(Page 9 of 17) - Jurisprudential Regimes and Obscenity Doctrine authored by Buchman, Jeremy.

•

Government: The dummy variables in this group contain information about the
government actor claiming that the work at issue is obscene. FED is coded 1 for cases
where a federal government actor brought obscenity charges, and 0 otherwise. The
dummy variable STATE is coded similarly. If state prosecutions draw upon more
localized notions of obscenity than federal prosecutions, then we would expect to see
Miller have a stronger effect on STATE than on FED.

•

Type of material being challenged: Dummy variables are coded 1 if the challenged
material included films or videos (FILM), primarily textual depictions (BOOK), or
magazines (MAGAZINE), and 0 otherwise. These variables attempt to capture, in
admittedly rough fashion, the perceived social value of the material in question. In
light of Miller’s third prong, media forms that include both mainstream works with
some erotic content and hard-core pornography should be affected less by Miller than
media forms where sexual content is predominantly explicit and pandering in its
explicitness.

My expectation, then, is that Miller will affect challenges to magazines

more than challenges to films or books, which include a wider range of material that
has sexual content but would be less likely to be deemed pornographic. Where the
case syllabus or majority opinion does not describe the challenged material, I rely on
the opinion of the last lower court to rule on the dispute or, where no citation to a
lower court was provided, to the party variables included in the Justice-Centered
Databases.

•

Community standards: In the absence of sufficiently precise—or, oftentimes, any—
polling data on different communities’ views regarding sexually explicit material, any
proxy for community standards can provide, at best, a rough approximation. My
coding of strict and lenient states (1 for strict, 0 for lenient) comes from Volokh
(2001, 167-68), who relies on marketing information provided to him by a
pornography distributor who sells nationwide through mail order or the Internet. This
distributor, like many others, forms a code list based on the approximate number of
obscenity prosecutions in a ZIP code grouping. (Information concerning the type of
prosecutions is harder for distributors to come by.) Because distributors who sell
nationwide risk prosecution in communities with a broad range of standards
concerning sexual explicitness (see Sable Communications of California, Inc. v. FCC
(1989)), they have strong incentives to know which communities are less permissive,
so that they can tailor their mailings in a manner that minimizes the odds of
prosecution. This proxy does assume that a state’s strictness or leniency remains
constant throughout the time frame under analysis, but in the absence of micro-level
data that can capture temporal shifts in public opinion concerning obscenity, any
other proxy will similarly rely on this assumption.

To test whether the post-Miller models differ from their pre-Miller counterparts, I follow Kritzer
and Richards’ (2003; also see Richards and Kritzer 2002) lead and employ a variation of the
Chow test (Hanushek and Jackson 1977) that compares the -2 log likelihood of the all-justices
regression to the -2 log likelihood of a regression in which the regime dummy is interacted with
the other independent variables.

I also perform a similar analysis using only the Miller justices.

If the addition of these interaction terms produces a statistically significant improvement in each

I do recognize the possibility that this distinction brings me perilously close to Stewart’s “I know it when I see it.”

I present separate regressions for the pre-Miller and post-Miller sets of cases, rather than the regression with the

regime dummy interactions, because the coefficients are easier to interpret, especially with respect to any changes in
independent variables as a result of Miller.