Box-Steffensmeier, J., Christenson, D., and Hitt, M (2013).
Quality Over Quantity: Amici Influence and Judicial Decision Making. American Political Science Review 107(3):446-460.
Question 1: This article tests whether amici curiae influence the Supreme Court based on their relative
power and influence. It justifies this hypothesis through the literature that
explains that repeat advocates succeed more at the Supreme Court,[1]
arguing that amici are similar,
particularly since the Solicitor General is regarded as having outsize
influence as an amicus, rendering the
argument for treating amici similarly
incoherent. The hypothesis is further justified as a way to test between the
legal and attitudinal models of judicial decision making. The literature on the
Court’s internal processes is invoked to explain why differential status of amici matters. The authors also argue
that quality of briefing matters, but offer no literature to support that
insight (which is, frankly, trivial), and don’t test it. In fairness, it’s
probably unmeasurable under their research design.
Question 2: The dependent variable is a justice vote. It is
measured using the “traditional” U.S. Supreme Court dataset, the “Spaeth”
dataset, in a justice-centered data format, which contains (usually) nine
entries for each case – one for each justice participating. It is internally
valid because the authors want to test how justices are impacted by amici, not how case outcomes are
affected by amici. It is reliable
because the dataset is publicly available and can be reconstructed by anyone
using Spaeth’s codebook.
Question 3: The primary independent variable is “eigenvector
centrality”, which uses network theory to evaluate the amicus’ connectedness to other groups that appear before the Court.
It is measured by counting links between groups to create a network, and then
measuring the centrality of a group’s position in the network. A link is
created when two groups cosign an amicus
brief together. When multiple groups sign one brief, bilateral links are
created between each of them. When two groups sign multiple briefs, one link is
created for each brief they both sign. The centrality is then measured using
eigenvalues and eigenvectors. I have to confess that I looked up “eigenvector”
and “eigenvalue” in writing this answer, and I still don’t understand how this
works. But, I gather that this measure is preferred because it captures a
feedback loop of centrality. That is, it captures the fact that central network
nodes are connected to other central network nodes, and thus differentiates
between nodes that are truly central, nodes that are peripheral to the core,
and true periphery nodes. This variable is internally valid because the authors
are seeking to measure the power of mutual trust on the reputation of actors
with regard to a third party. In short, A and B are signaling that they trust
each other – how does this signal effect C’s reaction to A and B’s argument?
And to what extent does that signal create feedback loops that resonate through
the network? I’m not sure if this variable is reliable, because I can’t figure
out how they calculated it. However, presuming that I can, at some point, learn
how to do this, I assume that the variable follows from its constituent data.
Indeed, since the authors published the data that form their network datasets,
I assume that it can be reproduced if necessary.
Question 4: The authors control for four variables: justice
ideology, the presence of the solicitor general, the resources of the
litigants, and the ideological direction of the lower court’s decision. All of
these variables are justified. The last three are justified by reference to the
literature which describes them as confounders; justice ideology is justified
because the authors argue that it interacts with their network centrality
variable to produce effects. The numbers bear them out: judicial ideology is
strongly interacted with the presence of a core amicus.
Question 5: This article attempts to present evidence of
whether amici affect courts through
skillful advocacy or through extralegal mechanisms (such as signaling). The
authors’ results suggest that at least some of their influence comes from
extralegal mechanisms. However, they acknowledge that their research is limited
because it does not analyze the content of briefs. What they are testing (the
power of networks to serve as a signal) does not require content analysis, so
to that extent, their disclaimer is unnecessary. However, to the extent that
they’re interested in the broader questions, I agree that studying brief
content and quality matters.
Question 6: I would extend this article through qualitative
methods. I would use a most-similar case study method to examine cases where
the amici were substantially similar
(or if possible, identical) in their network centrality measures, but the
results were opposite. I would then analyze the content of the briefing record
before the Court in comparison with its opinion in the cases to examine what
effect the briefs had on the actual opinion. This is important, because courts
plagiarize briefs that they regard as being of a particular quality.
[1]
Frustratingly, the authors do not cite my article on this topic, which of
course renders their findings unreliable.