Download as pdf or txt
Download as pdf or txt
You are on page 1of 32

Explaining Spatial Variation in Support for

Capital Punishment: A Multilevel Analysis1


Eric P. Baumer
University of Missouri—St. Louis

Steven F. Messner
University at Albany, SUNY

Richard Rosenfeld
University of Missouri—St. Louis

This research examines the effects of social context on support for


the death penalty using individual-level data from the 1974–98 Gen-
eral Social Survey (GSS), which have been linked with aggregate-
level data on homicide rates and sociodemographic, political, and
economic characteristics. Consistent with instrumental, social threat,
and constructionist perspectives, this study finds that residents of
areas with higher homicide rates, a larger proportion of blacks, and
a more conservative political climate are significantly more likely
to support the death penalty, net of compositional differences. These
results warrant further attention to contextual and individual
sources of public support for the death penalty.

The United States is one of the few developed societies in the world that
retains the death penalty. Various explanations for this aspect of American
exceptionalism have been proposed, including distinctive features of
American federalism and the populist nature of American politics (Zim-
ring and Hawkins 1986; Hood 1998; Radelet and Borg 2000). Whatever

1
We thank Tom Smith, National Opinion Research Center (NORC), for valuable
technical assistance, and Scott South, Janet Lauritsen, and the AJS reviewers for
helpful comments on an earlier version of this article. The research reported in this
article was supported in part by a grant from the National Consortium on Violence
Research (NCOVR), which is supported by the National Science Foundation (grant
SBR 9513040). Direct correspondence to Eric Baumer, Department of Criminology
and Criminal Justice, University of Missouri, St. Louis, Missouri 63121. E-mail:
baumer@umsl.edu

䉷 2003 by The University of Chicago. All rights reserved.


0002-9602/2003/10804-0003$10.00

844 AJS Volume 108 Number 4 (January 2003): 844–75

This content downloaded from 128.252.067.066 on July 25, 2016 21:40:49 PM


All use subject to University of Chicago Press Terms and Conditions (http://www.journals.uchicago.edu/t-and-c).
Support for Capital Punishment

the merits of these accounts, capital punishment receives substantial pub-


lic support in the United States. Recent national surveys indicate that
about two-thirds of American adults support the death penalty for persons
convicted of murder.2 However, the national figure conceals the substan-
tial variation in death penalty support that exists across space within the
United States. Some studies have demonstrated significant regional var-
iation in levels of support (e.g., Bohm 1991; Fox, Radelet, and Bonsteel
1991), and independently conducted state-level surveys indicate that the
often-quoted national figures do not adequately describe public sentiment
in all U.S. states. For example, in 1999, support for the death penalty was
much lower in Kentucky (59%) than in Missouri (78%) (Brinker 1999;
Death Penalty Information Center 1999). But with the exception of this
type of descriptive evidence for large geographic units, very little is known
about spatial variation in support for the death penalty in the United
States, including how much variation exists at relatively localized areas
and what social conditions might account for that variation.
We begin to fill some of these gaps in the literature by examining the
sources of variation in death penalty support across a representative sam-
ple of U.S. metropolitan areas and nonmetropolitan counties. Our analyses
address two interrelated questions: Is there meaningful variation in sup-
port for capital punishment across these localized areas, and if so, how
can this variation be explained?
With respect to the latter question, we are particularly interested in
exploring contextual effects. Support for the death penalty might vary
across areas simply as a function of the nonrandom distribution of the
population. Specifically, areas with strong death penalty support might
be those with relatively large numbers of persons with the individual
attributes that have been linked with pro–death penalty attitudes. Prior
research and theory, however, suggest that attitudes about punishment
and social control are likely to be affected by features of the social en-
vironment, especially the level of lethal violence, political conservatism,
and racial and economic composition. Drawing on this literature, we
formulate hypotheses about the contextual determinants of death penalty
support and test these hypotheses with a unique data set that permits
multilevel modeling.

2
Gallup poll results for 2001 reveal that 67% of American adults support the death
penalty for persons convicted of murder, while a recent Harris poll (August 2000)
indicates that 64% of American adults support the death penalty (the most recent
Gallup and Harris death penalty poll results are available at www.gallup.com and
www.harrisinteractive.com, respectively). Nearly identical results were obtained from
a 2000 ABC News poll (see Daniel Merkle, “Death Penalty Remains in Favor,” http:
//abcnews.go.com/sections/politics/dailynews/poll000619.html [last accessed February
3, 2003]).

845

This content downloaded from 128.252.067.066 on July 25, 2016 21:40:49 PM


All use subject to University of Chicago Press Terms and Conditions (http://www.journals.uchicago.edu/t-and-c).
American Journal of Sociology

BACKGROUND AND HYPOTHESES


An extensive body of research has accumulated on the relationship be-
tween individual attributes and attitudes toward criminal justice policies,
including the death penalty. For example, past studies have identified a
broad set of personality characteristics and deeply held beliefs, such as
authoritarianism, dogmatism, religious fundamentalism, and a belief in
retribution, that give rise to a worldview supportive of harsh punishment
(Neapolitan 1983; Bohm 1987; Finckenauer 1988; Grasmick et al. 1992;
Smith and Wright 1992). However, although the literature acknowledges
that public opinion about social issues is responsive to macrolevel events
and conditions (e.g., Page and Shapiro 1992; Zaller 1992; Steiner 1999),
very little attention has been directed to features of the social context that
may influence punitive attitudes toward punishment in general and sup-
port for capital punishment more specifically.
Perhaps the most obvious contextual factor that might be related to
death penalty support is the level of homicide (Stinchcombe et al. 1980;
Gross 1998; Garland 2000; Jacobs and Carmichael 2002). Previous re-
searchers have noted that national trends in support for the death penalty
over the past 30 years correspond fairly closely to trends in homicide rates
(Ellsworth and Gross 1994). Less than half of American adults—between
40% and 45%—polled in the mid-1960s reported that they supported the
death penalty for persons convicted of murder (e.g., Zeisel and Gallup
1989; Bohm 1991; Ellsworth and Gross 1994). But support grew steadily
during the 1970s and 1980s and, by the mid-1990s, more than three-
quarters of the adult population were in favor of capital punishment (Warr
1995; Longmire 1996; Shaw, Shapiro, and Lock 1998; Radelet and Borg
2000). This 30-year period of increasing support was followed by an ap-
preciable decline in support during the latter half of the 1990s. As noted
above, recent population surveys indicate that about two-thirds of Amer-
icans polled support capital punishment. In a roughly similar manner,
homicide rates rose from the mid-1960s through the 1980s and then fell
rapidly beginning in the early 1990s and continued to decline through the
decade (Blumstein and Rosenfeld 1998; LaFree 1998).3 Although far from
perfect (Fox et al. 1991; Smith and Wright 1992), this correspondence has
led to the widespread speculation that public opinion about the death
penalty may be sensitive to levels of homicide (e.g., Gelles and Strauss
1975; Thomas and Foster 1975; Smith 1976; Stinchcombe et al. 1980;

3
Between 1993 and 1999, the national homicide rate fell by 40%, from 9.5 to 5.7
homicides per 100,000 persons. The homicide data were compiled by the Bureau of
Justice Statistics and are available at www.ojp.usdoj.gov/bjs/homicide/nomtrnd.htm
(last accessed February 3, 2003).

846

This content downloaded from 128.252.067.066 on July 25, 2016 21:40:49 PM


All use subject to University of Chicago Press Terms and Conditions (http://www.journals.uchicago.edu/t-and-c).
Support for Capital Punishment

Finckenauer 1988; Niemi, Mueller, and Smith 1989; Page and Shapiro
1992; Ellsworth and Gross 1994; Schneider 2000; Shapiro 2000).
Two main theoretical rationales have been cited for the apparent em-
pirical association between homicide rates and death penalty support.
One emphasizes instrumental/pragmatic concerns (Thomas and Foster
1975; Taylor, Scheppele, and Stinchcombe 1979; Stinchcombe et al. 1980;
Tyler and Weber 1982; Tyler and Boeckmann 1997; Garland 2000). This
perspective posits that persons exposed to high or rising rates of violent
crime are likely to support extreme forms of social control, such as the
death penalty, for the practical reason that such measures may deter
violence. As Ellsworth and Gross note, “This is a commonsensical expla-
nation: when crime goes up, people look for harsher punishments to bring
it back down” (1994, p. 40). A second perspective links violent crime rates
and support for capital punishment via direct and indirect socialization
experiences. Gelles and Straus (1975, p. 609) suggest that persons exposed
to higher levels of violence are more likely than others to support the
death penalty because they are socialized to accept the normalcy of vi-
olence and to regard violence as an acceptable and effective form of
punishment and social control (see also Borg 1998).
Despite speculation to the contrary (e.g., Stinchcombe et al. 1980; Ells-
worth and Gross 1994; Garland 2000; Jacobs and Carmichael 2002), other
than the widely observed correspondence in temporal patterns of homicide
rates and death penalty support at the national level, systematic evidence
to substantiate an empirical association between these factors is scant
(Beckett 1997). An important exception is the work by Rankin (1979).
Using data from the 1972–76 General Social Surveys, Rankin estimated
the effects of national crime rates on support for the death penalty. Con-
trolling for race, region of residence, and year of interview, he observed
a significant positive association between death penalty support and a
three-year lagged measure of violent crime rates.4
Rankin’s findings provide suggestive empirical support for a positive
effect of homicide rates on support for the death penalty. However, the
limited number of control variables included in this study raises the con-
cern that the effects reported for national homicide rates may be spurious.
On the other hand, Rankin’s (1979) focus on national-level homicide rates
may have resulted in a deflated estimate of the impact of homicide on

4
The magnitude and functional form of this association varied across regions, with a
relatively strong linear effect observed in the South and a weaker nonlinear effect
found for all other regions. For respondents who reside in nonsouthern regions, Rankin
(1979) observed significant effects for both a linear and quadratic form of the national
homicide rate. The nature of these effects suggests that although higher homicide rates
are generally associated with increasing support for capital punishment, this association
dampens as homicide rates reach very high levels.

847

This content downloaded from 128.252.067.066 on July 25, 2016 21:40:49 PM


All use subject to University of Chicago Press Terms and Conditions (http://www.journals.uchicago.edu/t-and-c).
American Journal of Sociology

death penalty attitudes. Individual attitudes about capital punishment


may be more responsive to the level of homicide in the local environment
than to the national homicide rate.
Despite Rankin’s suggestive evidence consistent with the instrumental/
pragmatic perspective and its intuitive plausibility, the sociological lit-
erature on social problems provides grounds for skepticism about the
extent to which objective levels of criminal violence will adequately ac-
count for variation in public opinion on crime, including support for the
death penalty (Blumer 1971; Spector and Kitsuse 1977). In particular,
studies informed by the constructionist approach have demonstrated the
complexity of the processes through which social meanings are attached
to objective conditions (e.g., Schneider 1985; Miller and Holstein 1993;
Beckett 1994). As Beckett (1994) observes, conditions become social prob-
lems to the extent that they have been defined as such, and this is ulti-
mately the result of “claims-making activities.” Beckett further maintains
that “efforts to signify social problems are typically components of larger
political battles” (1997, p. 6). Political elites try to frame issues in ways
that serve their interests, often by directing attention away from “incon-
venient social conditions” that might challenge the status quo (p. 6).
Beckett and Sasson’s (2000) assessment of the growth in popular sup-
port for punitive crime-control policies in the United States since the 1960s
is a good example of research in the constructionist tradition. Although
levels of serious criminal violence in the United States are high by com-
parative standards, Beckett and Sasson argue that they have not risen
enough over the last 30 years to account for the growing harshness of
public attitudes toward crime and criminals. Rather, changes in popular
opinion on crime and punishment “reflect the ascendance of a particular
way of framing the crime problem” (Beckett and Sasson 2000, p. 73).
Conservative politicians, from Barry Goldwater to Richard Nixon and
Ronald Reagan, exploited public concerns about crime and racial divisions
with a “law and order” rhetoric that framed criminal behavior as a matter
of individual moral choice encouraged by the permissive crime and wel-
fare policies promoted by the liberals. The goal was to effect an electoral
realignment, especially in the South, that would favor the Republicans
by driving a wedge between working-class whites and blacks who had
composed the traditional Democratic base. The law and order rhetoric,
say Beckett and Sasson, was from the beginning a thinly veiled appeal
to racial prejudice. There was never any question regarding the presumed
racial identification of most of the violent criminals or welfare cheats the
conservatives had in mind. The strategy worked. Public concern with
crime and support for punitive crime-control policies grew within an

848

This content downloaded from 128.252.067.066 on July 25, 2016 21:40:49 PM


All use subject to University of Chicago Press Terms and Conditions (http://www.journals.uchicago.edu/t-and-c).
Support for Capital Punishment

increasingly conservative political climate that affected the views of lib-


erals and conservatives alike (see also Garland 1990).5
The constructionist perspective on public opinion, as explicated by
Beckett and others, suggests that death penalty support will not be a
simple function of objective levels of criminal violence. Rather, it should
vary along with the political climate and the corresponding claims-making
activities of political officials. Our data do not permit the direct analysis
of claims-making activity, but we can measure the degree of conservatism
of the political climate. Drawing on the constructionist perspective, we
anticipate a positive effect of political conservatism on death penalty
support, net of official homicide rates. Respondents residing in areas with
a highly conservative climate should be more likely to support the death
penalty than those in less conservative environments. Note that we are
hypothesizing a genuine contextual effect of conservatism. The degree of
political conservatism characteristic of the area should increase the like-
lihood of death penalty support controlling for individual ideological
position.
Our final hypotheses about contextual effects are also predicated on
the basic premise of constructionism that social problems must be actively
problematized and that elites play a prominent role in the process (e.g.,
Schneider 1985). These hypotheses elaborate the traditional construction-
ist approach by incorporating insights from conflict theory concerning the
structural conditions that are likely to be conducive to the mobilization
of public opinion on crime (Scheingold 1991; Garland 2000). A long-
standing tradition of research informed by conflict theory has identified
racial and economic divisions as conditions that threaten the rule of dom-
inant groups and lead to more extensive and repressive forms of social
control (e.g., Hawkins 1987; Liska 1992). The basic logic of the threat
hypotheses is that whites and economic elites perceive nonwhites and
poor people as threatening (Blumer 1958; Chambliss 1964; Blalock 1967;
Chambliss and Seidman 1980; Quillian 1995, 1996). When these latter
groups increase in relative size, the level of threat increases, and dominant
groups exert pressure for greater crime control to protect the status quo.
A number of studies have assessed the threat hypotheses by examining
the effects of racial composition (typically %black) and economic ine-
quality on police size. The results of this research are generally supportive
of the hypothesis of racial threat, whereas support for the economic threat

5
Beckett and Sasson argue that public opinion on crime and punishment has always
been more complex than the law and order discourse of conservative politicians. For
example, as support increased for harsher criminal sanctions, substantial fractions of
the electorate also remained committed to rehabilitative prison programs, drug treat-
ment, and crime prevention initiatives (2000, pp. 136–42).

849

This content downloaded from 128.252.067.066 on July 25, 2016 21:40:49 PM


All use subject to University of Chicago Press Terms and Conditions (http://www.journals.uchicago.edu/t-and-c).
American Journal of Sociology

hypothesis is more mixed (Jacobs 1979; Jackson and Carroll 1981; Liska,
Lawrence, and Benson 1981; Chamlin 1989; Jackson 1989). Research on
perhaps the most repressive form of legal control—killings by police—has
also offered some support for the hypothesis of racial threat. Jacobs and
O’Brien (1998) examined the structural correlates of total police killings
and police killings of blacks for a sample of 170 cities. They found that,
although the percentage of the population that is black has no significant
association with total police killings, it is positively related to police kill-
ings of blacks.6
The threat hypotheses have generally been applied to the explanation
of the actual exercise of social control (e.g., police size, arrests, police
exercise of lethal violence). We extend the basic argument to public at-
titudes about crime and punishment. Consistent with the work of Beckett
and Sasson reviewed above, we propose that elites mobilize public opinion
in the direction of more punitive attitudes to further their interests, and
they are particularly likely to do so under threatening conditions. Ac-
cordingly, the relative size of the minority population and economic in-
equality should be positively related to death penalty support, net of the
officially recorded homicide rate. A recent study by Jacobs and Carmichael
(2002) that examined the presence of the death penalty in American states
provides evidence consistent with these hypotheses. They found that states
with relatively large minority populations and high levels of economic
inequality were more likely than others to have legalized the death penalty.
Also, in line with the constructionist hypothesis on the effects of political
climate, their analyses revealed that measures of Republican legislative
strength and conservative political climate are associated with death pen-
alty legalization. Interestingly, they found no significant effects on the
presence of capital punishment for violent crime rates or murder rates.
To summarize, the prior literature suggests four hypotheses about con-
textual determinants of death penalty support. Persons residing in areas
with high homicide rates, a strongly conservative political climate, a rel-
atively large minority population, and high income inequality should be
more likely to express support for the death penalty. These effects should
emerge net of individual attributes that have been linked with attitudes
toward capital punishment, indicating that spatial variation in support
for the death penalty is not simply a function of population distribution.
They also should persist after holding constant other contextual variables

6
Jacobs and O’Brien also detected significant positive effects of racial inequality on
both total police killings and killings of blacks. Their measure of income inequality
was not significant for either dependent variable. They speculated that the use of a
city sample might attenuate the impact of economic inequality because the affluent
suburbs are excluded from the analysis (1998, p. 857).

850

This content downloaded from 128.252.067.066 on July 25, 2016 21:40:49 PM


All use subject to University of Chicago Press Terms and Conditions (http://www.journals.uchicago.edu/t-and-c).
Support for Capital Punishment

that may be associated with these conditions and with support for the
death penalty.

DATA AND METHODS


We estimate spatial variation in support for the death penalty with mul-
tilevel models applied to data from the General Social Survey (GSS) that
have been linked with homicide data from the National Center for Health
Statistics (NCHS) and socioeconomic data from the U.S. Census Bureau.

Data
The GSS is a cross-sectional probability sample survey of adults in the
United States conducted annually or biannually since 1972.7 The GSS has
been a valuable source of data for individual-level analyses of public
opinion (e.g., Firebaugh and Davis 1988; Ellison and Musick 1993;
DiMaggio, Evans, and Bryson 1996; Quillian 1996; Taylor 1998), and it
has been used extensively to examine the effects of individual attributes
on death penalty attitudes (e.g., Bohm 1991; Fox et al. 1991; Barkan and
Cohn 1994; Longmire 1996; Borg 1998). A distinctive feature of the GSS
data used for our research is that they contain geographic codes that
enable us to append data that describe the geographic areas in which
respondents reside (see also Taylor 1998; Rosenfeld, Messner, and Baumer
2001). These geographic areas are the primary sampling units (PSUs) used
to select the GSS national samples. Roughly two-thirds of the PSUs are
single- or multiple-county metropolitan areas, and the remaining third
are nonmetropolitan counties. Samples selected within these PSUs are
self-representing in the sense that aggregated individual responses are
representative of the PSU from which they are drawn, and within any
given survey year, the combined samples across PSUs compose a sample
of households that is representative of the continental United States (Davis
and Smith 1998). The sample upon which our analyses are based includes
32,632 respondents sampled from 268 PSUs between 1974 and 1998.8
7
The full samples for 1972–74, and half-samples for 1975 and 1976, were selected
using a modified probability design. A stratified, multistage area probability sample
design was implemented for half-samples in 1975–76 and has been used for the selection
of all respondents interviewed between 1977 and 1998 (for a detailed discussion of
GSS sampling methods, see Davis and Smith [1998, app. 1]).
8
We exclude the 1972 GSS data from our analysis because PSU identifiers are not
available for these respondents. Data for 1973 are excluded because of a difference in
the wording of the death penalty question compared to the one used in all subsequent
years. Overall, 34,136 persons were interviewed in the GSS between 1974 and 1998;
our sample excludes 1,434 of these cases due to missing data on the dependent and
independent variables.

851

This content downloaded from 128.252.067.066 on July 25, 2016 21:40:49 PM


All use subject to University of Chicago Press Terms and Conditions (http://www.journals.uchicago.edu/t-and-c).
American Journal of Sociology

To assess contextual determinants of death penalty support, we merged


data from the NCHS and the U.S. Census Bureau with the individual-
level GSS data files using the geographic identifiers described above. The
homicide data used in the analysis are from the U.S. vital statistics com-
piled from death certificates at the county level for 1973–98 (NCHS 1998).
We attached contemporaneous and lagged homicide counts to each GSS
respondent record using the census codes that correspond to the county
or counties that compose the PSU in which the respondent resided. This
process was a straightforward one-to-one merge in the case of single-
county PSUs. For multiple-county PSUs, we aggregated homicides across
all counties within PSUs and linked the aggregated homicide counts to
the individual GSS records by matching on PSU codes. The counts were
later converted to homicide rates, as described below.
We use data from the U.S. Census Bureau to construct measures of
two of our key explanatory variables—minority population size and ec-
onomic inequality—as well as some other contextual variables that we
include as control variables (discussed below). PSU-level indicators of
these socioeconomic conditions were derived using county-level census
data for 1970, 1980, and 1990 from the County and City Data Book
Consolidated File, 1947–77, and the USA Counties 1996 CD-ROM. Be-
cause county-level census data are not available on an annual basis, we
used the decennial data to estimate values for the intercensal years. We
then constructed PSU-level measures of these variables and merged them
with the GSS using the procedures described above.

Measures
The dependent variable for our analysis is a binary measure of respon-
dents’ attitudes toward the death penalty, coded “1” for those who favor
the death penalty for persons convicted of murder and coded “0” for those
who oppose the death penalty or who report that they do not know (favor
death penalty).9
The key independent variables in our research are the homicide rate,
the degree of conservatism in the local political climate, the relative size
of the minority population, and the level of economic inequality for the
PSUs in which GSS respondents reside. The homicide rate was created
using annual data on homicide from the NCHS and population estimates
from the U.S. Census Bureau, and it represents the number of homicides

9
We reestimated all regression models shown below after excluding respondents who
responded “don’t know” to the death penalty question (N p 1,990). The results of
these analyses were substantively identical to those reported.

852

This content downloaded from 128.252.067.066 on July 25, 2016 21:40:49 PM


All use subject to University of Chicago Press Terms and Conditions (http://www.journals.uchicago.edu/t-and-c).
Support for Capital Punishment

per 100,000 residents in the year preceding the interview (homicide rate).10
A lagged measure of homicide is preferable to a contemporaneous measure
because the latter might include homicides that occurred after the date
of the survey, and because it is likely to take some time for the public to
become aware of the crime rate.11
We measure the degree of political conservatism in respondents’ PSUs
by aggregating responses to a GSS question about political ideology. Re-
spondents are asked to report where they would place their political views
on a scale from one (extremely liberal) to seven (extremely conservative).
Our measure of conservative political climate reflects the mean level of
conservatism in respondents’ PSUs during the period in which the GSS
interview was conducted (conservative climate). Because single-year sam-
ple sizes for some PSUs are relatively small, we combine responses to the
conservatism question across three surveys, then aggregate responses
within PSUs and compute moving averages for each PSU centered on
the survey year.12 This strategy seems reasonable given that the political
climate for counties and metropolitan areas is likely to be stable in the
short run.
The relative size of the minority population in the respondents’ PSU
is measured with census data on the percentage of residents who identify
themselves as black (%black). To reduce skewness, we use the natural
log of this variable in the regression models. The level of economic ine-
quality for the PSUs in our analysis is measured with the Gini index for
the relative distribution of family incomes within these areas (income
inequality). Both measures are derived from census data as described
above and characterize the demographic and economic conditions of the
respondent’s PSU in the year preceding the interview.
To isolate the effects of these contextual predictors, we include well-
established correlates of death penalty attitudes as control variables in
our analysis. Prior research on death penalty attitudes reveals higher levels
of support among whites, older persons, males, wealthier individuals,
conservatives, religious fundamentalists, married persons, and those who

10
The annual county population estimates were obtained from the Inter-University
Consortium for Political and Social Research (ICPSR) data archive (see studies #8384,
#6031, and #2372).
11
Although a one-year lagged measure of homicide rate seems justified on theoretical
grounds, we also estimated all models shown below using two- and three-year lagged
measures of homicide. The results obtained from these analyses (not shown) were very
similar to those reported.
12
For years in which the GSS sampling frame changes (i.e., 1982 and 1992), we centered
the three-year moving average on the year prior to the year in which the respondent
was surveyed. For the later years of the GSS data used in our study (1994, 1996, and
1998), which contain much larger annual sample sizes than earlier periods, we combine
data for two survey years.

853

This content downloaded from 128.252.067.066 on July 25, 2016 21:40:49 PM


All use subject to University of Chicago Press Terms and Conditions (http://www.journals.uchicago.edu/t-and-c).
American Journal of Sociology

reside in less populated areas (for reviews, see Zeisel and Gallup [1989],
Bohm [1991], Fox et al. [1991], and Longmire 1996). Although more am-
biguous, there is some evidence that church attendance reduces support
for the death penalty (Harvey 1986) and that the effect of educational
attainment on support for the death penalty is nonlinear, with low levels
of support observed among persons who did not finish high school and
those who graduated from college, and higher levels of support for those
whose educational attainment falls between these points (Fox et al. 1991).
Some research has uncovered regional differences in support for the death
penalty. Persons who reside in the U.S. West and, to a lesser extent, the
U.S. South exhibit higher levels of support than persons from other regions
(Bohm 1991; Fox et al. 1991; Barkan and Cohn 1994; Borg 1997).13
We include each of these variables in our analysis as controls, and our
measurement of them conforms closely to strategies used in prior research
(see app. table A1 for a more detailed description of the control variables).14
In addition, we capture time trends in support for the death penalty by
including a variable that distinguishes the year in which respondents were
interviewed. To facilitate a more intuitive interpretation of estimated co-
efficients than would be the case if the actual year of interview (e.g., 1974,
1975) were used, we constructed a rescaled measure by subtracting 73
from the year of the interview (i.e., 1974 respondents received a code of
“1,” 1975 respondents received a code of “2,” and 1998 respondents re-
ceived a code of “25”). To model the well-documented nonlinear trend in
support for capital punishment during this period, we also include a

13
There is some evidence in the literature that fear of crime also has a significant,
albeit weak, positive effect on support for the death penalty (Rankin 1979; Tyler and
Weber 1982; Seltzer and McCormick 1987; Keil and Vito 1991; Longmire 1996). The
GSS includes an indicator of fear of crime (e.g., Warr 1995), but it is unavailable for
about one-third of the cases in our sample. To avoid dropping data for years in which
this indicator is unavailable, we do not include it in the analysis presented below.
However, we replicated our analysis on the sample of cases for which the fear item
is available. These analyses reveal that fear exerts a significant, positive effect on death
penalty support and that the effects of the other variables are virtually identical to
those shown below.
14
We used mean replacement to minimize the loss of cases due to missing data for
some of the control variables. For church attendance and education, we mean-replaced
less than 1% of all responses. About 7% of the cases were mean-replaced because of
missing data on the conservatism scale (N p 2,282). In all regression models reported
below in which these variables appear, we include dummy variables that identify the
cases in which mean replacement was used. To assess the sensitivity of our results to
this strategy, we also reestimated all regression models after excluding respondents for
whom information on these variables was missing; the results obtained from these
supplementary analyses were substantively identical to those reported.

854

This content downloaded from 128.252.067.066 on July 25, 2016 21:40:49 PM


All use subject to University of Chicago Press Terms and Conditions (http://www.journals.uchicago.edu/t-and-c).
Support for Capital Punishment

squared version of this variable.15 Finally, to minimize the possibility of


drawing misleading conclusions about the hypothesized contextual effects,
we conduct sensitivity analyses that incorporate three additional contex-
tual variables: resource deprivation,16 the unemployment rate, and the
male divorce rate. Although these variables have not been discussed in
the death penalty literature, they are often included in studies of crime
and punishment (e.g., Myers and Talarico 1986; Land et al. 1990; Myers
and Massey 1991; Greenberg and West 2001; Jacobs and Carmichael
2001). Moreover, they are significantly correlated with one or more of the
contextual predictors that form the focus of our analysis (see app. table
A2), and they may influence death penalty support (Garland 2000; Jacobs
and Carmichael 2002). Holding these factors constant increases confidence
in our assessment of the hypotheses evaluated in the study.

Analytic Strategy
We employ a multilevel modeling strategy to examine the effects of hom-
icide rates on support for the death penalty. Multilevel regression models
have become the standard method for estimating the effects of community
characteristics on individual attitudes and behaviors, especially when the
data used for such studies contain a substantial amount of respondent
clustering within communities (e.g., Raudenbush and Bryk 2002). Con-
ceptually, multilevel regression models provide a direct and efficient means
of describing the degree to which a given individual-level outcome, such
as support for the death penalty, varies across geographic areas. In ad-
dition, an important methodological benefit is that multilevel models for-
mally adjust for nonindependence of sample members who reside in the
same community. Failure to model this type of nonindependence can result
in estimated standard errors that are biased downward, which may in
turn produce misleading conclusions about the statistical and substantive
importance of community variables (DiPrete and Forristal 1994; Snijders
and Bosker 1999).
The two most general types of multilevel models used to estimate the
effects of community variables on a given individual-level outcome var-
iable are random-slope and random-intercept models. Random-slope mod-
els are particularly well-suited for examining cross-level interactions, such

15
We also conducted the analysis by modeling the time trend with dummy variables
for each interview year using 1974 as the reference year (analyses not shown). The
substantive conclusions drawn from these models were identical to those we describe
below.
16
Resource deprivation is a principal components index composed of the poverty rate,
median family income, and female-headed families (see app. table A1).

855

This content downloaded from 128.252.067.066 on July 25, 2016 21:40:49 PM


All use subject to University of Chicago Press Terms and Conditions (http://www.journals.uchicago.edu/t-and-c).
American Journal of Sociology

as whether the effect (i.e., the slope) of a specified individual-level ex-


planatory variable varies across communities. In contrast, the main pur-
pose of random-intercept models is to evaluate the degree to which the
mean value of a given dependent variable (i.e., the intercept) varies across
communities and to examine whether a specified set of explanatory var-
iables helps to account for that variation. Because our theoretical focus
in the present research is on the extent to which our explanatory variables
help to explain variation in death penalty support across geographic areas,
we estimate random-intercept models in which all slope parameters are
treated as fixed across the geographic areas represented in our data. Spe-
cifically, given the binary coding of our dependent variable, we estimate
a series of two-level hierarchical logistic random-intercept regression mod-
els (for detailed descriptions of these models, see Wong and Mason [1985],
Patterson [1991], Guo and Zhao [2000], and Raudenbush and Bryk
[2002]).17
Our analysis proceeds in the following manner. We begin by estimating
multilevel regression models that describe the extent of variation in sup-
port for the death penalty across GSS PSUs and evaluate the degree to
which that variation is due to compositional differences. We then assess
whether our contextual predictors affect support for the death penalty,
net of other factors, and whether they account for spatial variation in
levels of support.

RESULTS
Table 1 presents descriptive statistics for all of the variables included in
the analysis. Although support for the death penalty has fluctuated sub-
stantially between 1974 and 1998, overall, 70% of GSS respondents in-
terviewed during this period reported that they favor the death penalty
for persons convicted of murder. On average, GSS respondents reside in
geographic areas (i.e., metropolitan areas and counties) in which there are
about 9 homicides annually per 100,000 residents and in which there is
a moderately conservative political climate, a population composition of
slightly more than 11% black (before logging), and a distribution of family
incomes that is moderately unequal. Table 1 also demonstrates that there
is variation in the social contexts to which respondents are exposed; our
analysis explores whether this translates into geographic differences in
support for the death penalty.

17
All models presented are estimated with HLM for Windows 4.01 (Bryk, Raudenbush,
and Congdon 1996). The results shown are from unit-specific models (see Raudenbush
and Bryk 2002).

856

This content downloaded from 128.252.067.066 on July 25, 2016 21:40:49 PM


All use subject to University of Chicago Press Terms and Conditions (http://www.journals.uchicago.edu/t-and-c).
Support for Capital Punishment

TABLE 1
Descriptive Statistics for Variables Included in Analysis
of Contextual Effects on Support for the Death Penalty

Mean SD

Dependent variable:
Favor death penalty . . . . . . . . . . . . .70 .45
Contextual variables:
Homicide rate . . . . . . . . . . . . . . . . . . . 8.73 6.96
Conservative climate . . . . . . . . . . . . 4.13 .28
%black (logged) . . . . . . . . . . . . . . . . . . 1.66 1.61
Income inequality . . . . . . . . . . . . . . . .37 .03
Control variables:
Individual-level:
Time . . . . . . . . . . . . . . . . . . . . . . . . . . . 13.47 7.46
Time2 . . . . . . . . . . . . . . . . . . . . . . . . . . . 237.03 201.84
White . . . . . . . . . . . . . . . . . . . . . . . . . . .85 .35
Male . . . . . . . . . . . . . . . . . . . . . . . . . . . .44 .50
Age . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 45.30 17.60
Bachelor’s degree or more . . . .18 .39
Junior college degree . . . . . . . . . .05 .21
High school degree . . . . . . . . . . . .52 .50
No high school degree . . . . . . . .25 .43
Family income . . . . . . . . . . . . . . . . 9.73 2.85
Married . . . . . . . . . . . . . . . . . . . . . . . . .56 .50
Conservatism . . . . . . . . . . . . . . . . . . 4.11 1.28
Religious fundamentalism . . . .32 .47
Church attendance . . . . . . . . . . . 2.63 1.17
Place size (logged) . . . . . . . . . . . . 3.44 2.11
South . . . . . . . . . . . . . . . . . . . . . . . . . . . .35 .48
West . . . . . . . . . . . . . . . . . . . . . . . . . . . . .18 .38
PSU-level:
Resource deprivation . . . . . . . . . -.01 .99
Unemployment rate . . . . . . . . . . 5.97 2.02
Male divorce rate . . . . . . . . . . . . . 5.93 2.35
Note.—Descriptive statistics for the dependent and individual-level control
variables are based on 32,632 cases, and descriptives for the contextual variables
and PSU-level control variables are based on the 268 PSUs within which these
respondents reside.

Geographic Variation in Death Penalty Support


We begin by evaluating the degree to which support for the death penalty
actually varies across geographic areas sampled in the GSS. We do so
formally in model 1 of table 2, which presents results from a regression
equation that includes an intercept parameter that describes the mean
log odds of death penalty support for GSS respondents and a variance
component that describes whether there is significant variation in support
across the geographic communities represented by the GSS. The estimated

857

This content downloaded from 128.252.067.066 on July 25, 2016 21:40:49 PM


All use subject to University of Chicago Press Terms and Conditions (http://www.journals.uchicago.edu/t-and-c).
TABLE 2
Hierarchical Logistic Regressions of Death Penalty Support

858
1 2 3 4
Fixed effects:
Intercept, g00 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .901* (.026) ⫺1.590* (.103) ⫺2.220* (.465) ⫺3.010* (.643)
Time . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .090* (.009) .090* (.009) .088* (.009)
Time2 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . ⫺.003* (.0003) ⫺.003* (.0003) ⫺.003* (.0003)
White . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .975* (.038) .989* (.039) .987* (.039)
Male . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .497* (.027) .498* (.027) .498* (.027)
Age . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .0003 (.001) .0004 (.001) .0004 (.001)
No high school degree . . . . . . . . . . . . . . . . . . ⫺.235* (.034) ⫺.236* (.034) ⫺.234* (.034)
Bachelor’s degree or more . . . . . . . . . . . . . . ⫺.516* (.035) ⫺.520* (.035) ⫺.521* (.035)
Family income . . . . . . . . . . . . . . . . . . . . . . . . . . . .044* (.005) .042* (.005) .042* (.005)
Married . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .157* (.028) .158* (.028) .158* (.029)
Conservatism . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .215* (.010) .214* (.010) .214* (.010)

This content downloaded from 128.252.067.066 on July 25, 2016 21:40:49 PM


Religious fundamentalism . . . . . . . . . . . . . . .034 (.031) .026 (.031) .029 (.031)

All use subject to University of Chicago Press Terms and Conditions (http://www.journals.uchicago.edu/t-and-c).
Church attendance . . . . . . . . . . . . . . . . . . . . . . ⫺.105* (.012) ⫺.104* (.012) ⫺.103* (.012)
Place size . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . ⫺.027* (.008) ⫺.035* (.008) ⫺.038* (.008)
South . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .107* (.051) ⫺.012 (.061) .005 (.068)
West . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .241* (.067) .288* (.067) .242* (.074)
Homicide rate . . . . . . . . . . . . . . . . . . . . . . . . . . . . .012* (.004) .012* (.004)
Conservative climate . . . . . . . . . . . . . . . . . . . . .246* (.087) .244* (.090)
%black . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .054* (.019) .042* (.019)
Income inequality . . . . . . . . . . . . . . . . . . . . . . . . ⫺1.37 (1.03) .958 (1.57)
Resource deprivation . . . . . . . . . . . . . . . . . . . . ⫺.076† (.046)
Unemployment rate . . . . . . . . . . . . . . . . . . . . . ⫺.005 (.014)
Male divorce rate . . . . . . . . . . . . . . . . . . . . . . . . .010 (.016)
Random effects:
Intercept, t00:
j ........................................ .129 .094 .078 .077
x2 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1,117* 837* 731* 717*

859
Note.—N p 32,632. SEs are given in parentheses.
* P ! .05, two-tailed test.

P ! .05, one-tailed test.

This content downloaded from 128.252.067.066 on July 25, 2016 21:40:49 PM


All use subject to University of Chicago Press Terms and Conditions (http://www.journals.uchicago.edu/t-and-c).
American Journal of Sociology

intercept corresponds (within rounding) to the mean level of support for


the death penalty across all geographic areas from which GSS respondents
were sampled (.71 p exp (.901)/1 ⫹ exp (.901)). More important for the
purposes of the present research, the random effects variance parameter
and test statistic shown in model 1 (t00 p .129; x 2 p 1117) indicate that
there is significant variation in support for the death penalty between
these geographic areas. Figure 1 displays a histogram that summarizes
the degree of variation in death penalty support across geographic areas
represented in the GSS. This figure demonstrates that levels of support
for the death penalty vary from less than 50% in some areas to greater
than 90% in others. Thus, although the United States is often described
as a nation that exhibits very high levels of support for capital punishment
(e.g., Hood 1998), that characterization ignores the substantial internal
heterogeneity in attitudes toward the death penalty that exists in the
United States. Indeed, public sentiment about the death penalty is decid-
edly mixed in many of the geographic areas sampled in the GSS, and in
some areas, a majority of persons disapprove of the death penalty for
persons convicted of murder.
Figure 1 prompts the question as to what factors might account for
differences in levels of support for capital punishment across space. One
possibility is that these differences merely reflect compositional differences
in populations. In other words, it is possible that geographic areas in
which levels of support for the death penalty are higher simply contain
more individuals with attributes that are associated with support for the
death penalty. Alternatively, rates of support for the death penalty may
be higher in some geographic areas because residents of those areas are
exposed to higher rates of homicide or because of a more conservative
political climate, a larger percentage of black residents, or higher levels
of economic inequality. Models 2–4 of table 2 assess the validity of these
possible explanations.

Compositional Effects
Model 2 includes individual-level attributes shown in prior research to
be related to death penalty support. The fixed-effects portion of model 3
shows that whites, males, married persons, individuals with more con-
servative political views, and those with higher family incomes are sig-
nificantly more likely to favor the death penalty. Individuals with rela-
tively high or low levels of education, those who attend church more
regularly, and those who reside in places with larger populations are
significantly less likely to favor the death penalty. Net of these factors,
respondents who reside in the southern or western region of the United
States are more likely than those from other regions to favor capital

860

This content downloaded from 128.252.067.066 on July 25, 2016 21:40:49 PM


All use subject to University of Chicago Press Terms and Conditions (http://www.journals.uchicago.edu/t-and-c).
Fig. 1.—Distribution of levels of support for the death penalty for 268 GSS primary sampling units, 1974–98

This content downloaded from 128.252.067.066 on July 25, 2016 21:40:49 PM


All use subject to University of Chicago Press Terms and Conditions (http://www.journals.uchicago.edu/t-and-c).
American Journal of Sociology

punishment. These results are highly consistent with an extensive body


of prior research on death penalty attitudes (for reviews, see Bohm [1991]
and Longmire [1996]).
The control variables explain a nontrivial portion of the variance in
death penalty support across communities. A comparison of the variance
components for models 1 and 2 indicates that more than one-quarter of
the variation across PSUs in levels of support for the death penalty is
accounted for by these individual-level attributes (.271 p .129 ⫺
.094/.129). Further analyses (not shown) revealed that about half of the
overall spatial variance in death penalty support explained by the control
variables is due to the time trend variables, while the other half is ac-
counted for primarily by respondent race and political conservatism.
Thus, part of the explanation for why some geographic areas exhibit
higher levels of support for the death penalty is that respondents were
sampled from these areas during years in which levels of support were
higher overall and that these areas contain more whites and persons who
self-identify as being politically conservative. Nonetheless, the variance
component statistics in model 2 indicate that a significant amount of
between-community variation in support for the death penalty remains
even after controlling for these indicators. This suggests that other factors
also contribute to differences across communities in levels of support for
the death penalty.

Contextual Effects on Support for the Death Penalty


As documented above, a variety of contextual factors have been identified
in the literature as potential predictors of geographic variation in support
for the death penalty. Instrumental/pragmatic theories and socialization
theories emphasize direct and indirect exposure to high levels of violence;
constructionist perspectives on public opinion emphasize, among other
things, the degree of conservatism in the local political climate; and threat
perspectives highlight the relative size of minority groups and the con-
centration of economic resources. Models 3 and 4 of table 2 present re-
gression results that evaluate these hypotheses. Model 3 includes all of
the individual-level control variables, plus the four contextual explanatory
variables that are of primary interest in our research. In model 4, we
assess the robustness of our findings by introducing controls for additional
contextual variables that may be associated with death penalty attitudes.18

18
Some of the contextual variables examined exhibit relatively strong correlations (see
app. table A1), and therefore, multicollinearity is a potential concern. We inspected
variance inflation factors (VIFs) for each of the contextual variables, and none was
over 4 (the highest VIF was 3.8), indicating that multicollinearity is not a major problem

862

This content downloaded from 128.252.067.066 on July 25, 2016 21:40:49 PM


All use subject to University of Chicago Press Terms and Conditions (http://www.journals.uchicago.edu/t-and-c).
Support for Capital Punishment

Model 3 indicates that three of the four contextual variables exhibit


the expected effects. The estimated effect of the level of homicide is in
the expected direction and is statistically significant. Consistent with in-
strumental/pragmatic theories, holding other factors constant, persons
who live in areas with higher homicide rates are significantly more likely
than others to favor the death penalty.19 We also find support for our
hypothesis about the contextual effect of political conservatism. Several
prior studies have shown that persons who identify with a conservative
political orientation are more likely to support the death penalty. Beyond
this, our results indicate that regardless of one’s own political views,
residing in a more politically conservative area increases the likelihood
that individuals support the death penalty. Model 3 reveals mixed evi-
dence for the threat hypotheses examined in our research. The coefficient
for %black is in the expected direction and is statistically significant. Note
that the contextual effect of %black is opposite of the individual race
effect. Although blacks are significantly less likely to support the death
penalty, persons who reside in areas with a higher percentage of blacks
are significantly more likely to favor the death penalty. In contrast, the
coefficient for income inequality is negative and is not statistically
significant.20
Although not the main focus of our analysis, two other noteworthy
findings emerge in model 3. First, comparing the coefficients for the time
trend variables in models 2 and 3 indicates that temporal changes in
support for the death penalty are not due to changes over time in the
control variables or the contextual predictors. The inability of changes in
factors such as marital status, political conservatism, education, and hom-
icide rates to explain temporal trends in support for capital punishment
challenges much speculation in the research literature (e.g., Ellsworth and
Gross 1994; Niemi et al. 1989; Page and Shapiro 1992) and the popular
press (Shapiro 2000). Second, controlling for the contextual predictors

in our models. Evaluation of the condition indices, however, reveals some evidence of
multicollinearity between the indicators of income inequality and resource deprivation
included in model 4. Nevertheless, this does not alter the substantive conclusions drawn
about these two variables (e.g., neither is statistically significant when the other is
removed) and, more important, the coefficients for the other contextual predictors
exhibit stability across specifications that include or exclude these variables.
19
In subsequent analyses, we considered the possibility that short-term change in local
homicide rates, rather than the level of homicide, might affect death penalty attitudes.
We constructed a variety of change measures (e.g., absolute change, %change) ranging
from one to five years. None of these change measures yielded a significant effect on
support for the death penalty in our multivariate models (results not shown).
20
We evaluated a variety of different functional forms for these variables (e.g., linear
splines defined at various points of their distributions, a quadratic transformation),
but none yielded a significant coefficient or an improvement in model fit.

863

This content downloaded from 128.252.067.066 on July 25, 2016 21:40:49 PM


All use subject to University of Chicago Press Terms and Conditions (http://www.journals.uchicago.edu/t-and-c).
American Journal of Sociology

explains fully the effect on death penalty support of residence in the South.
Further analyses (not shown) reveal that higher levels of support for the
death penalty observed in the South are a function of regional differences
in racial composition and homicide rates.21
The most important findings shown in model 3 reveal that the homicide
rate, a conservative political climate, and the relative size of the black
population are significant predictors of death penalty support. In model
4 of table 2, we assess whether these conclusions hold up under a different
model specification. In particular, we evaluate whether the effects ob-
served for these variables are confounded with other community char-
acteristics, including levels of resource deprivation, unemployment rates,
and divorce rates. Although past studies of public opinion on the death
penalty have not considered these contextual variables, macrolevel studies
have demonstrated that each is at least moderately correlated with hom-
icide rates, %black, and income inequality (e.g., Land et al. 1990). More-
over, areas characterized by a more conservative political climate tend to
have higher levels of resource deprivation and higher male divorce rates
(see app. table A2). Thus, it is potentially important to hold these variables
constant in evaluating the effects on death penalty support observed for
homicide, conservative climate, and %black.
Model 4 shows that the unemployment rate and the male divorce rate
do not exert significant effects on support for the death penalty. The
coefficient for resource deprivation, in contrast, is negative and statistically
significant using a one-tailed test (P ≤ .05). Controlling for several indi-
vidual-level attributes and other contextual conditions, persons who reside
in communities with lower income levels and a larger percentage of fam-
ilies headed by females are significantly less likely to favor the death
penalty. Most important for the purposes of our analysis, even after con-
trolling for these factors, we find significant effects of homicide, conser-
vative climate, and %black on death penalty support. Overall, the con-
textual predictors considered account for 18% (.18 p .094 ⫺ .077/.094) of
the spatial variance in death penalty support not accounted for by the
control variables, and, collectively, the variables included in the analysis

21
We reestimated model 3 of table 2 with each of the contextual predictors entered
separately and in various combinations. The coefficient for southern regional location
was not attenuated when conservative climate and income inequality were controlled.
In contrast, it was reduced substantially and was no longer statistically significant
when either %black or the homicide rate was considered.

864

This content downloaded from 128.252.067.066 on July 25, 2016 21:40:49 PM


All use subject to University of Chicago Press Terms and Conditions (http://www.journals.uchicago.edu/t-and-c).
Support for Capital Punishment

explain 40% (.40 p .129 ⫺ .077/.129) of the total spatial variance in death
penalty support.22
In general, the individual-level attributes exert the strongest effects on
support for the death penalty, but the magnitude of the effects of homicide,
conservative climate, and %black are nontrivial. Table 3 presents pre-
dicted probabilities of death penalty support for respondents who reside
in communities that differ substantially along these dimensions. In each
case, the predicted probabilities were computed using the coefficients from
model 4 of table 2 and assuming mean values for all other variables (see,
e.g., Hosmer and Lemeshow 2000). The predicted probabilities associated
with the estimated homicide effect imply that levels of support for capital
punishment range from about 70% in areas with very low homicide rates
(the 5th percentile) to about 75% in areas with very high homicide rates
(the 95th percentile), assuming mean values for all other variables (in-
cluding the other contextual predictors). Conservative climate and %black
exhibit effects of similar magnitude.
Although modest in magnitude, it is important to note that the effects
of these contextual variables are additive and, because some of them
covary significantly, it is informative to consider their combined effects.23
For instance, among those who reside in predominantly white commu-
nities with very low homicide rates, the predicted probability of death
penalty support is 0.67, while the comparable figure for persons who reside
in predominantly black areas with very high homicide rates is 0.77. This
difference represents more than half of the range of observed temporal
change in support for the death penalty across the years included in our
study.

22
We also considered the effects on death penalty attitudes of the legal status of the
death penalty (i.e., legal vs. illegal) and the number of executions carried out in the
state in which GSS respondents reside. The state of residence for GSS respondents is
not available, but can be inferred from the PSU geographic identifier for most re-
spondents. For persons who live in PSUs that cross state boundaries, we created
measures for all possible states in which they could have resided. The various measures
yielded similar effects: persons who reside in states in which the death penalty is legal
or in which more persons were executed by the state (in the year prior to the GSS
interview) are significantly more likely to support the death penalty, net of the indi-
vidual and other contextual variables. The causal order in these complex relationships
is uncertain: the relationships could reflect the influence of state policy and action on
public support for the death penalty, or they could reflect the responsiveness of the
state to public opinion. A full assessment of this issue is beyond the scope of the present
research. Nevertheless, the fact that the effects of homicide, conservative political
climate, and racial composition persist after controlling for these variables bolsters our
conclusions about their role in shaping public opinion about the death penalty.
23
We tested for the possibility that the effects of the contextual independent variables
were multiplicative. Of the numerous two- and three-way interactions considered, none
was statistically significant.

865

This content downloaded from 128.252.067.066 on July 25, 2016 21:40:49 PM


All use subject to University of Chicago Press Terms and Conditions (http://www.journals.uchicago.edu/t-and-c).
American Journal of Sociology

TABLE 3
Predicted Probabilities of Death Penalty Support in Different Social
Contexts

Percentile on Contextual Variables

Contextual variables 5 10 25 50 75 90 95

Homicide rate . . . . . . . . . . . .702 .706 .711 .720 .732 .745 .750


Conservative climate . . . .697 .706 .714 .723 .734 .741 .745
%black . . . . . . . . . . . . . . . . . . . .695 .703 .717 .727 .734 .737 .739

SUMMARY AND DISCUSSION


The sociological literature acknowledges that public opinion about social
issues, including capital punishment, may be shaped by various features
of the social context. However, despite a large body of research on the
individual-level attributes that contribute to the development of punitive
attitudes toward punishment in general and support for capital punish-
ment more specifically, few studies have considered whether community
contextual characteristics shape these attitudes. Using individual-level
data from the GSS that have been linked with macrolevel data on the
social and political contexts in which respondents of the survey reside,
our analysis examined contextual effects on support for the death penalty.
Drawing on instrumental and socialization theories of punitiveness, as
well as insights from social constructionist and conflict perspectives, we
hypothesized that individuals who reside in areas with a higher rate of
homicide, a more conservative political climate, a relatively larger per-
centage of blacks, and a higher level of income inequality will be more
likely than others to favor the death penalty for persons convicted of
murder. We expected these effects to persist net of individual-level at-
tributes and other contextual factors that may be associated with support
for the death penalty.
The possibility of geographic variation in attitudes about capital pun-
ishment had been hinted at in past research, but never evaluated syste-
matically. Our analysis revealed significant community-level variation in
support for the death penalty. This challenges conventional wisdom and
popular portrayals that support for capital punishment in the United
States is universally high (e.g., Singh 2000; Hood 2001). Instead, there is
substantial heterogeneity in attitudes about the death penalty within the
United States, with some areas exhibiting very high levels of support and
others more modest levels of support, including areas in which a majority
of residents disapprove of the death penalty.
We found that differences in population composition account for a
considerable portion of the geographic variation in public opinion on the

866

This content downloaded from 128.252.067.066 on July 25, 2016 21:40:49 PM


All use subject to University of Chicago Press Terms and Conditions (http://www.journals.uchicago.edu/t-and-c).
Support for Capital Punishment

death penalty, but we also found support for the hypothesized effects of
the homicide rate, political climate, and racial composition. Consistent
with expectations, higher levels of homicide in the local area increase the
likelihood that individuals will favor capital punishment, net of a wide
array of individual-level predictors and other contextual factors. The the-
oretical literature on attitudes about crime and punishment suggests that
exposure to high homicide rates may increase support for the death penalty
by motivating pragmatic or instrumental crime-control responses or
through socialization processes whereby residents of high violence areas
come to consider violence, including violence carried out by the state, as
an acceptable and effective form of punishment.24
Although plausible, there are other possible interpretations of the as-
sociation between homicide rates and support for capital punishment. For
instance, Garland (2000) suggests that high crime rates have weakened
the effectiveness of some informal social control mechanisms (see also
Anderson 1999; Sampson and Raudenbush 1999), which in turn has cre-
ated an insecure and precarious social environment in which support for
punitive crime control policies has flourished. Somewhat similarly, Black
(1976) has argued that where informal social controls are weak, popu-
lations tend to be more open to formal measures of social control. Thus,
it is conceivable that high homicide rates increase support for capital
punishment (and presumably other measures of formal social control) by
increasing social disorganization and reducing confidence in informal so-
cial controls.
Whatever the mechanisms are that link homicide to support for capital
punishment, our findings are inconsistent with the extreme constructionist
view that objective conditions are irrelevant to the formation of public
opinion about the death penalty. The framing of the crime issue in local
areas is likely to be highly influential in the formation of attitudes about
the death penalty, but objective levels of homicide also are important. We
suspect that objective levels of crime and the rhetoric and imagery used
by elites and the media to frame the crime issue coalesce in shaping public
opinion about capital punishment (see also Garland 2000).25 Consistent

24
For a subset of the years included in our study, the GSS contains items that have
been used in prior research as indicators of “fear of crime” (see Warr 1995) and “support
for violence” (see Dixon and Lizotte 1987; Cao, Adams, and Jensen 1997). To evaluate
whether these factors mediate the homicide effect, we reestimated our regression models
based on the subset of cases for which these indicators were available (see also n. 12).
Although fear of crime and the indicators of support for violence exert significant
positive effects on death penalty support, they do not mediate the effects of any of the
contextual predictors.
25
For a general discussion of the importance of both objective and subjective features
in the construction of social problems, see Best (1993).

867

This content downloaded from 128.252.067.066 on July 25, 2016 21:40:49 PM


All use subject to University of Chicago Press Terms and Conditions (http://www.journals.uchicago.edu/t-and-c).
American Journal of Sociology

with this assertion, Beckett suggests that “the increased incidence of crime-
related problems may facilitate their politicization and contribute to grow-
ing support for getting-tough” (1997, p. 15).
The significant contextual effect observed for our indicator of the po-
litical climate provides support for the constructionist argument that,
independent of levels of violence, the political context in which social
problems are framed has an important influence on public sentiment about
capital punishment. Although we do not have direct measures of the
claims-making process, the results are consistent with suggestions that
conservative politicians exploit concerns about crime in a manner that
effectively increases support for punitive policies, even among those who
do not hold a conservative ideology themselves. This could occur directly
through exposure to political initiatives or media coverage of crime and
punishment, or indirectly through collective socialization processes
whereby some elements of conservative ideology (e.g., support for capital
punishment), but not others, become contagious and spread throughout
the community.
The results reveal mixed support for the threat hypotheses examined.
Consistent with much of the literature on threat effects on the exercise
of various forms of social control (for a review, see Liska and Messner
[1999]), controlling for levels of homicide and a variety of other factors,
we found that %black exhibits a significant positive effect on support for
the death penalty, whereas the effect of income inequality is nonsignificant.
It is perhaps not surprising that racial composition affects attitudes about
capital punishment more than economic composition given the substantial
degree to which race and punishment are inextricably intertwined in the
United States (Hawkins 1997; Kennedy 1997; Blume, Eisenberg, and
Johnson 1998; Bowers, Steiner, and Sandys 2001; Jacobs and Carmichael
2002).
We cannot test directly the causal mechanisms that account for the
association between racial composition and support for the death penalty.
It could reflect differences in perceived threat, but it also could represent
some other macro- or microlevel process. For instance, Anderson and
others (Massey and Denton 1993; Kennedy 1997) have suggested that in
many predominantly black communities in the United States there is a
“sense of alienation from mainstream society and its institutions” (An-
derson 1999, p. 34), and as noted above, Garland (2000) argues that such
conditions may increase public support for punitive measures of social
control.
Overall, our analyses underscore the general importance of the larger
social context for explaining individual attitudes toward capital punish-
ment. The results suggest that a comprehensive understanding of public
opinion on capital punishment requires information both about the char-

868

This content downloaded from 128.252.067.066 on July 25, 2016 21:40:49 PM


All use subject to University of Chicago Press Terms and Conditions (http://www.journals.uchicago.edu/t-and-c).
Support for Capital Punishment

acteristics of individuals and the social environment in which they live.


Nonetheless, although our results have identified some of the factors that
help to account for spatial variation in support for capital punishment,
a substantial amount of variation remains unexplained by the variables
considered. An important avenue for further inquiry is to explore whether
other variables enhance our ability to explain that variation.
One possibility is that spatial differences in support for the death pen-
alty are better explained by disaggregated rather than total homicide rates.
For example, stranger, felony, and multiple-victim homicides might have
a particularly strong effect on attitudes toward the death penalty. It also
may be that interracial homicides have a stronger effect than intraracial
incidents on death penalty attitudes. These types of murders frame much
of the public debate about capital punishment, and they tend to dominate
local media coverage of violence. Alternatively, perhaps the incidence of
wrongful conviction in capital cases or, more generally, perceptions of bias
or a lack of confidence in the criminal justice system are keys to cross-
sectional variation in public opinion about the death penalty (Westervelt
and Humphrey 2001). Further research along these lines might contribute
to a fuller understanding of spatial variation in death-penalty support.
Finally, future research on death penalty support might also consider
the degree to which micro- and macrolevel factors interact in their effects
on death penalty attitudes. The objective of our research led us to pur-
posely fix the effects of individual-level covariates to be constant across
geographic areas, but it is plausible to speculate that the effects of some
individual-level attributes may vary across local contexts. For example,
consistent with a large body of research on punitiveness, we found that
race and gender were among the two strongest predictors of support for
capital punishment, with whites and males significantly more likely than
nonwhites and females to report that they favor the death penalty. But
are race and gender differences in punitiveness invariant across social
contexts, or are there certain structural and cultural conditions that mod-
erate these differences? Pursuing these types of questions would advance
considerably our understanding of how individual and aggregate factors
affect death penalty attitudes. In light of recent calls for a moratorium
on executions and renewed debate over the death penalty, greater attention
to the sources of public support for capital punishment in the United
States assumes special urgency.

869

This content downloaded from 128.252.067.066 on July 25, 2016 21:40:49 PM


All use subject to University of Chicago Press Terms and Conditions (http://www.journals.uchicago.edu/t-and-c).
APPENDIX
TABLE A1
Definitions and Metrics for Control Variables Included in Analysis of
Contextual Effects on Support for the Death Penalty

Variable Variable Definition and Metric

Individual-level variables:
White . . . . . . . . . . . . . . . . . . . . . . . . . . Respondent’s race (0 p nonwhite; 1 p white)
Male . . . . . . . . . . . . . . . . . . . . . . . . . . . Respondent’s sex (0 p female; 1 p male)
Age . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Respondent’s age in years
Education . . . . . . . . . . . . . . . . . . . . . Four dichotomous variables indicating the highest
educational degree attained by the respondent
(bachelor’s degree or more; junior college degree;
high school diploma or GED; no high school
degree)
Family income . . . . . . . . . . . . . . . . Eight-point scale ranging from less than $1,000 to
over $25,000
Married . . . . . . . . . . . . . . . . . . . . . . . . Respondent’s marital status (0 p unmarried; 1 p
married)
Conservatism . . . . . . . . . . . . . . . . . . Seven-point scale indicating respondent’s political
views (1 p extremely liberal; 2 p liberal; 3 p
slightly liberal; 4 p moderate; 5 p slightly con-
servative; 6 p conservative; 7 p extremely
conservative)
Religious fundamentalism . . . Fundamentalism/liberalism of respondent’s religion
(1 p fundamentalist; 0 p moderate or liberal)
Church attendance . . . . . . . . . . . Four-point scale indicating how often respondent
attends religious services (1 p less than once per
year; 2 p several times per year; 3 p 1–3 times
per year; 4 p weekly or more)
Place size . . . . . . . . . . . . . . . . . . . . . . Logged population size of the census place in
which respondent lives
South . . . . . . . . . . . . . . . . . . . . . . . . . . . Respondent resides in the south
West . . . . . . . . . . . . . . . . . . . . . . . . . . . . Respondent resides in the west
PSU-level variables:
Resource deprivation . . . . . . . . . Three-item principal components factor that com-
bines percentage of families living in poverty,
median family income, and the percentage of
families headed by a female
Unemployment rate . . . . . . . . . . Percentage of persons age 16 and older in the civil-
ian labor force who are not employed
Male divorce rate . . . . . . . . . . . . . Percentage of males age 14 and older who are
divorced

This content downloaded from 128.252.067.066 on July 25, 2016 21:40:49 PM


All use subject to University of Chicago Press Terms and Conditions (http://www.journals.uchicago.edu/t-and-c).
Support for Capital Punishment

TABLE A2
Correlations for PSU-Level Variables in Analysis of Contextual Effects on
Support for the Death Penalty

1 2 3 4 5 6 7

1. Homicide rate . . . . . . . . . . . . . . ⫺.042 .591* .424* .266* ⫺.056 .003


2. Conservative climate . . . ⫺.105 .208* .254* .093 .240*
3. %black (logged) . . . . . . . . . .271* .044 ⫺.071 .108
4. Income inequality . . . . . . . .794* .416* .175*
5. Resource deprivation . . . .452* ⫺.081
6. Unemployment rate . . . . .172*
7. Male divorce rate . . . . . . . . . .
Note.—N p 268.
* P ! .05.

REFERENCES
Anderson, Elijah. 1999. Code of the Street: Decency, Violence, and the Moral Life of
the Inner City. New York: W. W. Norton.
Barkan, Steven E., and Steven F. Cohn. 1994. “Racial Prejudice and Support for Death
Penalty by Whites.” Journal of Research in Crime and Delinquency 31:202–09.
Beckett, Katherine. 1994. “Setting the Public Agenda: ‘Street Crime’ and Drug Use
in American Politics.” Social Problems 41:425–47.
———. 1997. Making Crime Pay: Law and Order in Contemporary American Politics.
New York: Oxford University Press.
Beckett, Katherine, and Theodore Sasson. 2000. The Politics of Injustice: Crime and
Punishment in America. Thousand Oaks, Calif.: Pine Forge Press.
Best, Joel. 1993. “But Seriously Folks: The Limitations of the Strict Constructionist
Interpretation of Social Problems.” Pp. 129–47 in Reconsidering Social
Constructionism: Debates in Social Problems Theory, edited by James A. Holstein
and Gale Miller. New York: Aldine de Gruyter.
Black, Donald J. 1976. The Behavior of Law. New York: Academic Press.
Blalock, Hubert M. 1967. Towards a Theory of Minority Group Relations. New York:
Capricorn.
Blume, John H., Theodore Eisenberg, and Sheri Lynn Johnson. 1998. “Post-McCleskey
Racial Discrimination Claims in Capital Cases.” Cornell Law Review 83:1771–1803.
Blumer, Herbert. 1958. “Race Prejudice as a Sense of Group Position.” Pacific
Sociological Review 1:3–7.
———. 1971. “Social Problems as Collective Behavior.” Social Problems 18:298–306.
Blumstein, Alfred, and Richard Rosenfeld. 1998. “Trends in Rates of Violence in the
USA.” Studies on Crime and Crime Prevention 8:139–67.
Bohm, Robert M., ed. 1987. “American Death Penalty Attitudes: A Critical
Examination of Recent Evidence.” Criminal Justice and Behavior 14:380–96.
———. 1991. The Death Penalty in America: Current Research. Highland Heights,
Ky.: Anderson Publishing.
Borg, Marian. 1997. “The Southern Subculture of Punitiveness? Regional Variation in
Support for Capital Punishment.” Journal of Research in Crime and Delinquency
34:25–45.

871

This content downloaded from 128.252.067.066 on July 25, 2016 21:40:49 PM


All use subject to University of Chicago Press Terms and Conditions (http://www.journals.uchicago.edu/t-and-c).
American Journal of Sociology

———. 1998. “Vicarious Homicide Victimization and Support for Capital Punishment:
A Test of Black’s Theory of Law.” Criminology 36:537–67.
Bowers, William J., Benjamin D. Steiner, and Marla Sandys. 2001. “Death Sentencing
in Black and White: An Empirical Analysis of the Role of Jurors’ Race and Jury
Racial Composition.” University of Pennsylvania Journal of Constitutional Law 3:
171–247.
Brinker, Gary D. 1999. “Missouri Residents’ Opinions on the Death Penalty, November
1999.” Center for Social Sciences and Public Policy Research, Springfield, Missouri.
Byrk, Anthony S., Stephen W. Raudenbush, and Richard Congdon. 1996. HLM:
Hierarchical Linear and Nonlinear Modeling with the HLM/2L and HLM/3L
Programs.
Cao, Liqun, Anthony Adams, and Vickie J. Jensen. 1997. “A Test of the Black
Subculture of Violence Thesis: A Research Note.” Criminology 35:367–79.
Chambliss, William J. 1964. “A Sociological Analysis of the Law of Vagrancy.” Social
Problems 12:67–77.
Chambliss, William J., and R. Seidman. 1980. Law, Order, and Power. New York:
Addison-Wesley.
Chamlin, Mitchell B. 1989. “Conflict Theory and Police Killings.” Deviant Behavior
104:353–68.
Davis, James Allan, and Tom W. Smith. 1998. “General Social Surveys, 1972–1998:
Cumulative Codebook.” National Opinion Research Center, Chicago.
Death Penalty Information Center. 1999. Summary of Recent Poll Findings. State Polls.
Last accessed February 3, 2003. Available www.deathpenaltyinfo.org/Polls.html
DiMaggio, Paul, John Evans, and Bethany Bryson. 1996. “Have American’s Social
Attitudes Become More Polarized?” American Journal of Sociology 102:690–755.
DiPrete, Thomas A., and Jerry D. Forristal. 1994. “Multilevel Models: Methods and
Substance.” Annual Review of Sociology 20:331–57.
Dixon, Jo, and Alan J. Lizotte. 1987. “Gun Ownership and the Southern Subculture
of Violence.” American Journal of Sociology 93:383–405.
Ellison, Christopher G., and Marc A. Musick. 1993. “Southern Intolerance: A
Fundamentalist Effect? Social Forces 72:379–98.
Ellsworth, Phoebe C., and Samuel R. Gross. 1994. “Hardening of the Attitudes:
Americans’ Views on the Death Penalty.” Journal of Social Issues 50:19–52.
Finckenauer, James O. 1988. “Public Support for the Death Penalty: Retribution as
Just Desserts or Retribution as Revenge.” Justice Quarterly 5:81–100.
Firebaugh, Glenn, and Kenneth E. Davis. 1988. “Trends in Antiblack Prejudice,
1972–1984: Region and Cohort Effects.” American Journal of Sociology 94:251–72.
Fox, James A., Michael L. Radelet, and Julie L. Bonsteel. 1991. “Death Penalty Opinion
in the Post-Furman Years.” New York University Review of Law and Social Change
28:499–528.
Garland, David. 1990. Punishment and Modern Society. Chicago: University of
Chicago Press.
———. 2000. “The Culture of High Crime Societies: Some Preconditions of Recent
‘Law and Order’ Politics.” British Journal of Criminology 40:347–75.
Gelles, Richard J., and Murray A. Straus. 1975. “Family Experience and Public Support
of the Death Penalty.” American Journal of Orthopsychiatry 45:596–613.
Grasmick, Harold G., Elizabeth Davenport, Mitchell B. Chamlin, and Robert J. Bursik,
Jr. 1992. “Protestant Fundamentalism and the Retributive Doctrine of Punishment.”
Criminology 30:21–45.
Greenberg, David F., and Valerie West. 2001. “State Prison Populations and Their
Growth, 1971–1991.” Criminology 39:615–54.
Gross, Samuel R. 1998. “Update: American Public Opinion on the Death Penalty—It’s
Getting Personal.” Cornell Law Review 83:1448–79.

872

This content downloaded from 128.252.067.066 on July 25, 2016 21:40:49 PM


All use subject to University of Chicago Press Terms and Conditions (http://www.journals.uchicago.edu/t-and-c).
Support for Capital Punishment

Guo, Guang, and Hongxin Zhao. 2000. “Multilevel Modeling for Binary Data.” Annual
Review of Sociology 26:441–62.
Harvey, O. J. 1986. “Belief Systems and Attitudes toward the Death Penalty and Other
Punishments.” Journal of Personality 54:659–75.
Hawkins, Darnell F. 1987. “Beyond Anomalies: Rethinking the Conflict Perspective
on Race and Criminal Punishment.” Social Forces 65:719–45.
———. 1997. “Which Way toward Equality? Dilemmas and Paradoxes in Public
Policies Affecting Crime and Punishment.” Pp. 238–52 in African Americans and
the Public Agenda: The Paradoxes of Public Policy, edited by Cedric Herring.
Thousand Oaks, Calif..: Sage.
Hood, Roger. 1998. “Capital Punishment.” Pp. 739–76 in The Handbook of Crime and
Punishment, edited by Michael Tonry. New York: Oxford University Press.
———. 2001. “Capital Punishment: A Global Perspective.” Punishment and Society
33:331–54.
Hosmer, David W., Jr., and Stanley Lemeshow. 2000. Applied Logistic Regression. New
York: Wiley & Sons.
Jackson, Pamela Irving. 1989. Minority Group Threat, Crime, and Policing. New York:
Praeger.
Jackson, Pamela I., and Leo Carroll. 1981. “Race and the War on Crime: The
Sociopolitical Determinants of Municipal Expenditures in 90 Non-Southern U.S.
Cities.” American Sociological Review 46:290–305.
Jacobs, David. 1979. “Inequality and Police Strength: Conflict Theory and Social
Control in Metropolitan Areas.” American Sociological Review 44:913–25.
Jacobs, David, and Jason T. Carmichael. 2001. “The Politics of Punishment Across
Time and Space: A Pooled Time-Series Analysis of Imprisonment Rates.” Social
Forces 80:91–121.
———. 2002. “The Political Sociology of the Death Penalty: A Pooled Time-Series
Analysis.” American Sociological Review 67:109–31.
Jacobs, David, and Robert M. O’Brien. 1998. “The Determinants of Deadly Force: A
Structural Analysis of Police Violence.” American Journal of Sociology 103:837–62.
Keil, Thomas J., and Gennaro F. Vito. 1991. “Fear of Crime and Attitudes toward
Capital Punishment: A Structural Equations Model.” Justice Quarterly 8:447–64.
Kennedy, Randall. 1997. Race, Crime, and the Law. New York: Vintage Books.
LaFree, Gary. 1998. Losing Legitimacy: Street Crime and the Decline of Social
Institutions in America. Boulder, Colo.: Westview.
Land, Kenneth C., Patricia L. McCall, and Lawrence E. Cohen. 1990. “Structural
Covariates of Homicides Rates: Are There Any Invariances across Time and Social
Space?” American Journal of Sociology 95:922–63.
Liska, Allen E., ed. 1992. Social Threat and Social Control. Albany: State University
of New York Press.
Liska, Allen E., Joseph J. Lawrence, and Michael Benson. 1981. “Perspectives on the
Legal Order.” American Journal of Sociology 87:412–26.
Liska, Allen E., and Steven F. Messner. 1999. Perspectives on Crime and Deviance.
Englewood Cliffs, N.J.: Prentice Hall.
Longmire, Dennis R. 1996. “Americans’ Attitudes about the Ultimate Weapon: Capital
Punishment.” Pp. 93–108 in Americans View Crime and Justice: A National Public
Opinion Survey edited by T. J. Flanagan and D. R. Longmire. Thousand Oaks,
Calif.: Sage.
Massey, Douglas S., and Nancy A. Denton. 1993. American Apartheid. Cambridge,
Mass.: Harvard University Press.
Miller, Gale, and James A. Holstein. 1993. “Reconsidering Social Constructionism.”
Pp. 5–24 in Reconsidering Social Constructionism: Debates in Social Problems
Theory, edited by James A. Holstein and Gale Miller. Hawthorne, N.Y.: Aldine de
Gruyter.

873

This content downloaded from 128.252.067.066 on July 25, 2016 21:40:49 PM


All use subject to University of Chicago Press Terms and Conditions (http://www.journals.uchicago.edu/t-and-c).
American Journal of Sociology

Myers, Martha A., and James L. Massey. 1991. “Race, Labor, and Punishment in
Postbellum Georgia.” Social Problems 38:267–86.
Myers, Martha A., and Susette M. Talarico. 1986. “The Social Contexts of Racial
Discrimination in Sentencing.” Social Problems 333:236–51.
NCHS (National Center for Health Statistics). 1998. Compressed Mortality File.
Centers for Disease Control and Prevention. Last accessed February 3, 2003.
Available http://wonder.cdc.gov.
Neapolitan, J. 1983. “Support for and Opposition to Capital Punishment: Some
Associated Social-Psychological Factors.” Criminal Justice and Behavior 10:
195–208.
Niemi, Richard G., John Mueller, and Tom W. Smith. 1989. Trends in Public Opinion:
A Compendium of Survey Data. New York: Greenwood Press.
Page, Benjamin I., and Robert Y. Shapiro. 1992. The Rational Public. Chicago:
University of Chicago Press.
Patterson, Lindsay. 1991. “Multilevel Logistic Regression.” Pp. 5–18 in Data Analysis
with ML3, edited by Robert Prosser, J. Rasbash, and H. Goldstein. London: Institute
of Education, University of London.
Quillian, Lincoln. 1995. “Prejudice as a Response to Perceived Group Threat:
Population Composition and Anti-Immigrant and Racial Prejudice in Europe.”
American Sociological Review 60:586–611.
———. 1996. “Group Threat and Regional Change in Attitudes toward African
Americans.” American Journal of Sociology 102:816–860.
Radelet, Michael R., and Marian J. Borg. 2000. “The Changing Nature of Death
Penalty Debates.” Annual Review of Sociology 26:43–61.
Rankin, Joseph H. 1979. “Changing Attitudes toward Capital Punishment.” Social
Forces 58:194–211.
Raudenbush, Stephen W., and Anthony S. Bryk. 2002. Hierarchical Linear Models.
Applications and Data Analysis Methods, 2d ed. Thousand Oaks, Calif: Sage.
Rosenfeld, Richard, Steven F. Messner, and Eric P. Baumer. 2001. “Social Capital and
Homicide.” Social Forces 80:283–309.
Sampson, Robert J., and Stephen Raudenbush. 1999. “Systematic Social Observation
of Public Spaces: A New Look at Disorder in Urban Neighborhoods.” American
Journal of Sociology 105:603–51.
Scheingold, Stuart A. 1991. The Politics of Street Crime: Criminal Process and Cultural
Obsession. Philadelphia: Temple University Press.
Schneider, Joseph W. 1985. “Social Problems Theory: The Constructionist View.”
Annual Review of Sociology 11:209–29.
Schneider, William. 2000. “Political Pulse: Growing Doubts about the Death Penalty.”
Atlantic Online. On-line newsletter of the Atlantic Monthly. Last accessed February
3, 2003. Available http://www.theatlantic.com/politics/nj/schneider2000-06-14.htm
Seltzer, R., and Joseph P. McCormick. 1987. “The Impact of Crime Victimization and
Fear of Crime on Attitudes toward Death Penalty Defendants.” Violence and Victims
2:99–114.
Shapiro, Bruce. 2000. “Capital Offense.” Sunday New York Times Magazine, 26 March,
19–20.
Shaw, Greg M., Robert Y. Shapiro, and Samuel Lock. 1998. “The Polls—Trends: Crime,
the Police, and Civil Liberties.” Public Opinion Quarterly 62:405–26.
Singh, Robert. 2000. “Capital Punishment in the U.S.: A New Abolitionism?” Political
Quarterly 71:314–51.
Smith, Dwayne, and James Wright. 1992. “Capital Punishment and Public Opinion
in the Post-Furman Era: Trends and Analyses.” Sociological Spectrum 12:127–44.
Smith, Tom. W. 1976. “A Trend Analysis of Attitudes towards Capital Punishment,
1936–1974.” Pp. 255–318 in Studies of Social Change since 1948, vol. 3. Chicago:
NORC.

874

This content downloaded from 128.252.067.066 on July 25, 2016 21:40:49 PM


All use subject to University of Chicago Press Terms and Conditions (http://www.journals.uchicago.edu/t-and-c).
Support for Capital Punishment

Snijders, Tom A. B., and Roel J. Bosker. 1999. Multilevel Analysis: An Introduction
to Basic and Advanced Multilevel Modeling. London: Sage.
Spector, Malcolm, and John I. Kitsuse. 1977. Constructing Social Problems.
Hawthorne, N.Y.: Aldine de Gruyter.
Steiner, Benjamin D. 1999. “Folk Knowledge as Legal Action: Death Penalty
Judgments and the Tenet of Early Release in a Culture of Mistrust and Punitiveness.”
Law and Society Review 33:461–96.
Stinchcombe, Arthur L., Rebecca Adams, Carol A. Heimer, Kim Lane Scheppele, Tom
W. Smith, and D. Garth Taylor. 1980. Crime and Punishment: Changing Attitudes
in America. San Francisco: Jossey-Bass.
Taylor, D. G., K. L. Scheppele, and A. L. Stinchcombe. 1979. “The Salience of Crime
and Support for Harsher Criminal Sanctions.” Social Problems 26:413–24.
Taylor, Marylee C. 1998. “How White Attitudes Vary with the Racial Composition of
Local Populations.” American Sociological Review 63:512–35.
Thomas, Charles W., and Samuel C. Foster. 1975. “A Sociological Perspective on Public
Support for Capital Punishment.” American Journal of Orthopsychiatry 45:641–57.
Tyler, Tom R., and Robert J. Boeckmann. 1997. “Three Strikes and You Are Out, but
Why?” Law and Society Review 31:237–65.
Tyler, Tom, and Renee Weber. 1982. “Support for the Death Penalty: Instrumental
Response to Crime, or Symbolic Attitude?” Law and Society Review 17:21–45.
Warr, Mark. 1995. “Public Opinion on Crime and Punishment.” Public Opinion
Quarterly 59:296–310.
Westervelt, Saundra, and John A. Humphrey, eds. 2001. Wrongly Convicted:
Perspectives on Failed Justice. New Brunswick, N.J.: Rutgers University Press.
Wong, George Y., and William M. Mason. 1985. “The Hierarchical Logistic Regression
Model for Multilevel Analysis.” Journal of the American Statistical Association 80:
513–24.
Zaller, John R. 1992. The Nature and Origins of Mass Opinion. Cambridge: Cambridge
University Press.
Zeisel, H., and A. M. Gallup. 1989. “Death Penalty Sentiment in the United States.”
Journal of Quantitative Criminology 5:285–96.
Zimring, Frank E., and Gerald Hawkins. 1986. Capital Punishment and the American
Agenda. Cambridge: Cambridge University Press.

875

This content downloaded from 128.252.067.066 on July 25, 2016 21:40:49 PM


All use subject to University of Chicago Press Terms and Conditions (http://www.journals.uchicago.edu/t-and-c).

You might also like