” is – i.e. the point on the court where the
majority and the dissent divided. So, if a decision is 5-4 in a conservative direction, then
the cutpoint is between the median justice and the Justice immediately to her left. If the
decision is 6-3 in a liberal direction, the cutpoint falls between the six most liberal
justices and the three in dissent. Knowing the agreement information, and the cutpoints,
the computer then does the best job it can of predicting which justices fell onto which
side of which cases. The model has proven reasonably accurate, correctly predicting
some 75% of the Justices’ votes
This may seem a bit of computational magic – and perhaps it is – but take a look
at Figure XX, which shows the positions of the Supreme Court Justices from 1994 to
2008. Because the scores are dynamic, each justice is represented by a line. (Compare
that to the points of the Segal-Cover scores.) Justice Thomas is on the far right; Justice
Stevens is on the far left. Interestingly, since 2001, Justices Thomas and Scalia have
seemed to diverge. Up through the 2004, Justice O’Connor is identified as the median
justice. This makes sense because of the many 5-4 majority coalitions in which she
participated, joining four other colleagues, sometimes those on her left, and sometimes
those on her right. Since the 2005 term, Justice Kennedy is the swing Justice on the
Court. Justices Alito and Roberts are squarely on the right, but not as conservative as
Justices Scalia and Thomas. If you knew much about the Supreme Court during this
period, you’d think that the Martin-Quinn model is doing a pretty good job. In fact, the
Martin-Quinn scores have proven to be quite predictive of the justices’ ideology relative
to one another, and thus extremely useful in empirical studies of the Supreme Court.
Indeed, Martin-Quinn scores are considered state-of-the-art and are widely used.21
21
For more technical details, see Andrew D. Martin and Kevin M. Quinn, Dynamic Ideal Point
Estimation via Markov Chain Monte Carlo for the U.S. Supreme Court, 1953-1999, 10 POL. ANALYSIS
Strict
Scrutiny
Rational
Basis
Brennan, Douglas,
Marshall, and White
Powell, Burger,
and Blackmun
Rehnquist
Intermediate
Scrutiny
10
Figure XX. Martin-Quinn measure of judicial conservatism, 1994-2008 terms. Andrew
D. Martin and Kevin M. Quinn, Dynamic Ideal Point Estimation via Markov Chain
Monte Carlo for the U.S. Supreme Court, 1953-1999, 10 POL. ANALYSIS 134 (2002).
Data obtained from http://mqscores.wustl.edu/measures.php (accessed August XX, 2010
134 (2002); Andrew D. Martin, Kevin M. Quinn, and Lee Epstein, The Median Justice on the U.S.
Supreme Court, 83 N.C. L. REV. 1275 (2005).
Term
Martin_QuinnConservatismMeasure
1994 1996 1998 2000 2002 2004 2006 2008
Rehnquist
Stevens
OConnor
Scalia
Kennedy
Souter
Thomas
Ginsburg
Breyer
Roberts
Alito
_3_2_101234
11
FOR DISCUSSION
1. Think about each of the methods described above. Can you see advantages?
Shortcomings? Think in particular about how useful each would be for various courts.
Would one method be useful for Supreme Court justices but less so for lower courts?
Vice versa?
2. With regard to political party, can you think of any value to this approach other
than as a proxy? Does appointing party of the President offer any insight as a dependent
variable standing alone?
3. How useful are Segal-Cover scores? Although Professor Segal and his colleagues
initially constructed their “scores” based on editorials in four leading publications, they
later expanded the set of newspapers to include the St. Louis Post-Dispatch and the Wall
Street Journal.22
Why do you suppose they did that? Also, Segal-Cover methodology
treats all paragraphs written by all editorial boards over a fifty year time period equally.
What are the potential downsides from doing this? Can you think of a better approach?
4. Unlike in the natural sciences when measuring things like length, weight, or
temperature is relatively easy, measuring latent (unobservable) concepts like ideology in
the social sciences is far more difficult. What do we have to assume to use political party
of the appointing President as a measure of ideology? What do we have to assume to use
the GHP measures? Which set of beliefs is more reasonable?
5. One interesting finding of the Martin-Quinn model is that sometimes a single
ideological dimension does just fine to explain behavior, even across different areas like
economics or federalism cases. But sometimes a single dimension doesn’t work very
well. The below graph orders the Justices based on their behavior during the entire 1993
Term – the Term when Smith v. United States was decided. You may remember that this
is one of the cases where the Court considered whether trading a gun for drugs was a
“use” under federal sentencing law. Justice O’Connor wrote the majority opinion in that
case. To show this, we have highlighted in red and placed above the line the dissenting
Justices in Smith, with the majority Justices appearing in black below the line:
22
Jeffrey A. Segal, et al., Ideological Values and the Votes of U.S. Supreme Court Justices Revisited, 52
J. POL. 812 (1995).
12
What do you see? Does the ideology of the justices explain their behavior in this case?
What do you suppose is doing the work? Is “law” what explains the outcome of this case?
D. Judges’ Decisions as the Dependent Variable
Following the lead of the attitudinal model, the working hypothesis of the studies
we have seen is that judicial ideology affects judicial decisionmaking. We have discussed
measuring the primary independent variable—ideology—in a variety of ways. next step
is to address the dependent variable—the thing we wish to explain. (As we will see in a
moment, we need to control for other independent variables that might also explain the
dependent variable.)
When coding any variable, a coding protocol can be assessed in two ways:
whether it is valid and whether it is reliable. A valid measure is one that is actually
measuring the thing that we hope to measure. For example, in the previous section we
discussed concerns that coding for the political party of the appointing President may not
be an accurate—or valid—measure of judicial ideology. A reliable measure is one that
consistently produces the same measure each time it is applied. A bathroom scale would
be reliable so long as it returned the same reading every time you stepped on it; it could
do so reliably whether or not it gave an accurate weight measure.
When coming up with a dependent variable (at least on a “collegial” court, i.e.,
one that has several judges hearing each case) the easiest thing to do is to sort judges or
justices into those who support the majority judgment and those who do not. That is what
Professor Pritchett did in his studies of the Roosevelt Court. One simply has to look at the
majority opinion, and see whether every other justice concurs in the disposition or not.
This approach is reliable and easily replicable. But even here there can be complications:
for example, when there are only partial concurrences for the disposition (e.g. when one
of the Justices notes neither to affirm or reverse the judgment below flat out, but would
remand to the lower court for decision on further issues.)
Despite its advantages in ease of application, Pritchett’s approach has serious
limitations. First, as noted, it can be employed only in a collegial court, not in a trial court
where only one judge is responsible for a decision. Second, while sorting judges into
majority and dissent shows agreement about the disposition of each case, it tells us
Stevens
Blackm
un
Souter
Ginsburg
Kennedy
O
'C
onnor
R
ehnquist
Scalia
Thom
as
_3 _2 _1 0 1 2 3
Martin_Quinn Scores for the 1993 T erm
13
nothing about the substance of the decisions. What did the court actually do? Who won
and who lost? Did the majority rule for the criminal defendant, or the state? Did the
corporation win, or the individual alleging employment discrimination? If we are
interested in learning about the effects of judicial ideology, we may want to know more
about judges’ decisions than studies like Pritchett’s can tell us. To that end, scholars have
developed other coding schemes, such as coding outcomes as liberal or conservative.
1. Revesz and the D.C. Circuit
Recall Professor Revesz’s study of decisionmaking on the D.C. Circuit. Revesz’s
intuition was that the judges’ ideologies—measured by the party of the appointing
President—affected their decisions. His dataset consisted of environmental cases
involving the EPA.
In order to examine his intuition that the judges’ decisions were influenced by
their ideologies, Revesz had to measure how the judges actually ruled in each case, in a
way that captured something that might have been affected by ideology. What would you
do? In most cases in the circuit courts of appeal, the court is approving (affirming) or
disapproving (reversing) the decision of the lower court. In most EPA cases, however, the
party challenging the EPA’s action appeals directly to the D.C. Circuit from the agency
decision, without going through the district court first. The D.C. Circuit then decides
whether to affirm or reverse what the EPA has done. One option for Revesz, then, was to
code the D.C. Circuit’s decisions for whether they affirmed or reversed the EPA. Such an
approach would be fairly straightforward. Can you see the problem though? It wouldn’t
tell us much about the ideological valence of the court’s decisions. Sometimes the agency
will do things that will favor environmentalists, and sometimes it will do things that will
favor industry groups. A coding method that focused on whether the D.C. Circuit
affirmed the agency’s decision couldn’t distinguish between decisions that were proenvironment
and decisions that were pro-industry.
What Professor Revesz did was focus on who was challenging the EPA’s action.
Rather than just coding affirmances or reversals, Revesz classified each challenger as
either an industry challenger (who sought to limit regulation) or an environmental
challenger (who sought to increase environmental protection). For each case, he then
coded whether the judge voted to reverse the EPA or not—in other words, whether the
judge agreed with the challenger’s position. For industry challenges, reversing the EPA
would a pro-business disposition; for environmental challenges, reversing the EPA would
be a pro-environment decision. We might plausibly call the first type of decision
“conservative” and the second type “liberal.”
As noted, Revesz’s study confirmed his intuitions about the effects of judicial
ideology on decisions in environmental cases. He found that, beginning in the 1980s,
Republican appointees were more likely to reverse EPA decisions when challenged by
industry than their Democratic-appointee colleagues. When environmental groups
challenged EPA decisions, judges appointed by Democratic presidents were more likely
14
to reverse than those appointed by Republicans. Revesz concluded that “ideology
significantly influences judicial decisionmaking on the D.C. Circuit.”23
2. Supreme Court and Court of Appeals Databases
Professor Revesz’s study of the D.C. Circuit focused on one issue-area:
environmental law. Revesz reasoned that decisions favoring environmental groups would
appeal to political liberals, and decisions favoring industry would appeal to
conservatives. Using a similar approach, other scholars have coded decisions across a
range of different issue-areas as “liberal” or “conservative” based on who won and who
lost. The most salient example of this is the Supreme Court Database
(http://supremecourtdatabase.org), first developed and released in the 1980s by the
attitudinalist Professor Harold Spaeth. The Supreme Court Database is a treasure trove of
information about every case decided by the Supreme Court since 1946. It includes
information about who the litigants were, how the case got to the Supreme Court, the
issues and legal provisions considered in the case, who won, who lost, and the manner in
which each justice voted. (A group of scholars currently is backdating the Supreme Court
Database to the beginning of the Supreme Court’s first decision.)
In each case in the Supreme Court Database, Professor Spaeth and his research
team coded every Justice’s vote as either liberal or conservative. Each case has an issue
area, such as criminal procedure, or labor-management relations. And each issue area has
its own rule for labeling outcomes liberal or conservative, as shown below.24
1. In the context of issues pertaining to criminal procedure, civil rights, First Amendment,
due process, privacy, and attorneys, [votes are coded as liberal if they are (and
conservative if they are not)]
• pro-person accused or convicted of crime, or denied a jury trial
• Pro-civil liberties or civil rights claimant . . .
• Pro-child or juvenile
• Pro-indigent
• Pro-Indian
• Pro-affirmative action
• Pro-neutrality in establishment clause cases
23
Revesz, Environmental Regulation, Ideology, and the D.C. Circuit, supra, at 1719.
24
Online Code Book, SUPREME COURT DATABASE, http://supremecourtdatabase.org/
documentation.php?var=decisionDirection (last visited Apr. 25, 2013).
15
• Pro-female in abortion
• Pro-underdog
• Anti-slavery
• Anti-government in the context of due process, except for takings
clause cases where a pro-government, anti-owner vote is considered
liberal . . .
• Violation of due process by exercising jurisdiction over nonresidents
• Pro-attorney
• Pro-accountability and/or anti-corruption in campaign spending
• Pro-privacy vis-à-vis the 1st Amendment where the privacy invaded is
that of mental incompetents
• Pro-disclosure in Freedom of Information Act issues except for
employment and student records
2. In the context of issues pertaining to unions and economic activity, [votes are coded as
liberal if they are (and conservative if they are not)]
• pro-union except in union antitrust where liberal = pro-competition
• anti-business
• anti-employer
• pro-competition
• pro-liability
• pro-injured person
• pro-indigent
• pro-small business vis-a-vis large business
• pro-state/anti-business in state tax cases
• pro-debtor
• pro-bankrupt
• pro-Indian
16
• pro-environmental protection
• pro-economic underdog
• pro-consumer
• pro-accountability in governmental corruption
• pro-original grantee, purchaser, or occupant in state and territorial land claims
• anti-union member or employee vis-a-vis union
• anti-union in union antitrust
• anti-union in union or closed shop
• pro-trial in arbitration
• pro-state in state and territorial land claims
3. In the context of issues pertaining to judicial power, [votes are coded as liberal if they
are (and conservative if they are not)]
• pro-exercise of judicial power
• pro-judicial “activism”
• pro-judicial review of administrative action
4. In the context of issues relating to federalism, [votes are coded as liberal if they are
(and conservative if they are not)]
• pro-federal power
• pro-executive power in executive/congressional disputes
• anti-state
5. In the context of issues pertaining to federal taxation, [votes are coded as liberal if they
are (and conservative if they are not)] pro-United States
6. In interstate relations, miscellaneous, and private law issues, [votes are coded as
unspecifiable] for all such cases.
A similar database, developed by Professor Donald Songer, collects information
about decisions in the federal courts of appeals
17
(http://artsandsciences.sc.edu/poli/juri/appct.htm). Like the Spaeth database, the Songer
database codes decisions by issue area, and classifies each judge’s decision as “liberal” or
“conservative.” Unlike the Supreme Court database, however, the Court of Appeals
database classifies some decisions as “mixed”—meaning that “the judge voted for an
intermediate outcome, for example to affirm a criminal conviction but reduce the
sentence; in other words, he cast a liberal vote on one issue and a conservative vote on
another in the same case.”25
FOR DISCUSSION
1. After looking at the coding rules for the Supreme Court Database, can you see
any difficulty? Do you agree with all the coding choices? Sometimes it is not easy or
obvious whether a Supreme Court decision is “liberal” or “conservative,” which can lead
to errors of classification. Can you identify any cases in which this might be a problem?
Think about Lawrence v. Texas, striking down a Texas anti-sodomy law, or a case in
which the Supreme Court strikes down hate speech rules as a violation of the First
Amendment. See Ernest A. Young, Just Blowing Smoke? Politics, Doctrine, and the
Federalist Revival After Gonzales v. Raich, 2005 Sup. Ct. Rev. 1, 11-12 (2005); [cite
Anna Harvey article here]
2. What would you do if a case involves more than one issue? United States v.
Lopez involved a challenge to a federal law that prohibited possession of a firearm within
proximity to a school. In Lopez, the Court struck down the federal law in question and, in
so doing, invalidated a criminal conviction. Lopez is an important cases on the (legal)
issue of federalism, but it also involved a criminal proseuction. The Spaeth database
codes a case as “liberal” if it reverses a criminal conviction and “conservative” if it limits
national power on federalism grounds. Which issue should prevail? (The Spaeth database
codes Lopez as a federalism case, and accordingly as conservative.) Is it possible to
resolve such intra-case conflicts without sacrificing either the reliability or the validity of
the measure? Professor Ernie Young has argued that the answer is no:
When the prearranged criteria conflict, one of two things seems likely
to happen: The Coders may resolve the conflict based on some gestalt
judgment that would be difficult to articulate and defend.
Alternatively, the coders might analyze the doctrinal issues in the case
to determine which dimension of the decision (e.g., Lopez’s restriction
of national power or its windfall to the individual defendant) is more
salient. But this approach accords a critical role to doctrine that the
attitudinalists generally deny. In any event, the indeterminacy of
coding cases . . . casts considerable doubt on the attitudinal model’s
great strength—its strong predictive success.26
3. Can you think of other ways that you might be able to take a set of opinions and
create a dependent variable that didn’t rely on a liberal-conservative coding?
25
Landes & Posner Rational Judicial Behavior 776.
26
Young, supra, at 12.
18
4. Professor Revesz’s solution required him to determine whether a challenger is an
industry challenger or an environmental challenger. Just as judges don’t wear blue or red
robes to tell us their ideology, litigants don’t come with labels attached either. How
would you figure out what type of challenger a particular litigant was in a case? What do
you suppose Professor Revesz did?
5. Can you think of a type of study where looking at affirmances and reversals
would be more interesting than looking at who won or who lost?
6. Some studies that look at collegial courts—courts like the U.S. Supreme Court or
the circuit courts where panels of judges decide cases—use the outcome of the case as
the dependent variable. Other studies, like Professor Revesz’s, treat the decisions reached
by each judge in the case as the dependent variable. Which do you think is better? In
which situations would you prefer one over the other?
E. Control Variables, and Law
1. The Problem of Omitted Variable Bias
It is not enough to have a dependent variable we want to explain and an
independent variable that explains it. Why? Suppose we are interested in testing the effect
of the length of felony sentences on crime rates. We us the average sentence for felonies
as our explanatory or independent variable and the statewide crime rate as our dependent
variable. We find that crime rates are indeed lower in states with longer felony
sentences, and that the relationship is a strong one – the longer the sentences, the more
we bring down crime rates. So, our hypothesis is proven, right?
Not necessarily. Suppose that another researcher points out that state
unemployment rates might have causal effect. So, we add that as another independent
variable. And upon doing so we learn that the relationship between unemployment rates
and crime rates is itself strong and statistically significant. But our results also suggest
that the relationship between felony sentences and crime rates is much weaker than we
originally thought, though still significant. What have we learned? That to address crime
rates we may need to do something about unemployment, and that though felony
sentences may matter, it may not be nearly as much as we originally thought.
This is the problem of omitted variable bias. It is what empiricists mean when
they say they are “controlling for” factors other than the primary independent variable.
They are seeking to make sure that what is being observed is not an artifact of something
other than what they hope to study. They want to make sure they didn’t leave some
explanatory factor out. If some factor other than the independent variable helps explain
the dependent variable, then results that seem not to be statistically significant may be
(and vice versa).
The same problem arises in empirical studies of judging. If we don’t compare
decisions in like cases, any findings might solely be due to the cases we see, not any true
19
underlying differences in behavior. Suppose we wanted to test whether more conservative
judges are more award harsher sentences than more liberal judges. And so we compare
the records of Judge Reynolds, a conservative Republican appointee, and Judge Drew, a
liberal Democratic appointee. We find that Judge Reynolds gives sentences that on
average are 10 years longer than those of Judge Drew. Would it make sense to conclude
that the conservative imposes harsher sentences. Well, maybe. But what if we
subsequently learned that Reynolds heard felony cases and Drew misdemeanor cases.
That fact alone might explain the different sentences, confounding our conclusion about
ideology.
One way to handle this problem in practice is to use something called regression
models. Regression models let the researcher isolate the effect of key independent
variables—like ideology—on the dependent variable while holding all other variables
constant.
But if one does not include the right set of control variables, things can go very
wrong. Judge Edwards has emphasized this point:
“Omitted variable bias” refers to the possibility that factors which are
correlated with the variables studied, but are not included in a study’s
analytic framework, actually explain the data better than do the factors
considered by the investigator. A variable omitted from the study,
which is itself related to one or more of the included variables, may
explain the data better than the included variables. This problem is
most extreme in any sort of univariate analysis: such an analysis, as its
name indicates, tests for only one variable, and therefore always
allows the possibility that one or more of the multitude of other
potentially relevant variables, because of its correlation with an
included variable, explains the outcome better than does the variable
tested.27
Omitted variable bias is an important problem. Simply put, if a scholar does not
control for the right set of independent variables, any inferences drawn from a statistical
analysis might be wrong (a simpler term for “biased”). Yet omitted variables are only
consequential if they are related to the dependent variable. Any empirical study can be
criticized for “leaving something out,” just as every doctrinal study can be criticized for
not synthesizing every case. The weight we can put on the conclusions of a study depend
on how thorough and careful a job the analyst does in subjecting their central independent
variable to careful scrutiny of possible alternative explanations.
27
Edwards, Collegiality and Decision Making on the D. C. Circuit, supra, at 1350-1351.
20
2. Controlling for Law
There is one type of independent variable that is especially important to consider
when studying judicial decisionmaking: law.28
If the law itself has any bite—and, as we
saw in Chapter Two, there are many reasons to think it does—it is important to take the
law into account when studying differences in behavior. Otherwise law itself, not
ideology, might explain differences in behavior.
Judge Edwards and a co-author find that many empirical studies of judging are
lacking in this regard:
Legal scholars remain interested in trying to use empirical methods—
most notably the statistical analysis of case outcomes—to understand
the effect of extralegal factors on appellate decisionmaking. In our
view, the principal problem with such empirical legal analyses is that
they cannot distinguish between legal and extralegal factors without
considering and accurately accounting for the most important
determinants of appellate decisionmaking: (1) the case records on
appeal, (2) the applicable law, (3) controlling precedent, and (4)
judicial deliberations. By failing to take account of these core
determinants—in part, perhaps, because they cannot be easily or
accurately measured—the field of empirical legal studies fails to
provide a nuanced understanding of how legal and extralegal factors
interact to generate judicial decisions, and likely overemphasizes
extralegal factors.29
Some scholars have recognized the need to have some control variable for legal
doctrine. The problem is that it is not so easy. Stop and think a moment. How would you
control for law? What would the variable be?
One good ieas of how to incorporate law as an independent variable is, ironically,
Professor Jeffrey A. Segal’s work on search and seizure. We say ironically because Segal
today is one of the primary proponents of the attitudinal model. Yet, in his early work he
seemed very attuned to capturing law in his regressions. During the 1960s and 1970s, the
Warren Court begin pushing the law of search and seizure to the left, for example by
applying the exclusionary rule to the states in Mapp v. Ohio, 367 U.S. 643 (1961). (The
exclusionary rule says that courts must throw out any evidence of guilt that was seized as
a result of an unlawful search or seizure.) Richard Nixon ran for president in 1968 on a
law-and-order platform, and after he was elected had the opportunity to appoint four
justices to the Supreme Court, including Chief Justice Burger in 1969. From 1962-1981
the Supreme Court decided over one hundred cases dealing with search and seizure.
28
Professor Edwards also has emphasized another important (and often omitted) variable:
deliberation on collegial courts. We discuss that issue in detail in Chapter XX.
29
Harry T. Edwards & Michael A. Livermore, Pitfalls of Empirical Studies that Attempt to
Understand the Factors Affecting Appellate Decisionmaking, 58 DUKE L. J. 1897, 1899 (2009).
21
Understanding how their rules fit together coherently by reading the cases and focusing
on legal rationales was difficult, to say the least.30
Rather than read each and every Supreme Court search and seizure case and try to
make sense of them all inductively—something that the law professors of the day could
do with only limited success—Professor Segal developed what he called a “legal model”
of Supreme Court decisionmaking. He studied all of the search and seizure cases from
1962-1981. The dependent variable in the analysis was whether the Court upheld the
search. He then coded a number of legal factors that he posited (from reading some of the
cases) might affect decisionmaking: the nature of the intrusion (if the search took place at
the defendant’s home, his business, his car, or on his person in public), the extent of the
intrusion (if it was a full search and seizure or a limited intrusion such as a stop and frisk
or detentive questioning), prior justification (whether there was a warrant or probable
cause), arrest (if it was incident to a lawful arrest, after a lawful arrest, or related to an
unlawful arrest), and some exceptions such as whether the search and seizure was after a
hot pursuit. Of course, Segal also considered the political composition of the Court.31
Segal’s model proved remarkably successful. Using both legal and political
factors, his model correctly classified over three-quarters of the cases used in the study,
and the model was able to correctly forecast six of the seven cases decided during the
1981 term of the Court. Professor Segal concluded:
[T]he model has taken one of the Court’s most difficult areas, one
characterized as “inconsistent,” “bedeviling,” and even “a mess” and
found for the first time a clear and logical form. The legal model was
quite satisfactory, yet evidence suggests that the Court is not immune
from considering extralegal characteristics too. Perhaps the reason for
the bedevilment is that those who have been bedeviled have been too
reliant on the case method as the means for analyzing these decisions.
Such reliance seems reasonable enough. The opinion delivered in any
given decision more often than not relies upon a single aspect of the
case, for example, was there probable cause, or is a search several
hours after a lawful arrest reasonable? Yet the decision itself is
dependent upon many factors—presumably those mentioned above—
that go largely undiscussed in the main body of the opinion. In a
probable cause determination, it is all very well to know that probable
cause was sufficiently established, but when it comes to understanding
the decision itself, we must know where the search took place, the
30
See Ronald Dworkin, Fact Style Adjudication and the Fourth Amendment: The Limits of Lawyering,
48 INDIANA. L. J. 329, 329 (1973) (calling the law in this area, “a mess”).
31
Jeffrey A. Segal, Supreme Court Justices as Human Decision Makers: An Individual-Level Analysis
of the Search and Seizure Cases, 48 J. POL. 938-944 (1986); Jeffrey A. Segal, Predicting Supreme Court
Cases Probabilistically: The Search and Seizure Cases, 1962-1981, 78 AM. POL. SCI. REV. 891 (1984).
22
extent of the intrusion, and whether any exceptions were present, as
well as the makeup of the Court.32
Two other models incorporating legal doctrine as an independent variable of
Supreme Court decisionmaking are worthy of note. Professors Tracey George and Lee
Epstein published an important study of Supreme Court death penalty cases from the
1971 though 1988 terms.33
They modeled the justices’ decisions on whether or not to
uphold a judgment imposing death. The authors hypothesized that case outcomes will be
influenced by a number of doctrinal factors: whether the killing was intentional, whether
the jury was selected to be biased in favor of the death penalty, whether results of stateinitiated
psychiatric examinations were considered, and whether the mitigating or
aggravating circumstances of a particular case had been considered. George and Epstein
found that the Court reacted consistently to doctrinal cues as hypothesized, indicating that
in this area, using the variables they identified, over some of the cases in their domain, a
legal model is effective in explaining outcomes (although the composition of the Court
matters as well).
Another study by Professors Mark Richards and Herbert Kritzer incorporated law
in a novel way. They argued that “the central role of law in Supreme Court decision
making is not to be found in precedents that predict how justices will vote in future cases.
Rather, law at the Supreme Court level is to be found in the structures the justices create
to guide future decision making….”34
Richards and Kritzer developed the concept of a
jurisprudential regime, which refers to “a key precedent, or a set of related precedents,
that structures the way in which the Supreme Court justices evaluate key elements of
cases in arriving at decisions in a particular legal area.”35
Based on their reading of the
freedom of expression law, they suggested that two 1972 cases, Chicago Police
Department v. Mosley, 408 U.S. 92 (1972), and Grayned v. City of Rockford, 408 U.S.
104 (1972), established the speech-protective content-neutrality regime. Using a handful
of political variables and jurisprudential variables, with interactions at the time of the
regime shift, their empirical models show that these two cases fundamentally shifted the
jurisprudential approach in this area of law; i.e., these cases defined new a jurisprudential
regime. They have applied this approach to other areas of law as well.36
32
Segal, Predicting Supreme Court Cases Probabilistically, supra, at 899-900.
33
Tracey E. George & Lee Epstein, On the Nature of Supreme Court Decision Making, 86 AM. POL.
SCI. REV. 323 (1992).
34
Mark J. Richards & Herbert M. Kritzer, Jurisprudential Regimes in Supreme Court Decision
Making, 96 AM. POL. SCI. REV. 305, 306 (2002).
35
Id. at 308.
36
Herbert M. Kritzer & Mark J. Richards, Jurisprudential Regimes and Supreme Court
Decisionmaking: The Lemon Regime and Establishment Clause Cases, 37 LAW & SOC’Y REV. 827
(2003); Herbert M. Kritzer & Mark J. Richards, The Influence of Law in the Supreme Court’s
Search-and-Seizure Jurisprudence, 33 AM. POL. RES. 33 (2005); Mark J. Richards, Joseph L. Smith &
Herbert M. Kritzer, Does Chevron Matter?, 28 LAW & POLICY 444 (2006). However, some scholars
have noted problems with this approach. See Scott. Reconsidering the Impact of Jurisprudential
Regimes, 87 SOCIAL SCIENCE QUARTERLY 380 (2006); Jeffrey R. Lax & Kelly T. Rader, Legal
Constraints on Supreme Court Decision Making: Do Jurisprudential Regimes Exist?, 72 J. POL. 273
(2010); Herbert M. Kritzer and Mark J. Richards, Taking and Testing Jurisprudential Regimes
23
FOR DISCUSSION
1. Incorporating law into a study of judicial decisionmaking requires substantive
knowledge about an area of law. To incorporate law effectively, it is typically necessary
to focus on a single area of law. Why do you suppose that’s the case? Can you see how
developing law as an independent variable is likely to be costly for empirical researchers?
2. There is another difficulty, which becomes apparent when looking more closely at the
George and Epstein model. One feature of their model is that the effectiveness of the
legal factors in explaining outcomes decreased as time passed. (This was also true of
Professor Segal’s study.) Do you have any insight into why this might be the case? In a
common law system like ours, the law is changes with every decision. Does this affect
the efficacy of including legal factors as independent variables? Ultimately what is
needed is a way to “model” law so that it can be used as an independent variable. But
such a variable is going to need to be dynamic and changeable over time.
F. What We Know (and Don’t Know) About Ideology and
Judging
At this point we’ve covered the major steps of empirical studies: choosing
independent and dependent variables, and constructing (or choosing) a dataset. We’ve
seen some of the things that can go wrong at each step, and what the available measures
might leave out. Now we can start to come to grips with the evidence of ideological
voting.
1. On Generalization
Taken as a whole, what do we know about ideology and judicial decisionmaking?
To answer this question, we must generalize from a set of specific studies (warts and all)
to behavior generally. Generalization is the process of taking information about specifics
and drawing conclusions about the situation across the board. Generalization is
something people do all the time. Indeed, every doctrinal study of law takes a small set of
cases and generalizes from them to understand what the law is.
Generalizations can be reasonable or unreasonable. Unreasonable generalizations
are a bad thing, and are usually called overgeneralization. Can you see why, for example,
taking a study about decisionmaking on the Supreme Court and reaching the conclusion
that all judges behave ideologically would be unreasonable. Similarly, a study about
highly controversial abortion cases in the federal district courts will likely tell us very
Seriously: A Response to Lax and Rader, 72 J. POL. 285 (2010); Jeffrey R. Lax and Kelly T. Rader,
The Three Prongs of a Jurisprudential Regimes Test: A Response to Kritzer and Richards, 72 J. POL.
289 (2010).
24
little about run of the mill bankruptcy cases. Reasonable generalizations are those that
involve judges, in similar courts, deciding similar cases. At bottom, the context in which
the study takes place is the context to which we can generalize. Only by looking across
multiple studies in multiple domains can we get a clear picture of what judges do.
Studies about judges in a single circuit court in a particular type of case in a short
time period tell us something about the precise period of study. But scholars usually are
interested in broader, general questions about judicial decisionmaking, public policy, or
law. We conjecture that many readers of Professor Revesz’s article may have drawn
conclusions concerning decisionmaking on the D.C. Circuit generally, and maybe on the
entire federal courts of appeals, or even all judges. Judge Edwards is critical of this sort
of generalization:
There is absolutely no reason to believe, nor does Revesz ever suggest,
that his limited sample of environmental cases is representative of the
D.C. Circuit’s jurisprudence generally. I almost hesitate to make such
a self-evident point, but, by the overly broad and heedless language of
his ‘conclusions,’ Revesz leaves me no choice.37
Throughout his study, Professor Revesz takes pains to point out that his
conclusions are about these environmental cases—and no others—before the D.C. Circuit
in this time period. His title makes that perfectly clear. At that same time, Professor
Revesz uses these findings to speak about the D.C. Circuit generally. Why?
I do not mean to suggest by this discussion that environmental cases
are necessarily unique in giving rise to the types of ideological
decisions described in my article. There is no compelling reason to
believe that environmental cases would be exceptional in this regard.
My own intuition is that a similar pattern would be found with respect
to a broader universe of cases in which different administrations had
markedly different policy preferences over issues of relatively high
political salience.38
Ultimately his assertion is empirical. What do you think? What does Revesz’s
study tell us about decisionmaking in the D.C. Circuit generally? What about other circuit
courts?
2. The Supreme Court
To make reasonable generalizations from the specific studies, it’s important to
compare those that look at judging on comparable courts. We’ll start with the U.S.
Supreme Court, which sits at the top of our judicial hierarchy.
37
Edwards, Collegiality and Decision Making on the D. C. Circuit, supra, at 1344.
38
Revesz, Ideology, Collegiality, and the D. C. Circuit: A Reply to Chief Judge Harry T. Edwards, supra,
at 813-814.
25
As noted, the strongest advocates in favor of the hypothesis that judicial ideology
affects case outcomes are Professors Jeffrey Segal and Harold Spaeth. They are the
primary proponents of the attititudinal model. They use Segal-Cover scores as the
independent variable, and votes by the justices on the U.S. Supreme Court as the
dependent variable.39
In each area of law, Segal and Spaeth classify the vote each Justice
casts as either liberal or conservative, using the methods we discussed above. Then, for
each Justice, they calculate the percentage of the time that the Justice votes liberally in
the issue area. (This type of variable—where information is combined across a whole
bunch of cases—is called an aggregate variable.)
With these two variables in hand, the next step is to relate them. Does the
percentage of liberal votes of the justices relative to one another seem to be correlated
with how liberal or conservative the Segal-Cover score indicated each justice was?
One easy way to evaluate this question is graphically, with something called a
scatterplot. See Figure XX, which shows four separate scatterplots; one for each issue
area. Each point in each pane of the figure represents a Justice. The dependent variable
(Percent Liberal Voting) is on the y-axis; the independent variable (Segal-Cover Score) is
on the x-axis. Each point shows percent liberal voting and the Segal-Cover score for each
Justice.
To see how well these variables are related to one another, each panel also
contains something called a “regression line,” which summarizes the relationship
between the two variables. If ideology and aggregate voting were unrelated to one
another, what shape do you suppose the relationship between these two variables should
take? If you answered “a flat line with zero slope” you would be correct. In the lowerright
Federalism panel, you’ll see a very shallow line. In this pane, the relationship is
statistically insignificant: ideology doesn’t seem to be explaining variation in behavior
across justices. On the other hand, in Civil Liberties and Civil Rights cases, the lines have
a very steep slope. Indeed, over sixty percent of the variability in voting is explained by
the Segal-Cover scores. In other words, if you knew nothing other than the Segal-Cover
score of each justice, you could correctly predict over sixty percent of the justices’ votes.
In these two panes, the relationships are statistically significant. In Economics cases, the
relationship isn’t as strong, so the line has a lesser slope. In this pane the relationship is
significant, but the percentage of explained variance is less.
These findings show pretty convincingly that ideology is significantly related to
behavior, at least in the areas of civil rights and civil liberties, and to a lesser extent in
economics cases. This is consistent with the findings of Lee Epstein and Carol Mershon,
whose separate study shows that Segal-Cover scores are significantly related to voting
behavior on the Supreme Court, except in federalism, judicial power, and tax cases.40
39
Jeffrey A. Segal & Albert D. Cover, Ideological Values and the Votes of U.S. Supreme Court
Justices, 83 AM. POL. SCI. REV. 557 (1989).
40
Lee Epstein & Carol Mershon, Measuring Political Preferences, 40 AM. J. POL. SCI. 261 (1996).
There is a great deal of additional evidence about ideology on the Supreme Court. For a sampling,
see Glendon Schubert, THE JUDICIAL MIND 275 (1965); David W. Rohde & Harold J. Spaeth
26
Figure XX. Effectiveness of Segal-Cover scores across issue areas for justices serving
during the 1946-2004 terms. The Segal Cover scores are taken from Jeffrey Segal’s
website http:// www.stonybrook.edu/polsci/jsegal/qualtable.pdf. The voting data come
from Lee Epstein et al., Table 6-4: Aggregate Liberal Voting of Judges, 1946-2004
Terms, THE SUPREME COURT COMPENDIUM, 4th
ed. (2006) at 534-537.
This approach, however, does not control for law in the sense we described in the
previous section—that is, simply comparing Segal-Cover scores to votes does not
incorporate other information about the case (e.g., the salient legal facts) that we would
expect to be important predictors of legal decisions. In an attempt to remedy this
problem, Professors Segal and Spaeth expanded the case-facts approach that Professor
Segal pioneered in the search and seizure study we discussed above. The earlier study
had looked at search and seizure cases decided by the Supreme Court between 1962 and
1981. Segal and Spaeth extended the study through 1998, and examined the effect of
ideology (measured by the Segal-Cover scores) while controlling for the same casespecific
factors (e.g., the location of the search, the timing of the search, and the presence
SUPREME COURT DECISION MAKING 157 (1976); and Andrew D. Martin & Kevin M. Quinn, Dynamic
Ideal Point Estimation via Markov Chain Monte Carlo for the U.S. Supreme Court, 1953-1999, 10 POL.
ANALYSIS 134 (2002).
Segal Cover Score
PercentageLiberalVoting
20
40
60
80
100
0.0 0.2 0.4 0.6 0.8 1.0
●
●
●
●
●
●
●
●
●
●
●
●
● ●
●
●
●
●
●●
●
●
●
●
●
●
●
●
●
●
●
●
Economics
●
● ●●●
●
●
●
●
●
● ●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
Federalism
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●● ●
●
●
●
●
●
●
●
●
●
●●
Civil Liberties
0.0 0.2 0.4 0.6 0.8 1.0
20
40
60
80
100
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
● ●
●
●
●
●
●
●
●
●
●
●
●
Civil Rights
27
of a warrant) that Professor Segal used in his earlier work. They found that, even after
controlling for legal factors, ideology has a strong and statistically significant effect.41
These findings, along with the aggregate vote analysis shown in Figure XXX, constitute
some of the strongest empirical evidence for the attitudinal model.
Segal and Spaeth’s basic result—that ideology matters even after we control for
case-specific facts—has held for issues beyond search and seizure. As mentioned earlier,
Professors George and Epstein studied the Supreme Court’s decisions in death penalty
cases between 1972 and 1992, and their results indicated that legal factors were effective
predictors of judicial decisions. In their study, they estimated three models: one model
with only legal factors, another model with extralegal factors (including a measure of the
number of Nixon, Ford, and Regan appointees on the court to capture judicial ideology),
and a combined model that includes both the legal factors and the extralegal factors. The
combined model provides the best predictive power. In short, their results underscore that
law matters, but other factors, including ideology, play an important role, as well.42
The Kritzer and Richards study described in the previous section reached similar
conclusion. Kritzer and Richards studied how the influence of legal factors changed after
a landmark Supreme Court decision. Their analysis spanned a series of studies and many
issue areas: free press, free expression, free speech, Establishment clause, search and
seizure, and administrative law. They measured ideology using Segal-Cover scores, and
they controlled for issue area-specific case factors. Across all areas of law they studied,
they concluded that both law and ideology significantly affect judicial decisions.43
3. Circuit Courts of Appeals
What about the lower courts? Judges in the Courts of Appeals are sandwiched in
the middle of the judicial hierarchy. Their legal opinions may be reviewed by the
Supreme Court, and they oftentimes review the work of district court judges. Compared
with their colleagues on the Supreme Court, what would your hypothesis be about
ideological voting on the Courts of Appeals?
Many studies have been undertaken that look at circuit court judges. We’ve
already seen the conclusions Professor Revesz drew in environmental cases in the D.C.
Circuit. But some scholars have gone beyond that set of cases and that court. When
Senator Schumer called for considering ideology when confirming federal judges, one of
41
Segal & Spaeth, SCAMR, supra.
42
Tracey E. George & Lee Epstein, On the Nature of Supreme Court Decision Making 86 AM. POL.
SCI. REV. 323 (1992).
43
Herbert M. Kritzer & Mark J. Richards, The Influence of Law in the Supreme Court’s Search-andSeizure
Jurisprudence, 33 AM. POL. RES. 33 (2005); Herbert M. Kritzer & Mark J. Richards,
Jurisprudential Regimes and Supreme Court Decisionmaking: The Lemon Regime and Establishment
Clause Cases, 37 LAW & SOC’Y REV. 827 (2003); Mark J. Richards, Joseph L. Smith & Herbert M.
Kritzer, Does Chevron Matter?, 28 LAW & POLICY 444 (2006); Mark J. Richards & Herbert M.
Kritzer, Jurisprudential Regimes in Supreme Court Decision Making, 96 AM. POL. SCI. REV. 305
(2002).
28
the studies he relied on was an early version of a study by Professor Cass Suntein and his
colleagues that posed the question: Are Judges Political?44
Like Revesz, Professor Sunstein and his colleagues used the party of the
appointing President to construct the independent variable: ideology of the judge. They
then structured their analysis around a hypothesis that should sound familiar. They
describe their ideological voting hypothesis in the following way:
In ideologically contested cases, a judge's ideological tendency can be
predicted by the party of the appointing president; Republican
appointees vote very differently from Democratic appointees.
Ideologically contested cases involve ... issues ... such as affirmative
action, campaign finance, federalism, the rights of criminal defendants,
sex discrimination, piercing the corporate veil, racial discrimination,
property rights, capital punishment, disability discrimination, sexual
harassment, and abortion.45
To investigate their hypothesis, the authors then had to construct a dataset. Here’s
what they did:
We examined a total of 4958 published majority three-judge panel
decisions, and the 14,874 associated individual judge's votes, in areas
involving abortion, capital punishment, the Americans with
Disabilities Act, criminal appeals, takings, the Contract Clause,
affirmative action, Title VII race discrimination cases brought by
African-American plaintiffs, sex discrimination, campaign finance,
sexual harassment, cases in which plaintiffs sought to pierce the
corporate veil, industry challenges to environmental regulations, and
federalism challenges to congressional enactments under the
Commerce Clause... To keep the inquiry manageable, our investigation
is limited to recent time periods (sometimes from 1995 to the present,
though sometimes longer, certainly when necessary to produce a
sufficient number of cases in a particular category).46
With each case in hand, the next step was to code their dependent variable. For
each area of law they coded whether the vote each judge cast in each case was
“stereotypically liberal” or not. In close calls, an outcome was coded “stereotypically
liberal” if the party on the liberal side gets any relief.
44
Cass R. Sunstein, et al., ARE JUDGES POLITICAL? AN EMPIRICAL ANALYSIS OF THE FEDERAL
JUDICIARY (2006); Cass R. Sunstein, David Schkade & Lisa Michelle Ellman, Ideological Voting on
Federal Courts of Appeals: A Preliminary Investigation, 90 VA. L. REV. 301, 352 (2004); Confirmation
Hearings on Federal Appointments: Hearings Before the Senate Comm. On the Judiciary, 107th Cong.
765 (2002) (statement of Sen. Schumer).
45
Sunstein, Schkade & Ellman, Ideological Voting on Federal Courts of Appeals, supra, at 304.
46
Id. at 311-313.
29
Finally, to test their hypotheses, Sunstein et al. compared the behavior of judges
appointed by Republican and Democratic judges. We show their main findings with
regard to ideological voting in Figure XX. In every area except Takings Clause,
Federalism, and Criminal cases, there are statistically significant differences in behavior
between judges of the two parties. In some areas of law, these differences are quite large:
in Affirmative Action the difference is twenty-six percentage pages, and in both Abortion
and Capital Punishment the differences are greater than twenty percentage points.
Figure XX. Percent liberal votes for thirteen areas of law comparing Republican and
Democrat judges, from the Sunstein, et al., study of the circuit courts. CITE: Cass R.
Sunstein, David Schkade, and Lisa Michelle Ellman. 2004. "Ideological Voting on
Judge's Party
PercentageLiberalVotes
20
40
60
80
100
Republican Democrat
Campaign Finance
Republican Democrat
Affirmative Action
Republican Democrat
EPA
Republican Democrat
Sex Discrimination
Contracts Corporate Veil ADA
20
40
60
80
100
Abortion
20
40
60
80
100
Capital Punishment Title VII Federalism Criminal
20
40
60
80
100
Takings Clause
30
Federal Courts of Appeals: A Preliminary Investigation." Virginia Law Review. 90: 301-
354 at 352. The differences in all panels are statistically significant except Takings
Clause, Federalism, and Criminal cases.
Here is a sampling of other studies of the circuit courts:
• In an early pair of studies, Professor Sheldon Goldman examined the effect of
party affiliation on voting in the U.S. Courts of appeals in the 1960s. He found
that the political party of the appointing President was a significant predictor of
voting behavior in some issue areas (like labor and personal injury). At the same
time, he found no significant ideological effects in other issue areas (such as tax
and eminent domain).47
Goldman’s general conclusions—that the party affiliation
of the appointing president predicts judicial decisions in some, but not all, issue
areas—is consistent with the findings of the study by Sunstein, et al., that we
reviewed above.48
• In another important work, Professor Frank Cross used a random sample of circuit
court opinions from 1925 to 1991, across all areas of law, to examine deference to
lower courts and ideological voting on the U.S. Courts of Appeals.49
Unlike the
Sunstein and Goldman studies, Professor Cross’s study attempted to control for
the law. Deference to lower courts—a major part of Cross’s inquiry—often is
mandated by the operative standard of review that governs appeals to the courts of
appeals. Deference, in other words, is a legal factor as opposed to an ideological
influence external to law. Cross found that judges on the circuit court regularly
defer to their colleagues on the lower courts. The next most important factor in
explaining outcomes is the political party of the appointing President. But the
magnitude of the party effect is much smaller than deference to the lower courts,
and, just as with Professor Revesz’s study, there is a great deal of unexplained
variance once party is taken into account.
• Another study that attempted to control by law focused on obscenity cases in the
courts of appeals from 1957 to 1990. Professors David Songer and Susan Haire
concluded that “at least for obscenity cases, significant presidential appointment
effects exist even after adding controls for several significant case facts,
differences in legal issues raised by the litigants, and differences in the nature of
the litigants. These effects are both statistically significant and of a nontrivial
magnitude.”50
47
Sheldon Goldman, Voting Behavior on the United States Courts of Appeals, 1961-1964, 60 AM. POL.
SCI. REV. 374 (1966); Sheldon Goldman, Voting Behavior on the United States Courts of Appeals
Revisited, 69 AM. POL. SCI. REV. 491 (1975).
48
Cass R. Sunstein, et al., ARE JUDGES POLITICAL? AN EMPIRICAL ANALYSIS OF THE FEDERAL
JUDICIARY (2006).
49
FRANK B. CROSS, DECISION MAKING IN THE U.S. COURTS OF APPEALS (2007).
50
Donald R. Songer and Susan Haire, Integrating Alternative Approaches to the Study of Judicial
Voting: Obscenity Cases in the U.S. Courts of Appeals, 36 AM. J. POL. SCI. 963, 977 (1992).
31
• In another study, Professors Songer, Martha Humphries Ginn, and Tammy A.
Sarver studied tort cases heard by the circuit courts as part of their diversity
jurisdiction between 1960 and 1988. In diversity cases, federal courts decide
questions involving state law, and the courts of appeals are—for all practical
purposes—the court of final review, because the Supreme Court will not take
cases that turn on state law. In such cases, the authors argued, “appeals court
judges are in the same institutional position as Supreme Court justices except for
the power to select which cases they will consider on the merits” because there is
virtually no Supreme Court review of these cases, state courts cannot review
them, and Congress has never passed a law overturning a decision in this issue
area.51
Their model included an impressive number of control factors, including
resource differences between the litigants, state tort law, state ideology, and
judicial ideology. Their measure of ideology is a composite measure that
combines the political party of the judge, the party of the appointing president,
and a number of different measures of presidential ideology. Their findings reveal
that, in diversity tort cases, both law and ideology matter. [I wasn’t able to figure
out how the authors purport to identify the political party of the judge. Do you
guys know? After having spent so much time on how to ID ideology, I think we
ought to be explicit about this kind of thing.]
4. District Courts
Finally, there are the federal district courts. Studies of judging in the district
courts are few and far between. The Rowland and Carp study and the Ashenfelter, et al.
study discussed in Section C.2 investigated the importance of political party of the judge
in the district courts.52
When looking at the entire docket, Ashenfelter et al. found no
effects of political ideology. However, studying only published opinions in civil liberties
cases, Rowland and Carp found significant differences between judges of different
parties.
While there are comparatively fewer studies that examine the role of ideology in
the decisionmaking process of District Court judges, two studies are illustrative:
• Professor Randall Lloyd studied published district court decisions in
reapportionment cases between 1964 and 1983. His results indicate that a district
court judge is 55% more likely to vote against a reapportionment plan drawn up
by the opposing political party than a nonpartisan plan, even when controlling for
51
Donald R. Songer, Martha Humphries Ginn, & Tammy A. Sarver, Do Judges Follow the Law
When there is No Fear of Reversal?, 24 JUST. SYS. J. 137, 139 (2003).
52
Professors Rowland and Carp derived the political-party measure from Who’s Who and other
biographical sources. (p.295). Professor Ashenfelter et al. explain that “[s]tandard biographical
dictionaries and other sources supplied the political party . . . .” Orley Ashenfelter et. al., Politics and
the Judiciary: The Influence of Judicial Background on Case Outcomes, 24 J. Legal Stud. 257, 266
(1995). [Should we add a brief section discussing these biographical sources on judges?]
32
the political party of the judge. Moreover, he finds that Democrats are 36% more
likely to vote against any reapportionment plan than Republicans.53
• Professor Paul Collins, Jr. and his coauthors studied decisionmaking in 48,000
published alien petitions, criminal rights, discrimination, freedom of expression
and religion, habeas-corpus, military exclusion, and privacy cases decided
between 1938 and 2004 to determine if district court decisionmaking differs in
times of war. While they found no evidence of a wartime effect in their statistical
analysis, their results indicate that judges appointed by Republican presidents are
11% more likely than Democratic appointees to reach a conservative decision in
these cases, even after controlling for the preferences of the circuit court and the
U.S. Supreme Court. However, the analysis doesn’t include any case- or lawspecific
factors, nor does it differentiate between issue areas.54
FOR DISCUSSION
1. Taken as a whole, these studies support the intuition that the ideology of the judge
matters in situations when judges have discretion. Supreme Court justices show the most
ideological behavior, circuit court judges less, and trial court judges only in the rare
circumstance when they are presented with novel legal questions. These findings are
consistent with the meta-analysis—a type of statistical study that combines results from a
bunch of other studies—conducted by Professor Daniel Pinello. Pinello looked at every
study between 1959 and 1998 that analyzed the effects of ideology (usually gauged by
the political party of the appointing President) on judging. He found that in the federal
courts, the effects of ideology are greatest in the Supreme Court, less in the circuit courts,
and quite small in the district courts.55
As Professors Gregory Sisk and Michael Heise
write,
Does ideology matter in the selection and decisionmaking of federal
judges? Unquestionably. The expanding literature of empirical
research on the subject, including our own, makes it starkly clear that
the partisan or political dimension is, among others, an inescapable
element in understanding judicial behavior. But, at least at the lower
federal court level, ideology explains only part of judicial behavior and
tends to emerge in certain narrowly defined sets of cases in studies
designed to tease out those marginal effects.56
53
Randall D. Lloyd, Separating Partisanship from Party in Judicial Research: Reapportionment in the
U.S. District Courts, 89 AM. POL. SCI. REV. 413 (1995). Professor Lloyd explains that “[p]arty of the
judge was found in published sources, such as ‘Who’s Who in American Law.’”
54
Paul M. Collins, Jr. et al., International Conflicts and Decision Making on the Federal District
Courts, 29 JUST. SYS. J. 121 (2008).
55
Daniel R. Pinello, Linking Party to Judicial Ideology in American Courts: A Meta-analysis, 20 JUST.
SYS. J 219 (1999).
56
Gregory C. Sisk & Michael Heise, Judges and Ideology: Public and Academic Debates about Statistical
Measures, 99 NW. U. L. REV. 743, 793-794 (2005).
33
Do you find it surprising that there is more evidence of ideological judging on the
Supreme Court relative to the lower courts? Why do you think this is?
2. Some of the better studies try to take law into account as an independent variable. As
we’ve made clear, this is no easy task. On the Supreme Court, the findings of ideological
effects are robust whether one includes law or not. Studies that look at the circuit courts
and control for law find some modest ideological effects in some issue areas, just like
Professor Revesz did. And the studies that look at district court decisionmaking only find
ideological effects in the vanishingly rare cases when opinions are published (and even
then, not all the time). Assume that these findings are valid. Do they show that Supreme
Court Justices are willfully ignoring the law? Or is there another explanation?
Judge Posner writes that “[t]he Court is awash in an ocean of discretion” and
argues that Supreme Court cases continually present new twists, making precedent less
than determinative.57
This is entirely consistent with Judge Cardozo’s claim that the
indeterminacy law itself provides space for judicial discretion. Even Professor Brian
Tamanaha—a critic of empirical studies of judging—recognizes the amount of discretion
Supreme Court justices have:
The Supreme Court hears about one-tenth of 1 percent of federal court
cases, a miniscule number winnowed and selected out from the
massive total. These are legally uncertain cases or cases that involve
bad rules or bad results. They often have political salience. . . .
Inevitably the personal views of justices will have an impact in a
measurable proportion of their decisions.58
It is precisely for this reason that Professor Tamanaha concludes:
The region of uncertainty is where judges render decisions with the
least legal guidance, and where judges’ particular mix of legal and
social views has the most leeway and impact – though still in a context
thick with legal norms. This is why the political views of Supreme
Court justices, who hear the highest proportion of legally uncertain
cases, manifest a much stronger relationship with their legal decisions
(although significantly short of a complete alignment) in comparison
with lower court judges.59
Indeed, at some level the most interesting question is why we don’t see
ideological effects in more Supreme Court cases. First, there are a lot of unanimous
decisions on the Supreme Court. Can these be explained by ideology? If not, what does
explain them? There also are many cases where the votes don’t align with common
57
See Richard A. Posner HOW JUDGES THINK at 272.
58
Brian Z. Tamanaha, BEYOND THE FORMALIST-REALIST DIVIDE: THE ROLE OF POLITICS IN
JUDGING 197 (2009).
59
Id. at 190.
34
political views, like the guns-and-drugs cases we examined in Chapter Two. What
explains outcomes like these?
3. Notice that in any number of these studies, in some issue areas like federalism, or
judicial power, ideology doesn’t relate to behavior? Do you have any ideas why? Do you
think it means that these issues aren’t political? That the cases are easy? Or perhaps that
there are flaws in the way the dependent or independent variables are constructed?
4. Coding remains a problem in many of these studies, for reasons we have
examined. For example, Sunstein et al. decided to code “any relief” as a win. Do you
agree with that decision? Suppose a criminal sentenced to thirty years in prison is able to
get his sentence reduced by a week due because of a misapplication of law. Does it make
sense to treat this type of relief as equivalent to an entire conviction being overturned?
Does this have implications for the reliability or validity of the measure? Do you think
this approach is more reliable than other possible approaches?
5. Finally, there are important gaps in our knowledge, which point to important areas for
future study. First, and most important, almost all of these studies look at public law—
law that governs the relationship between individuals and their government, not between
private individuals. Indeed, most studies that find an ideological effect are looking at civil
liberties cases. Can you see why selecting only these sorts of cases may introduce some
selection bias in our conclusions?
The truth is, we simply don’t know how much ideology might (or might not)
affect behavior in private law disputes. Yet stability in private law is of utmost
importance to structuring relationships, economic an otherwise. As Dean Roscoe Pound
once wrote:
The purpose of matured law is not merely to do justice in each
concrete cause but also to furnish a rule in advance of action by which
men may be guided with assurance in the complicated transactions of
modern business; to insure stability of industrial undertakings, to
provide assured constancy of the conditions under which property is
held and business carried on… They may conceivably perform these
functions without law. But if they are to perform them according to
law, they must perform them in a certain, uniform way. For certainty
and uniformity are the essential attributes of law.60
Second, and related, we also don’t know much empirically about common law
decisionmaking. Professor Revesz’s study looked at behavior in cases involving
administrative rulings. Many studies look at decisionmaking in constitutional cases, or
those that involve statutes. Today the U.S. Supreme Court rarely hears common law cases
(although constitutional cases come closest to being a species of common law). We might
expect that in common law cases—where judges are, essentially, adapting law to the facts
of each case, while maintaining fidelity to existing precedent and broad principles—
60
Roscoe Pound, The Enforcement of Law, 20 GREEN BAG 401, 408 (1908).
35
judges will exhibit all sorts of ideological behavior. But Judge Posner disputes this
intuition:
Since common law is explicitly judge created, one might think it
would be bound to be even more “lawless” than constitutional
adjudication. In fact it is more stable, more objective, more lawlike,
and less like “real” legislation than constitutional law is. It deals with
subjects in which there is usually a high degree of consensus in both
society at large and the judiciary, enabling judges of diverse
backgrounds and political commitments to reason from common
premises to conclusions that command broad support within the
judiciary and the larger political community. That persons of diverse
backgrounds, values, and so forth can be brought to agree on a matter
is the practical meaning of “objective.”61
61
Richard A. Posner, HOW JUDGES THINK, 82-83 (2008). For a recent study, see Mark P. Gergen and
Kevin M. Quinn, Common Law Judicial Decision Making: The Case of the New York Court of Appeals,
1900-1950, Paper Presented at the 2011 Emory Conference on Institutions and Lawmaking.