Vous êtes sur la page 1sur 28

595420

research-article2015
IJOXXX10.1177/0306624X15595420International Journal of Offender Therapy and Comparative CriminologySchaible and Altheimer

Article
International Journal of
Offender Therapy and
Social Structure, Anomie, and Comparative Criminology
2016, Vol. 60(8) 936­–963
National Levels of Homicide © The Author(s) 2015
Reprints and permissions:
sagepub.com/journalsPermissions.nav
DOI: 10.1177/0306624X15595420
ijo.sagepub.com

Lonnie M. Schaible1 and Irshad Altheimer2

Abstract
Merton’s “Social Structure and Anomie” seeks to explain how “socio-cultural”
influences exert “definite pressures” to engage in non-conformity. Despite having
a significant influence, few studies have assessed the degree to which Merton’s
propositions explain cross-national variation in levels of crime. Using data on national
levels of homicide, data from the World Values Survey, and other structural controls,
the present study assesses the degree to which deinstitutionalization, demoralization,
and blocked opportunity interact to explain crime cross-nationally. Results provide
a high degree of support for Merton’s assertion that societal types characterized by
relatively high levels of materialism and/or demoralization or deinstitutionalization
suffer from higher levels of homicide. However, there is less support for Merton’s
assertion that inequality interacts with various societal patterns of means/ends
integration in a meaningful way. Findings and implications for the utility of classical
anomie as a general macro-level theory are discussed.

Keywords
cross-national, criminology, homicide, strain, anomie, Merton, culture

Merton’s (1938; revised in 1949, 1957, 1964, and 1968) “Social Structure and Anomie”
is among the most significant criminological articles ever published, promoting sub-
stantial debate, empirical research, and inspiration for several prominent contempo-
rary theoretical frameworks. Despite this, relatively little empirical literature has
directly addressed the propositions of classical anomie theory. Specifically, although
some studies have examined the widely recognized micro-level “deviant adaptations”

1University of Colorado Denver, USA


2Rochester Institute of Technology, NY, USA

Corresponding Author:
Lonnie M. Schaible, School of Public Affairs, University of Colorado Denver, 1380 Lawrence Street, Suite
500, Denver, CO 80217, USA.
Email: lmschaible@gmail.com
Schaible and Altheimer 937

(conformity, innovation, ritualism, retreatism, and rebellion), few studies, to date,


have directly examined Merton’s macro-level propositions, and none have assessed
how social structures and cultural attributes of societies identified by Merton jointly
produce different levels of crime.
At the core of Merton’s macro-level arguments regarding societal levels of crime is
the notion that societies with a relative balance between cultural pressures for success
and institutional control of the means for success, should benefit from lower levels of
deviance. In contrast, inasmuch as societies suffer from an imbalance between their cul-
turally prescribed means and the pressures for success, Merton suggests that “antisocial
behavior” is likely to “ensue on a considerable scale.” In seeking to clarify these societal
level relationships, Merton (1938) explicitly identifies a number of “cultural patterns” of
means/goals integration (p. 673), likely to result in different levels and types of societal
crime. These types have not yet been explored or discussed in the empirical literature,
but they would seem to have considerable value in understanding how opportunity is
likely to interact with cultural and structural features of various societies.
Using data on national levels of homicide, data from the World Values Survey
(WVS), United Nations, World Bank, World Health Organization, International
Labour Office, and CIA World Factbook, the present study assesses hypotheses regard-
ing the explanation of societal levels of crime central to Merton’s (1938) seminal state-
ment in “Social Structure and Anomie.” Specifically, we assess how several “cultural
patterns” of means/ends (mal)integration identified by Merton influence levels and
patterns of crime at the national level both independently and in interaction with mea-
sures of opportunity, while controlling for other relevant factors. Findings and impli-
cations for the utility of Merton’s typology of societies as a general macro-level and
cross-national theory are discussed.

A Re-Examination of Merton’s Anomie Theory


A key tenet of classical anomie theory is that cultural pressures to pursue material
goals result in the subordination of means for attaining these. Drawing on this tenet,
the dominant approach in the literature has been to suggest that the degradation of
means by over-emphasis on goals is almost invariant—that is, it is virtually impossible
for a society to have both a strong value of material goals and strongly institutionalized
means for achieving those (Baumer, 2007; Baumer & Gustafson, 2007; Bernard, 1987;
Messner, 1988; Messner & Rosenfeld, 1994; Messner, Thome, & Rosenfeld, 2008).
However, Merton (1938) takes a softer view of the causal relationship than is ordinar-
ily recognized, stating,

To say that these two elements, culture goals and institutional norms, operate jointly is
not to say that the ranges of alternative behaviors and aims bear some constant relation to
one another. The emphasis upon certain goals may vary independently of the degree of
emphasis upon institutional means. (p. 673)

Thus, Merton implies that although there likely is a strong connection, various combi-
nations of means and goals do inevitably exist in the cultural and structural variation
938 International Journal of Offender Therapy and Comparative Criminology 60(8)

observed across societies. Consistent with this premise, Merton (1938) further
acknowledges three possible “cultural patterns” (p. 673) that societies are likely to fall
into: stable, deinstitutionalized, and ritualist. Consideration of each of these types
raises interesting questions about their implications for explaining various outcomes,
especially with regard to how they interact with opportunity structures.
According to Merton, in “stable” societies, a relative balance exists between the
emphasis on material goals and the acceptable means for attaining them. Such societ-
ies neither over-emphasize material success goals, nor compel members of such soci-
eties to defiantly adapt when goals are unfulfilled. Although some individuals may
aspire to and pursue more lofty material goals—the general pressures to achieve
wealth (e.g., “The American Dream”) are likely subordinated and only heralded inas-
much as they are a product of virtuous means. Similarly, when one is blocked from
attaining goals, the deviant adaptations identified by Merton (innovation, retreatism,
ritualism, and rebellion) are minimized as viable options—because conformity is val-
ued in balance with achievement. As such, individuals are compelled to pursue success
by accepted and conventional means (i.e., education, hard work, etc.), without feeling
undue pressure to “win”—that is, it is not so much about whether you win or lose, it is
about how you play the game. As a benefit, these societies are likely to be character-
ized by a relatively stable social order, and lower levels of antisocial behavior. In
contrast to stable societies, Merton suggests that two types of imbalance between
goals and means are likely to produce divergent consequences in interaction with other
factors, especially opportunity.
In “deinstitutionalized” societies, pursuit of material goals is “limited only by tech-
nical rather than institutional considerations” (Merton, 1938, p. 673). Moreover,
Merton (1938) argues,

The process whereby exaltation of the end generates a literal demoralization, i.e., a
deinstitutionalization, of the means is one which characterizes many groups in which the
two phases of the social structure are not highly integrated. The extreme emphasis upon
the accumulation of wealth as a symbol of success in our own society militates against
the completely effective control of institutionally regulated modes of acquiring a fortune.
(p. 675)

This conforms to most traditional and contemporary interpretations of Mertonian


theory, which argue that excess emphasis on material goals generally corresponds with
the subordination of conventional means for achieving these. Thus, it is argued, amid
strong pressures for material success, members of deinstitutionalized societies freely
deviate from conventional means to achieve material symbols of status and success. This
is particularly true when conventional means for attaining success are structurally
blocked for considerable portions of the population. Specifically, blocked or limited
opportunity plays a clear and central role as a conditioning factor—in other words, amid
a strong emphasis on material success, actors are compelled to engage in deviant adapta-
tions inasmuch as they are blocked from achieving success by conventional means.
Beyond the more familiar “stable” and “deinstitutionalized” types of societies,
Merton also offers the possibility of “ritualist” societies, in which rigid adherence to
Schaible and Altheimer 939

means without concern for material goals produces a “tradition-bound, sacred society
characterized by neo-phobia” (i.e., strict observance of accepted means and traditional
roles is the goal; Merton, 1938, p. 673). The possibility of such a “cultural pattern”
raises interesting questions. Of particular interest is how the central concept of oppor-
tunity is likely to interact within such a cultural context. Seeking to provide some
insights into this conundrum, Merton (1938) states as follows:

A high frequency of deviate behavior is not generated simply by “lack of opportunity” or


by this exaggerated pecuniary emphasis. A comparatively rigidified class structure, a
feudalistic or caste order, may limit such opportunities far beyond the point which obtains
in our society today. It is only when a system of cultural values extols, virtually above all
else, certain common symbols of success for the population at large while its social
structure rigorously restricts or completely eliminates access to approved modes of
acquiring these symbols for a considerable part of the same population, that antisocial
behavior ensues on a considerable scale. (p. 680)

Here, Merton suggests that the specific cultural patterns of means/goals integration
observed in a society are likely to be critical in determining whether blocked opportu-
nity matters as a crime generating force. This means that in “ritualist” societies, that do
not emphasize material success for all (i.e., universalism) and, in fact, strongly empha-
size observance of traditional roles, blocked opportunities are theoretically rendered
neutral as a criminogenic force. In fact, when examined globally, we may expect these
societies, despite their inegalitarian social structures, to benefit from relatively low lev-
els of “antisocial behavior” or patterns of crime. Alternatively, they may reflect a differ-
ent emphasis or substantive pattern than is observed in other societies (e.g., honor
killings in enforcement of rigid norms intended to reinforce the existing social order).
Taken together, Merton’s notion of cultural patterns suggests interesting implica-
tions for both theory and research that have not yet been explored. Specifically, con-
trary to past research and theoretical development, it suggests that the relationship
between means and goals is not invariant. Furthermore, it suggests that different pat-
terns of means/goals integration are likely to substantively condition the effects of key
intervening variables such as blocked opportunity. Stated another way, although
blocked opportunity may be highly consequential in deinstitutionalized societal types,
it may have relatively benign effects in either stable or ritualistic types of societies,
which both have a stronger emphasis on preserving traditional means. Thus, drawing
on Merton’s thinking, it seems important to account for cultural context (means/goals
integration), when cross-nationally examining the effects of opportunity structures on
levels of crime. We now turn our attention to a discussion of extant research on classi-
cal anomie theory.

Empirical Studies of Anomie


Despite the prominence of classical anomie theory, there have been few macro-level tests
(Pratt & Cullen, 2005). Most research has focused on the micro-level processes specified
by Merton, especially his typology of deviant adaptations. Only a few macro-level studies
940 International Journal of Offender Therapy and Comparative Criminology 60(8)

have explicitly tested Merton’s propositions (Baumer & Gustafson, 2007); however, as a
result of the lack of suitable data, no study has assessed the cross-national relevance of
Merton’s macro-level propositions.
Of the existing cross-national research, much of it has focused on the measurement
and causes of Durkheimian anomie or normlessness, which closely resembles Merton’s
notion of the breakdown of institutionalized means. For example, Cao (2004) found
that demographic measures influenced rates of anomie across nations and that anomie
in the United States did not differ significantly from other English-speaking nations. In
another article, Cao (2007) found that employment, having children, and life satisfac-
tion were negatively associated with anomie among respondents in China. In addition,
R. Zhao and Cao (2010) found that individuals living in nations undergoing rapid soci-
etal change experienced higher levels of anomie. Similar conclusions were drawn by
Thorlindsson and Bernburg (2004), who found that community political participation
influenced levels of anomie among juveniles in Iceland. More importantly, none of the
research discussed above directly informs the relationship between anomie and crime.
Existing macro-level tests of the relationship between anomie and crime rates fall
into three categories. The first involves national case studies that examine how rapid
social change has influenced crime rates within individual nations. For example, Liu
(2005) found that crime rates increased in China from 1978 to 1999, a peak period of
economic growth and change in China, for both economically and non-economically
motivated crimes. Interestingly, Liu (2005) found that rates of economically motivated
crimes rose faster than rates of non-economically motivated crime. Similarly, L. Zhao
(2008) found that rapid social change in China was associated with an explosion in
property crime and political corruption. Taken together, these studies suggest that rapid
social transformations lead to increases in many different types of crime. Examination
of data from other nations yields similar conclusions. Recent research also reveals that
socio-political changes in former Soviet Bloc nations correspond with increased rates
of crime (Kim & Pridemore, 2005; Stamatel, 2009). Admittedly, these studies are not
direct tests of classical anomie theory. They do, however, inform our understanding of
the theory if we assume that one outcome of the social changes occurring in these
nations is the emergence of a culture of self-interest and pursuit of material wealth.
Indirect support for classical anomie can also be drawn from cross-national research
on theories that share conceptual overlap with classical anomie. For example, Baumer
and Gustafson (2007) note that both classical and institutional anomie assert that crime
will be higher in societies characterized by a strong cultural emphasis on monetary suc-
cess goals and a weak emphasis of pursuing those goals through legitimate means.
Thus, the robust effects revealed in research on institutional anomie theory (i.e.,
Messner & Rosenfeld, 1997; Pratt & Cullen, 2005; Savolainen, 2000) provide indirect
support that a cross-national examination of Merton’s original arguments should receive
empirical support. Relatedly, in a recent test of Bonger’s (1916) theory, Antonoccio and
Tittle (2007) found that nations with higher levels of demoralization have higher rates
of homicide. Bonger’s (1916) notion of demoralization, which is characterized by an
egoistic moral environment where individuals place their selfish interests over the
needs of others, shares remarkable resemblance to Merton’s arguments.
Schaible and Altheimer 941

Perhaps, the most comprehensive test of classical anomie theory has been per-
formed by Baumer and Gustafson (2007), who examined Merton’s core propositions
using 77 geographic units within the General Social Survey (GSS). Baumer and
Gustafson (2007) found support for the proposition that instrumental crime was higher
in geographic areas characterized by a high commitment to monetary success and a
low commitment to legitimate means to attain that success. Less support was found,
however, for the proposition that the interaction between commitment to monetary
success and low commitment to legitimate means is moderated by other features of the
social structure. More specifically, job availability, economic attainment, and eco-
nomic inequality were all found not to moderate the commitment to monetary success/
low commitment to legitimate means interaction. Support for the notion that Baumer
and Gustafson’s (2007) findings can be generalized to the cross-national level is found
in recent research by Chamlin and Sanders (2013). Their research did not include a
direct measure of anomie but found that measures of acceptance of material success
interacted with levels of relative and absolute deprivation, respectively, to influence
cross-national variation in drug trafficking. These findings lend support to Merton’s
argument that the cultural and structural components of the social system interact to
influence macro-level variation in crime.
Taken together, the existing macro-level research on classical anomie lends sup-
port to several key tenets of classical anomie theory. Specifically, societies with high
anomie, or a decline in observance of institutional means, tend to suffer from rela-
tively higher levels of crime. Similarly, societies dominated by an egoistic or materi-
alistic environment also suffer from higher levels of crime. However, the research
remains ambiguous on other key factors. Specifically, although inequality and other
indicators of structural obstacles are consistent predictors of aggregate crime rates,
studies have been inconsistent regarding the hypothesized interaction between
blocked opportunities, institutionalized means, and materialism. Furthermore, no
study to date has examined the degree to which different patterns of means-ends inte-
gration affect levels of crime in a cross-national sample that has sufficient variation
to represent the various types identified by Merton. Given this, the present study
seeks to begin to fill this gap and further develop Merton’s theoretical project started
almost 80 years ago.

Hypotheses
Based on the theory and empirical research presented above, we hypothesize the
following:

Hypothesis 1 (H1): Stable societal types, which are characterized by a balanced


emphasis on material goals and means for achieving them, will benefit from lower
levels of homicide.
Hypothesis 2 (H2): Deinstitutionalized or demoralized societal types, which are
characterized by an over-emphasis on material goals and the subordination of tradi-
tional means for achievement, will suffer from higher levels of homicide.
942 International Journal of Offender Therapy and Comparative Criminology 60(8)

Hypothesis 3 (H3): Ritualistic societies, which have a strong emphasis on tradi-


tional means and a de-emphasis of material goals, will not exhibit a significant
relationship with levels of homicide (due to the possibility of high levels of intra-
family violence, which offset the benefits of traditionalism in reducing interper-
sonal violence).
Hypothesis 4 (H4): Measures of inequality in the opportunity structure will only
significantly interact with deinstitutionalized or demoralized societal types to pro-
duce multiplicatively higher levels of homicide.

Data and Method


To assess the degree to which Mertonian anomie theory explains variation in homicide
rates at the national level, we utilize aggregated indicators from the most recent year
relevant data are available (WVS, 2009; see also Inglehart & Baker, 2000).1 The WVS
is a data collection effort begun by the European Values Survey group in 1981 and
expanded from roughly 25,000 respondents from 21 countries in Wave 1 (1981-1984) to
more than 100,000 respondents from 87 countries across all waves.1 Individual partici-
pants were contacted by a professional survey organization in their home countries and
asked to respond to a series of questions about their values and beliefs toward a range of
local, regional, and national issues and institutions. Demographic information was also
solicited from participants. All interviews were conducted in the native language of
respondents, after which the data were coded and compiled into publicly available data
sets.2 Samples within each country are intended to be representative of the national adult
(18 and older) population, with weights provided to correct for over-sampling of under-
represented groups and cross-national differences in sample size (intra-nation responses
scaled to n = 1,500). Although the sample is disproportionately Western and developed,
a significant number of less developed and non-Western nations are represented.
To control for correlates of cross-national homicide, data from WVS aggregated to
the national level as weighted averages were merged into a data set with other struc-
tural indicators commonly used in past studies (e.g., Bjerregaard & Cochran, 2008a,
2008b; Messner & Rosenfeld, 1997; Savolainen, 2000). Structural indicators were
drawn from the United Nations, World Bank, World Health Organization, International
Labour Office, and CIA World Factbook. Despite limitations discussed widely else-
where, these data provide sufficient consistency to permit analysis of variation between
nations (LaFree, 1999; Neapolitan, 1997). Consistent with past cross-national research
on homicide, the number of nations for which most data were available was 35. This
sample size is comparable with other published cross-national studies (Nivette, 2011).
Data used in the analysis have been provided in the Appendix.

Dependent Variable
The bulk of cross-national research has examined homicide rates (per 100K) as a
dependent variable, because less serious forms of instrumental crime are likely to suf-
fer from inconsistencies in legal definitions and reporting across cultural contexts. In
Schaible and Altheimer 943

contrast, homicide is considered to be the most reliable indicator of antisocial behavior


across historical and cultural units, as its definition has remained relatively consistent.
Moreover, given its seriousness, it is very likely to be reported or detected and reflected
in official statistics. Paradoxically, most critiques of Mertonian anomie theory have
noted one of its primary limitations as the inability to account for non-instrumental
crime (Kornhauser, 1978); however, Merton offers a much broader scope for classical
anomie theory than has traditionally been recognized. Specifically, although the driv-
ing theoretical forces are deeply intertwined with economic factors, and have conse-
quently been presumed to primarily drive “instrumental” crime, Merton (1938)
suggests that “antisocial behavior ensues on a considerable scale” (p. 680) when cul-
tural emphases on material goals is blocked by structural inequality. This viewpoint
resonates with developments in the contemporary literature exploring anomie theory
and its intellectual progeny, which have only recently started to seriously explicate
how anomie can account for other forms of non-instrumental criminal violence via
their impact on societal institutions (see especially, Messner & Rosenfeld, 2007; Stults
& Baumer, 2008). Considering Merton’s concern with “anti-social behavior” noted in
the quote above and contemporary efforts to expand beyond instrumental crime—
critiques of anomie theory being limited to instrumental crimes take on less relevance
(Stults & Baumer, 2008). Specifically, inasmuch as a society lacks sufficient social
constraints resulting from structural imbalances, this is likely to have broad conse-
quences for all types of crime. Consider, for example, the relationship between con-
centrated disadvantage and homicide rates in the United States. Furthermore, previous
research on institutional anomie theory—which shares considerable overlap with clas-
sical anomie—has examined homicide as the dependent variable; thereby establishing
the utility homicide as a dependent variable in tests of anomie theories (Bjerregaard &
Cochran, 2008a, 2008b; Messner & Rosenfeld, 1997; Savolainen, 2000). Thus, con-
sistent with the cross-national literature and recent developments in research on ano-
mie theory, we treat the national homicide rate (per 100K) as our outcome measure.
Specifically, we utilize the average United Nations (UN) homicide rate (per 100K) for
the years 2005 to 2008 (the most recent years of availability for WVS data).

Measuring Materialism and Anomie


As discussed earlier, Merton’s notion of cultural patterning has not been empirically
assessed. To operationalize this concept, we developed two unique sets of typologies
representing the three types of societies discussed in Merton’s seminal work (stable,
deinstitutionalized, and ritualistic). To create these typologies, it was first necessary to
identify variables measuring the independent dimensions of materialism (goals) and
anomie (means), which combine to define the aforementioned societal types.
The first step in developing the typologies was to carefully examine the variables
available in WVS data. Using theory to guide inclusion of items in factor analysis with
varimax rotation (not presented here in the interest of space), three variables were iden-
tified as capturing the independent dimensions of materialism and anomie critical to
Merton’s definition of societal types. One item measures materialism (single item), and
944 International Journal of Offender Therapy and Comparative Criminology 60(8)

two items measure alternative dimensions of anomie that are consistent with Merton’s
theory and subsequent empirical measurement—demoralization (eigenvalue = 3.13)
and deinstitutionalization (eigenvalue = 2.76). These measures of societal dimensions
and the process for combining them into typologies are discussed in detail below.

Materialism.  Societal value of economic success is featured prominently in Merton’s


theoretical framework. However, when considered in a contemporary global context,
measurement of materialism is complicated. Here, we use a single item from the WVS
that asked respondents to report the degree to which they felt increased future empha-
sis on money and material possessions was good (2), didn’t mind (1), or bad (0). In
doing so, we capture the full range of possible cultural emphases on materialism,
including the possibility of antimaterialistic values that might be associated with more
traditional or ritualistic societies.

Dimensions of anomie.  Equally prominent in Merton’s theoretical framework is the


concept of anomie. In as much as a society has de-emphasized conventional methods
of attaining culturally prescribed goals, anomie is indicated. A number of items within
WVS measure the degree of acceptance or rejection of institutionalized means. Factor
analyses (available on request) indicated that two separate dimensions of anomie char-
acterized the data. As such, summated scales were created to measure each of these
distinct conceptual operationalizations of anomie—demoralization and deinstitution-
alization (as discussed below).

Demoralization.  Drawing on the work of R. Zhao and Cao (2010) and supported by
factor analysis using varimax rotation, we operationalize demoralization as the level of
support for minor infractions that avert economic costs or afford economic benefits to
those who undertake them. Specifically, respondents were asked the degree to which each
of the following was “justifiable” (1-10): claiming undeserved benefits, avoiding fares on
public transport, cheating on taxes, and accepting a bribe in the course of one’s duties.
This is consistent with Merton’s notion that in unstable societies, individuals are willing
to compromise means when pursuing economically beneficial ends. When combined,
these indicators comprise a scale with an alpha reliability of .867 (eigenvalue = 3.13).

Deinstitutionalization. Within the WVS, a number of indicators assess the degree to


which hard work is considered essential for individuals and society. A stronger work
ethic, as evidenced by these measures, is indicative of a society that has a strong value
of institutionalized means as envisioned by Merton. After reversing the coding to
reflect deinstitutionalization (5 = strongly disagree; 1 = strongly agree), we combine
the following Likert-type items into a summated scale of deinstitutionalized means
(eigenvalue = 2.76, α = .892): need a job to develop talents, humiliating to receive
money without working for it, people who do not work turn lazy, work is a duty toward
society, and work should come first even if it means less spare time. This is consistent
with Merton’s notion that stable societies are characterized by a relatively strong
emphasis on pursuing material ends through conventional means.
Schaible and Altheimer 945

Operationalizing Societal Types


Utilizing the indicators of cultural materialism, and the two dimensions of anomie
(demoralization and deinstitutionalization) described above, two separate sets of
typologies were developed to capture each dimension of anomie. The general
approach to operationalization entailed identifying societies that had imbalances
between materialism and anomie—which, when combined, define the respective
societal types identified by Merton (deinstitutionalized OR demoralized | stable |
ritualist).
Given that our goal was to assess societal deviations from normal, all three vari-
ables were examined for normality prior to creation of typologies (using histograms
and Tukey’s, 1977, ladder of powers). It was determined that the base measures of
demoralization and deinstitutionalization were statistically normal; however, material-
ism was negatively skewed. To normalize materialism, the variable was cubed (^3),
prior to creation of the typologies utilized in the analysis. Two sets of typologies were
created using the variables noted above—one set based on the demoralization indica-
tor and paired with the materialism^3 variable and another set based on the deinstitu-
tionalization indicator and paired with the transformed materialism variable (see
detailed discussion below). This resulted in two sets of dummy variables for each
societal type. Taken together, these six dummy variables have the benefit of parsimo-
niously capturing the joint influences of both materialism and anomie and capturing
alternative conceptualizations of anomie reflected in the empirical literature.3
Operationalization of each type is discussed, respectively, as the “demoralization-
based” typology (includes demoralization indicator discussed above) and the “deinsti-
tutionalization-based” typology (includes deinstitutionalization indicator discussed
above).

Demoralization-Based Typology
Using the materialism^3 and demoralization variables, dichotomies representing each
societal type were created as follows.

Demoralized.  Any society exhibiting a value one standard deviation above the mean on
either materialism or demoralization was coded as 1, with all other values being coded
as 0. This captures the existence of a societal level imbalance in which either material-
ism or the acceptability of engaging in deviant behaviors for personal gain is abnor-
mally high. Of all nations included in the sample, 28.6% (n = 10) fell within this
typology.

Stable (using demoralization).  All societies, within one standard deviation of the mean
on both materialism and demoralization were coded as 1, with all other values being
coded as 0. This captures the existence of a relative balance between cultural goals of
materialism and constraints on the means for achieving those goals. Of all nations
included in the sample, 45.7% (n = 16) fell within this typology.
946 International Journal of Offender Therapy and Comparative Criminology 60(8)

Ritualist (using demoralization).  Any society exhibiting a value one standard deviation
below the mean of either materialism or demoralization was coded as 1, with all other
values being coded as 0. This captures the existence of a means/goals balance that
either de-emphasizes material goals or over-emphasizes strong prohibition of deviant
means for achieving these goals. Of all nations included in the sample, 25.7% (n = 9)
fell within this typology.

Deinstitutionalization-Based Typology
Using the materialism^3 and deinstitutionalization variables, dichotomies represent-
ing each societal type were created as follows.

Deinstitutionalized. Any society exhibiting a value one standard deviation above the


mean on either materialism or demoralization was coded as 1, with all other values
being coded as 0. This captures the existence of a societal level imbalance in which
either over-emphasis of materialism or the de-emphasis of achieving these goals
through conventional means has occurred. Of all nations included in the sample, 20%
(n = 7) fell within this typology.

Stable (using deinstitutionalization).  All societies, within one standard deviation of the
mean on both materialism and demoralization, were coded as 1, with all other values
being coded as 0. This captures the existence of a relative balance between cultural
goals of materialism and conventional means for achieving those goals. Of all nations
included in the sample, 54.3% (n = 19) fell within this typology.

Ritualist (using deinstitutionalization).  Any society exhibiting a value one standard devia-
tion below the mean of either materialism or demoralization was coded as 1, with all
other values being coded as 0. This captures the existence of a means/goals imbalance
that either de-emphasizes material goals or over-emphasizes conventional means for
achieving these goals. Of all nations included in the sample, 25.7% (n = 9) fell within
this typology.

Measuring Inequality and Opportunity


Classical anomie theory posits a critical role for inequality. Specifically, inasmuch as
a society has an inequitable distribution of wealth, this indicates a potentially uneven
opportunity structure that is likely to contribute to anomie and higher levels of crime.
All published macro-level studies of classical anomie have used indicators of inequal-
ity of opportunity as a proxy for blocked opportunity (Bennett & Basiotis, 1991;
Chamlin & Cochran, 2005, 2006; Cochran & Bjerregaard, 2012; Nivette, 2011). In
seeking to assess the influence of opportunity specified by anomie theory, past studies
have used a variety of indicators to assess inequality: the Gini coefficient, ratio of
income between the richest and poorest segments of the population, Gross Domestic
Product (GDP) or Gross National Income (GNI) per capita, and purchasing power
Schaible and Altheimer 947

parity (Altheimer, 2007, 2008; P. Blau, 1977; J. Blau & Blau, 1982; Krohn, 1976; Pratt
& Godsey, 2003; Savolainen, 2000). A debate exists in the literature about which mea-
sure is superior, and any measure is likely to have its own unique problems.
Seeking to assess diverse indicators of inequality, we combine two common indica-
tors of inequality: the Gini coefficient and a ratio-based measure of income inequality.
A past problem with the Gini coefficient has been that it is not widely available across
nations for a consistent block of years; however, the Standardized World Income
Inequality Database (SWIID) has compiled data on the Gini coefficient for 171
nations. Corresponding with the time frame for which the homicide rate and WVS data
were most recently available, the average of Gini coefficients for data available
between 2005 and 2008 was used. As an additional measure of inequality, we use the
ratio of median incomes between the richest and poorest 20% used by Pratt and Godsey
(2003) and Altheimer (2007, 2008). Seeking to maximize the number of data points
available, we combine these indicators into a scale (α = .90, eigenvalue = 1.7).

Control Variables
Sex ratio.  Merton takes no clear position on the role of demographic factors in crime;
however, sex ratio has been shown to be an important predictor of cross-national rates
of crime (Barber, 2000; Guttentag & Secord, 1983; Messner & Sampson, 1991;
Schacht, Rauch, & Mulder, 2014; South & Messner, 1987). Specifically, although
counter-intuitive, it has been found that a higher ratio of females-to-males leads to
higher levels of crime when examined cross-nationally. As such, it is standard practice
to include sex ratio (males to females) as a control variable within cross-national anal-
yses of homicide (Altheimer, 2007, 2008; Guttentag & Secord, 1983; Messner &
Sampson, 1991; Pratt & Godsey, 2003; Savolainen, 2000; Schaible & Hughes, 2011;
South & Messner, 1987).

Development. Although Merton does not posit a clear role for development per se,
variability in development is likely to have an impact within a modern global context.
This has been widely supported in the cross-national literature, which has found consis-
tent effects of modernization and development on anomie (R. Zhao & Cao, 2010) and
levels of crime (Bennett, 1991; LaFree & Drass, 2002; Messner, 1982, 1986). Our mea-
sure of modernity parallels Chamlin and Cochran’s (2006). Using factor analysis with
varimax rotation, a single component with an eigenvalue of 4.7 was identified. The fol-
lowing items loaded highly on this scale (α = .95): % urban, % 15-24, phone lines per
capita, electricity consumption (KWH per capita); average years of education for adults;
Human Development Index (HDI) components (2005): GDP per capita; % of GDP spent
on health, social, education programs; life expectancy index. In addition, infant mortal-
ity, which Pridemore (2008, 2011) has identified as an important indicator of poverty/
absolute deprivation, also loaded highly with the factors above and was included in the
scale. Because a population estimate of individuals in their crime prone years is included,
it also serves as an important control for the influence of the age structure of the popula-
tion (Gartner, 1990; Gartner & Parker, 1990; Pampel & Gartner, 1995).
948 International Journal of Offender Therapy and Comparative Criminology 60(8)

Analytical Procedures
The process underlying the generation of crime is more likely a Poisson process than
a random one (D’Unger, Land, McCall, & Nagin, 1998; Land, McCall, & Nagin,
1996; Osgood, 2000). Moreover, because homicide is among the rarest of all criminal
events, both the number and rate are likely to take on a Poisson distribution. In other
words, within a discrete period of time, most aggregate units will have both a small
count and rate of homicide, while a few will have unusually high rates of homicide.
Consider, for example, the distribution of homicide across census tracts in the United
States—within any given year, most have a very low or zero value, although a very
small minority have a very high count and/or rate (see Lee & Ousey, 2001; Osgood,
2000; Osgood & Chambers, 2000). The same applies to homicide rates examined
cross-nationally. In other words, most countries have very low rates of homicide in
any given year; however, a very small fraction of countries has very high levels of
homicide. Thus, the distribution of homicides, whether expressed as a rate or count,
tends to approximate a Poisson distribution more than they do a normal distribution
(Chamlin & Cochran, 2006). This poses a number of problems from a statistical
perspective.
The primary consequence of the skewed distribution of rare events such as
homicide rates is that it produces abnormal error and covariance structures that are
likely to be non-linear, contribute to heteroskedasticity, and violate traditional
assumptions of normality. As such, traditional ordinary least squares (OLS) meth-
ods are inappropriate. Although OLS with log transformed homicide rates have
been utilized to overcome the problems noted above, such approaches risk failing
to capture the complex and non-linear error structures that are likely to exist within
rare event data (Gardner, Mulvey, & Shaw, 1995; Wooldridge, 1997). As such, con-
temporary criminologists have looked to the broader family of generalized linear
models (GLM or GLiM), which are specifically designed to address the more com-
plicated error structures of rare event data (McCullagh and Nelder, 1989). The fol-
lowing quote from Maume and Lee (2003, p. 1158) succinctly summarizes the
rationale:

Current practice in the macrolevel criminological literature is to employ a Poisson-based


estimation strategy (see Lee & Ousey, 2001; Osgood, 2000; Osgood & Chambers, 2000).
Poisson estimators are well suited for rare event data because even in the face of skewed
distributions, they still provide more efficient estimates than ordinary least-squares
regression.

Although Poisson and negative binomial models (both extensions of GLM) are
generally deemed superior for handling rare event data, there is very little guidance on
which variant is best for any given situation. Key considerations are whether equidis-
persion exists in the data (variance is equal to the mean) and which model best accounts
for error that cannot be explained by the model. To address the problem of over-dis-
persion and model error with count data, most criminologists have utilized negative
binomial regression. Chamlin and Cochran (2006) note,
Schaible and Altheimer 949

Because there is no a priori reason to assume that cross-national differences in homicide


rates can be completely accounted for by the conventional Poisson regression model, it is
considered prudent to also calculate the negative binomial variant of this regression
model. (p. 244)

Although negative binomial regression has generally been utilized to examine


count data, it has also been adapted to rate data (Hilbe, 2011; Long 1997). This is done
through a relatively simple transformation, which involves including a term in the
equation that logs the deflator that would ordinarily be utilized for count data (i.e.,
population). Given the strong positive skew of our dependent variable, the superior
capabilities of GLM models to account for resultant complex error structures, and the
capacity for negative binomial models to account for rate data and error not explained
by the model, we use negative binomial regression to examine homicide rates (per
100K) as the dependent variable.
Drawing on the literature and reasoning noted above, we apply negative binomial
regression to assess the degree to which coefficients for the relationship between soci-
etal types and homicide rates are significant in the expected directions, while control-
ling for known covariates of homicide.4 Based on H1, we expect a significant negative
relationship between “stable” types and homicide. Based on H2, we expect a signifi-
cant positive relationship between “demoralized or deinstitutionalized” types and
homicide. Based on H3, we expect no significant relationship between “ritualist” type
societies and homicide. Similarly, we expect inequality and other indicators of blocked
opportunity to exhibit significant independent effects on levels of homicide; however,
we expect that effects of inequality will be conditioned by the societal patterns of
means-ends integration, in a manner consistent with Merton’s discussion noted previ-
ously. Specifically, consistent with H4, we expect a significant effect for the multipli-
cative interaction between both demoralized and deinstitutionalized types and
inequality, but no significant interactions between inequality and stable or ritualist
types. Thus, inequality is only likely to have a strong and clear conditioning effect in
deinstitutionalized/demoralized societies.

Results
Descriptive statistics and bivariate correlations are presented in Table 1. Descriptive
statistics indicate that all variables (except homicide) were approximately normally
distributed after transformations were applied. Most societies were categorized as
“stable” across both measures of means/ends integration. When utilizing the demoral-
ization-based indicator of societal type, 45.7% of societies were classified as “stable”;
similarly, when utilizing the deinstitutionalization-based indicator, 54.3% of societies
were stable. Only 20% of societies were categorized as deinstitutionalized, although
28.6% fell into the demoralized category. In all, 25.7% of societies fell into the “ritual-
ist” category under both measures.
All control variables were significantly correlated with the UN homicide rate
(2005-2008) in the expected direction, except sex ratio that did not emerge as a
950
Table 1.  Descriptive Statistics and Correlations.
M SD Minimum Maximum (1) (2) (3) (4) (5) (6) (7) (8) (9) (10) (11) (12)

(1) UN homicide rate (2005-2008) 8.04 11.24 0.67 44.10 1.00  


(2) Inequality Scale ((1 − Max)3 − 1) −288.3 94.1 −485.9 −61.1 .60** 1.00  
(3) Development index 0.13 0.71 −1 1.81 −.42* −.36* 1.00  
(4) Sex ratio (females:males) 50.63 1.17 48.12 53.85 .05 .07 .29 1.00  
(5) Materialism3 (continuous) 9.13 3.78 1.14 15.97 −.27 .11 .49** .19 1.00  
(6) Demoralization (continuous) −0.08 0.84 −1.76 1.55 .31 .35* .00 .14 −.02 1.00  
(7) Deinstitutionalization (continuous) −0.10 0.80 −1.59 2.02 .08 .16 −.72** −.36 −.24 −.35 1.00  
(8) Stable type (demoralization-based) 0.40 0.50 0 1 −.18 −.12 .23 −.14 .11 −.12 −.11 1.00  
(9) Stable type (deinstitutionalization- 0.49 0.51 0 1 .19 .40* −.03 .13 .11 .46* −.09 .26 1.00  
based)
(10) Demoralized type 0.29 0.46 0 1 .35* .18 −.27 .14 −.39* .70** −.06 −.52** .14 1.00  
(11) Deinstitutionalized type 0.20 0.41 0 1 .03 −.34* .07 .10 −.49** −.01 −.35* −.12 −.49** .32 1.00  
(12) Ritualist type (demoralization-based) 0.26 0.44 0 1 −.13 −.02 .16 .12 .53** −.39* .09 −.48* −.31 −.37* −.29 1.00
(13) Ritualist type (deinstitutionalization- 0.26 0.44 0 1 −.17 −.10 −.04 −.18 .32 −.37* .42* −.08 −.57** −.37* −.29 .55*
based)
Note. UN = United Nations.
Schaible and Altheimer 951

significant variable in the bivariate correlations. Only the deinstitutionalized societal


type was significantly correlated with the UN homicide rate. The bivariate correlations
with the two indicators of “stable” societies were in opposite directions but not signifi-
cant. Both indicators of “ritualist” (demoralization- and deinstitutionalization-based)
societies were correlated with homicide but non-significant. Both development and
deinstitutionalization were significantly negatively correlated with the inequality
index. Dummies for societal type were significantly negatively correlated with one
another. As such, it was necessary to enter the type of dummies separately in subse-
quent equations to avoid multicollinearity. No variance inflation factor (VIF) test
exceeded three for any of the regression equations.5
Further analyses were conducted to select the appropriate statistical procedure.
Likelihood ratio tests of the assumption that α = 0 were all beyond p < .001, indicating
that over-dispersion was present, and that negative binomial models were the most
appropriate for addressing underlying data issues. Similarly, Akaike information crite-
rion (AIC) levels were low comparable with alternative models (results not shown)
and indicate that the model fit was superior for these data. As further support for our
findings, results from OLS with Huber–White correction of standard errors, over-
dispersed Poisson models, and negative binomial results were all nearly identical and
led to the same substantive conclusions.
In the models presented in Table 2, each societal type is entered separately with all
controls. Control variables exert effects on homicide in the expected directions across
all models. Consistent with past literature (Altheimer, 2007, 2008; Bennett, 1991; P.
Blau, 1977; J. Blau & Blau, 1982; Chamlin & Cochran, 2005, 2006; LaFree, 1999;
Messner, 1982, 1986; Neapolitan, 1997; Nivette, 2011; Pratt & Cullen, 2005;
Savolainen, 2000), higher inequality consistently predicts higher homicide rates when
controlling for other factors; similarly, higher levels of development correspond with
lower rates of homicide across societies. Consistent with the counter-intuitive finding
of past literature (Antonoccio & Tittle, 2007; Messner & Sampson, 1991, Schaible &
Hughes, 2011; South & Messner, 1987), sex ratio is a strong and consistent predictor
of levels of homicide with a higher ratio of females to males predicting higher levels
of homicide.
The effect of societal types is consistent with the effects hypothesized by Merton.
Specifically, in support of H1, those societies with the most balanced integration of
means and ends—the stable types—benefit from lower levels of homicide when con-
trolling for other known correlates. However, only the demoralization-based indicator
of “stable” type societies emerges as significant. Similarly, in support of H2, homicide
rates are significantly higher in societies that have high materialism paired with either
excessive levels of demoralization or deinstitutionalization. Finally, in support of H3,
neither typology for ritualist societies exhibits a significant relationship with homicide
rates. Taken together, the findings are consistent with Merton’s notion that imbalance
in the social and cultural structure need not necessarily result in higher homicide rates,
but only tends to do so when there is an imbalance in which either means are de-
emphasized or materialism excessive.
Table 2.  Negative Binomial Regression for Independent Effects.

952
M1 M2 M3 M4 M5 M6 M7

  coef/se coef/se coef/se coef/se coef/se coef/se coef/se


Inequality Scale ((1 − Max)3 − 1) 0.008*** 0.008*** 0.008*** 0.008*** 0.009*** 0.008*** 0.008***
(0.001) (0.001) (0.001) (0.001) (0.001) (0.001) (0.001)
Development index −1.266*** −1.234*** −1.239*** −1.227*** −1.186*** −1.264*** −1.259***
(0.237) (0.242) (0.239) (0.240) (0.232) (0.237) (0.234)
Sex ratio (females:males) 0.229*** 0.181** 0.235*** 0.190*** 0.202*** 0.230*** 0.222***
(0.103) (0.098) (0.101) (0.096) (0.099) (0.104) (0.104)
Stable type (demoralization-based) −0.428**  
  (0.228)  
Stable type (deinstitutionalization-based) −0.243  
  (0.257)  
Demoralized type 0.433***  
  (0.201)  
Deinstitutionalized type 0.624***  
  (0.275)  
Ritualist type (demoralization-based) −0.019  
  (0.259)  
Ritualist type (deinstitutionalization-based) −0.166
  (0.263)
Intercept −7.790* −5.043 −7.739* −5.978 −6.248 −7.834* −7.430
(5.251) (5.049) (5.173) (4.863) (5.057) (5.289) (5.307)
/lnalpha −1.712*** −2.063*** −1.775*** −2.064*** −1.934*** −1.708*** −1.702***
(0.505) (0.621) (0.526) (0.615) (0.558) (0.506) (0.498)
AIC 184.590 183.451 185.707 182.363 181.468 186.584 186.191
Observations (n) 35 35 35 35 35 35 35

Note. AIC = Akaike information criterion.


*p < .15. **p < .10. ***p < .05.
Schaible and Altheimer 953

In Table 3, we examine support for H4, or the degree to which cultural patterns
multiplicatively interact with inequality in the opportunity structure. The patterns of
relation between control variables and homicide rates remain consistent with the pre-
vious models and past research. Similarly, the independent effects of cultural patterns
are largely unchanged by the entry of interaction variables (although independent
effects of deinstitutionalization are no longer significant). However, contrary to H4,
deinstitutionalization and inequality interact to have a significant negative effect on
homicide rates, rather than the expected positive effect. Taken together, these findings
are contrary to Merton’s propositions, but consistent with past findings examining
lower units of analysis (Baumer & Gustafson, 2007). This suggests that the effects of
inequality are largely independent of the means/ends integration. In other words, the
effects of inequality do not vary significantly by the cultural patterns of means/ends
mal-integration. This is consistent with past literature, which has consistently failed to
provide clear support for the existence of relationship between inequality and anomic
conditions at the macro-level.

Discussion
This study examined Merton’s anomie theory in a cross-national sample of nations.
Four hypotheses were tested on the link between the patterns of means/end integration
discussed by Merton and homicide. The findings were generally supportive of Merton’s
propositions. In partial support of H1, stable societies benefit from lower levels of
homicide with both typologies for “stable” possessing a negative valence but only the
demoralization-based stable type achieving significance. In support of H2, deinstitu-
tionalized and demoralized societies were found to have significantly higher rates of
homicide. H3 was supported by the absence of a significant relationship between ritu-
alist types and homicide. However, in clear contradiction to H4, interactions between
inequality and cultural type did not support the existence of interactive effects. We
believe that these findings contribute to the advancement of classical anomie and
cross-national criminology and have implications for research in this area.
The findings from this exploratory study, when considered with those recently
reported by Baumer and Gustafson (2007), suggest that Merton’s theory merits further
examination in the criminological literature. Although revisions of the theory have
garnered attention in recent years, the macro-level propositions of classical anomie
theory have not received adequate attention. Our results suggest that Merton’s theory
still holds promise for expanding knowledge about cross-national variation in crime
and should not yet be thrown in the criminological dustbin.
Particularly noteworthy was strong support for Merton’s proposition that homicide
is higher in deinstitutionalized and demoralized societies and lower in stable societies,
with ritualist societies exhibiting no clear relationship to levels of homicide. This pro-
vides credence to the notion that cultural patterns have direct implications for societal
levels of homicide and raises questions about what other cultural patterns might be
important for explaining cross-national variation in homicide. Our typology of cultural
patterns explained homicide as well as other well-established correlates of violence.
Table 3.  Negative Binomial Regressions for Interaction Effects.

954
M1 M2 M3 M4 M5 M6

  coef/se coef/se coef/se coef/se coef/se coef/se


Inequality Scale ((1 − Max)3 − 1) 0.008*** 0.008*** 0.007*** 0.007*** 0.007*** 0.008***
(0.001) (0.001) (0.002) (0.002) (0.001) (0.002)
Development index −1.236*** −1.229*** −1.240*** −1.223*** −1.449*** −1.276***
(0.244) (0.239) (0.233) (0.221) (0.246) (0.246)
Sex ratio (females:males) 0.181** 0.193*** 0.244*** 0.256*** 0.231*** 0.220***
(0.099) (0.097) (0.103) (0.104) (0.096) (0.104)
Stable type (demoralization-based) −0.416*  
(0.275)  
Stable (demoralization-based) × Inequality −0.000  
(0.002)  
Demoralized type 0.397**  
  (0.228)  
Demoralized type × Inequality 0.001  
  (0.002)  
Ritualist type 0.016  
  (0.258)  
Ritualist type × Inequality −0.004  
  (0.004)  
Stable type (deinstitutionalization-based) −0.280  
  (0.262)  
Stable type (deinstitutionalization-based) 0.005  
  (0.004)  
Deinstitutionalized type 0.359  
  (0.281)  

(continued)
Table 3.  (continued)

M1 M2 M3 M4 M5 M6

  coef/se coef/se coef/se coef/se coef/se coef/se


Deinstitutionalized type × Inequality −0.012***  
  (0.005)  
Ritualist type (deinstitutionalization-based) −0.167
  (0.263)
Ritualist type (demoralization-based) × Inequality 0.001
  (0.005)
Intercept −7.650* −8.215** −10.744*** −11.430*** −10.244*** −9.545**
(4.997) (4.910) (5.221) (5.284) (4.905) (5.292)
/lnalpha −2.053*** −2.034*** −1.767*** −1.738*** −2.158*** −1.710***
(0.632) (0.609) (0.519) (0.498) (0.591) (0.502)
AIC 185.444 184.236 187.818 185.804 179.325 188.134
Observations (n) 35 35 35 35 35 35

Note. All variables were centered prior to creation of interaction terms; AIC = Akaike information criterion; se = standard error; coef = coefficient.
*p < .15. **p < .10. ***p < .05.

955
956 International Journal of Offender Therapy and Comparative Criminology 60(8)

This suggests that both cultural and structural factors are important in explaining macro-
variation in homicide, especially at the societal level. Earlier cross-national research has
been particularly effective at identifying the structural predictors of homicide, but less is
known about cultural predictors or their intersections. This may be due to underdevel-
oped theoretical articulations of the cultural process that influence homicide rates, as
well as the fact that it has been difficult to operationalize the important predictors of
cultural processes at the cross-national level. Results of this study suggest that research-
ers can continue to look to classical anomie theory for guidance on understanding the
link between macro-level processes and homicide. Furthermore, the development of the
WVS and other data sets is making the examination of such processes more feasible.
Greater consideration of cultural processes will provide a more complete understanding
of the relationship between social organization and violence (Karstedt, 2001).
The findings also provide support for the methodological approach taken in this
study. The small samples, commonly used in cross-national criminological research,
often limit degrees of freedom and the ability of researchers to test for complex statis-
tical interactions. Similarly, they necessitate special statistical treatment. This neces-
sitates innovative approaches to examining such relationships. One way to get around
this problem—and the approach taken here—is to create typologies that combine com-
plex theoretical constructs while also increasing degrees of freedom. The increasing
statistical sophistication of the field may cause some to view simpler approaches sus-
piciously, but making greater use of typologies provides a practical solution to the
small sample dilemma that researchers often face in cross-national analyses.
Although deinstitutionalization directly affected homicide, interactions with inequal-
ity in the opportunity structure did not emerge as significant, as predicted by classical
anomie. These findings mirror those recently reported by Baumer and Gustafson (2007)
and lead to questions concerning why no such interactions were found. Two explana-
tions are provided here. First, it is possible that the likelihood of an interaction between
deinstitutionalization and blocked opportunity is contingent on the nature of the blocked
opportunity. Merton (1938) suggests the manner that that social structure predisposes
toward crime in deinstitutionalized societies is contingent on two factors:

First, such antisocial behavior is in a sense “called forth” by certain conventional values
of the culture and by the class structure involving differential access to the approved
opportunities for legitimate, prestige-bearing pursuit of the culture goals . . . The second
consideration is of equal significance. Recourse to the first of the alternative responses,
legitimate effort, is limited by the fact that actual advance toward desired success-
symbols through conventional channels is, despite our persisting open-class ideology,
relatively rare and difficult for those handicapped by little formal education and few
economic resources. (p. 679)

This statement suggests that high levels of economic inequality may not elevate
rates of crime in deinstitutionalized societies if there is some degree of opportunity to
advance toward prestige-bearing culturally defined success symbols. This may also
suggest that measures of economic inequality used here, as well as in most cross-
national criminological research, do not adequately account for the type of blocked
Schaible and Altheimer 957

opportunity noted by Merton. The second possible explanation for why no interaction
was found is that deinstitutionalization simply does not interact with blocked opportu-
nity to influence rates of homicide in the manner proposed by Merton. If this finding
continues to emerge in research on classical anomie theory, then it may be necessary
for the theory to be refined.

Limitations
Despite the contribution that this study provides to the criminological literature, several
important limitations should be noted. First, this study examined total rates of homicide
as the dependent variable. Merton’s original propositions, however, specifically linked
anomie to instrumental crimes. Future research on this topic should examine more pre-
cise indicators of homicide that are disaggregated by motive, as well as other forms of
instrumental crime. Second, it is plausible that the degree to which crime is reported
across societies is a function of structural and cultural processes. Thus, the approach to
crime reporting may be different in demoralized or deinstitutionalized societies when
compared with more stable societies. In light of this fact, future research on anomie
should examine crime data from self-report surveys and other unofficial sources data.
Third, as noted above, more sophisticated measures of social stratification may be nec-
essary to tease out potential interactions between blocked opportunity and deinstitu-
tionalization. Future research should use alternative indicators of blocked opportunity
that better account for societal levels of social stratification. Fourth, the number of
nations examined, although similar to previous cross-national studies, was small. This
inevitably limits the ability to conduct statistical tests and risks failing to detect real
effects because of weak power. Although we did find significant effects, future research
should look to examine Merton’s propositions in a larger sample of nations. Finally, the
data from WVS utilized to operationalize cultural features of societies were only avail-
able for one time span. As such, time-series analysis was precluded. Future analyses
should seek to examine how transitions in societal types over time have a differential
influence on levels of homicide and other types of crime.

Conclusion
Notwithstanding these limitations, we believe that our study makes an important con-
tribution to the extant cross-national criminological literature. To our knowledge, this
study represents the first attempt to test the cross-national propositions of classical
anomie theory. Undoubtedly, Merton was a scholar with profound insights and much
to offer in understanding macro-level processes in crime. Unfortunately, these insights
have remained relatively unexplored. Although the present study tests some of these
insights, many questions remain to be answered, especially about the relationship
between cultural patterns identified by Merton, their interaction with opportunity
structures, and consequences for crime and other social problems. We hope that in
addressing a number of important questions here, the present study will serve as base
and fount for such an exploration.
958
Appendix
UN Homicide Inequality Sex ratio Stable type Stable type Ritualist type Ritualist type
Rate index ((1 − (females: Development (demoralization- (deinstitutionalization- Demoralized Deinstitutionalized (demoralization- (deinstitutionalization-
(2005-2008) Max)3 − 1) males) index based) based) type type based) based)

Albania 3.300 −344.029 50.559 −0.246 0 0 1 1 0 0


Argentina 5.475 −162.279 50.957 0.518 1 1 0 0 0 0
Bangladesh 2.750 −326.464 49.410 −0.796 0 0 0 0 0 0
Bosnia and 1.800 −295.977 51.887 0.192 0 1 0 0 1 0
Herzegovina
Brazil 22.550 −95.582 50.705 0.250 0 1 1 0 0 0
Bulgaria 2.400 −340.887 51.544 0.517 0 0 1 1 0 0
Chile 3.575 −210.618 50.537 0.451 0 0 0 0 1 1
China 1.325 −269.359 48.124 −0.154 1 1 0 0 0 0
Finland 2.375 −428.039 51.025 1.350 1 0 0 1 0 0
Georgia 7.450 −246.677 52.848 −0.157 0 1 0 0 1 0
Germany 0.975 −418.874 51.051 1.344 1 1 0 0 0 0
Guatemala 44.100 −140.195 51.240 −0.472 0 1 1 0 0 0
India 3.450 −351.094 48.283 −0.766 0 1 1 0 0 0
Indonesia 8.100 −281.630 50.055 −0.423 0 0 0 1 0 0
Jordan 1.550 −335.995 48.621 −0.128 1 0 0 0 0 1
Kyrgyzstan 8.650 −248.272 50.676 −0.472 1 1 0 0 0 0
Mexico 9.950 −169.597 50.698 0.208 0 1 1 0 0 0
Republic of 6.850 −283.585 52.458 −0.158 0 0 1 1 0 0
Moldova
Morocco 1.550 −306.871 50.829 −0.452 1 0 0 0 0 1
Norway 0.675 −375.010 50.364 1.805 0 0 0 0 1 1
Peru 11.125 −165.278 49.865 −0.073 1 1 0 0 0 0
Philippines 6.950 −279.049 49.620 −0.356 0 1 1 0 0 0
Poland 1.350 −310.121 51.698 0.495 1 1 0 0 0 0
Vietnam 1.600 −331.685 50.741 −0.372 1 0 0 0 0 1
Slovenia 0.825 −425.459 51.194 0.945 1 1 0 0 0 0
South Africa 38.300 −61.075 50.750 −0.156 1 1 0 0 0 0

(continued)
Appendix (continued)
UN Homicide Inequality Sex ratio Stable type Stable type Ritualist type Ritualist type
Rate index ((1 − (females: Development (demoralization- (deinstitutionalization- Demoralized Deinstitutionalized (demoralization- (deinstitutionalization-
3
(2005-2008) Max) − 1) males) index based) based) type type based) based)

Spain 1.075 −331.634 50.757 1.190 0 0 0 0 1 0


Sweden 1.000 −485.845 50.426 1.572 1 0 0 1 0 0
Turkey 4.100 −289.313 49.736 0.161 0 0 0 0 1 1
Uganda 36.300 −293.734 49.963 −1.004 0 0 1 1 0 0
Ukraine 6.550 −324.399 53.848 0.372 0 1 1 0 0 0
Macedonia, TFYR 2.250 −280.596 50.037 0.191 1 1 0 0 0 0
Egypt 0.750 −358.883 49.692 −0.296 0 0 0 0 1 1
United Republic 24.500 −320.055 50.209 −0.939 0 0 0 0 1 1
of Tanzania
Uruguay 6.050 −200.825 51.747 0.556 0 0 0 0 1 1

Note. UN = United Nations; TFYR = The Former Yugoslavian Republic.

959
960 International Journal of Offender Therapy and Comparative Criminology 60(8)

Declaration of Conflicting Interests


The author(s) declared no potential conflicts of interest with respect to the research, authorship,
and/or publication of this article.

Funding
The author(s) received no financial support for the research, authorship, and/or publication of
this article.

Notes
1. These figures are based on the authors’ calculations using data obtained from the WVS
Association (http://www.worldvaluessurvey.org/).
2. A master questionnaire was prepared in English prior to translation into the national lan-
guage of respondents. To check the accuracy of translation and identify potential problems,
the translated version of the questionnaire often was translated back into English by an
independent party and pre-tested with pilot samples (World Values Survey, 2009).
3. Several nations exhibited patterns that were not consistent with Merton’s three identified
types, but logically possible (e.g., Spain had high anomie, but low materialism, whereas
Bangladesh and Indonesia had high materialism, but low anomie and deinstitutionaliza-
tion). These three exceptions were coded as 0 in all analyses because they did not conform
to the three specific types discussed by Merton.
4. Although we presented results using negative binomial regression, it is noteworthy that
results were nearly identical when any of the following methods were applied: ordinary
least squares (OLS) with Heteroskedasticity-Consistent Estimator 3 (HC3) correction of
standard errors, over-dispersed Poisson regression with either deviance or chi-square, and
negative binomial regression.
5. The conservative threshold for VIF is considered to be four; however, a number of stud-
ies have suggested that regression is relatively robust to multicollinearity (see especially
O’Brien, 2007).

References
Altheimer, I. (2007). Assessing the relevance of ethnic heterogeneity as a predictor of homicide
at the cross-national level. International Journal of Comparative and Applied Criminal
Justice, 31, 1-20.
Altheimer, I. (2008). Social support, ethnic heterogeneity, and homicide: A cross-national
approach. Journal of Criminal Justice, 36, 103-114.
Antonoccio, O., & Tittle, C. (2007). A cross-national test of Bonger’s theory of criminality and
economic conditions. Criminology, 45, 925-958.
Barber, N. (2000). The sex ratio as a predictor of cross-national variation in violent crime.
Cross-Cultural Research, 34(3), 264-282.
Baumer, E. (2007). Untangling research puzzles in Merton’s multilevel anomie theory.
Theoretical Criminology, 11, 63-93.
Baumer, E., & Gustafson, R. (2007). Social organization and instrumental crime: Assessing
the empirical validity of classic and contemporary anomie theories. Criminology, 45,
617-663.
Bennett, R. (1991). Development and crime: A cross-national, time-series analysis of compet-
ing models. The Sociological Quarterly, 32, 343-363.
Schaible and Altheimer 961

Bennett, R., & Basiotis, P. (1991). Structural correlates of juvenile property crime: A cross-
national, time-series analysis. Journal of Research in Crime & Delinquency, 28, 262-287.
Bernard, T. (1987). Testing structural strain theories. Journal of Research in Crime &
Delinquency, 24, 262-280.
Bjerregaard, B., & Cochran, J. (2008a). Want amid plenty: Developing and testing a cross-
national measure of anomie. International Journal of Conflict and Violence, 2, 182-193.
Bjerregaard, B., & Cochran, J. (2008b). A cross-national test of institutional anomie theory: Do
the strength of other social institutions mediate or moderate the effects of the economy on
the rate of crime? Western Criminology Review, 9(1), 31-48.
Blau, J., & Blau, P. (1982). The cost of inequality: Metropolitan structure and violent crime.
American Sociological Review, 114-129.
Blau, P. (1977). Inequality and heterogeneity. New York, NY: Free Press.
Bonger, W. (1916). Criminality and economic conditions (H. P. Horton, Trans.). Boston, MA:
Little, Brown.
Cao, L. (2004). Is American society more anomic? A test of Merton’s theory with cross-national
data. International Journal of Comparative and Applied Criminal Justice, 28, 15-31.
Cao, L. (2007). Returning to normality: Anomie and crime in China. International Journal of
Offender Therapy and Comparative Criminology, 51, 40-51.
Chamlin, M., & Cochran, J. (2005). Ascribed economic inequality and homicide among modern
societies: Toward the development of a cross-national theory. Homicide Studies, 9, 3-29.
Chamlin, M., & Cochran, J. (2006). Economic inequality, legitimacy, and cross-national homi-
cide rates. Homicide Studies, 10, 231-252.
Chamlin, M., & Sanders, B. (2013). Falsifying Merton’s macro-level anomie theory of profit-
motivated crime: A research note. Deviant Behavior, 34, 961-972.
Cochran, J., & Bjerregaard, B. (2012). Structural anomie and crime: A cross-national test.
International Journal of Offender Therapy and Comparative Criminology, 56, 203-217.
D’Unger, A., Land, K., McCall, P., & Nagin, D. (1998). How many latent classes of delinquent/
criminal careers? Results from mixed Poisson regression analyses. American Journal of
Sociology, 103, 1593-1630.
Gardner, W., Mulvey, E., & Shaw, E. (1995). Regression analyses of counts and rates: Poisson,
overdispersed Poisson, and negative binomial models. Psychological Bulletin, 118,
392-404.
Gartner, R. (1990). The victims of homicide: A temporal and cross-national comparison.
American Sociological Review, 55, 92-106.
Gartner, R., & Parker, R. (1990). Cross-national evidence on homicide and age structure of the
population. Social Forces, 69, 351-371.
Guttentag, M., & Secord, P. (1983). Too many women? The sex ratio question. Beverly Hills,
CA: Sage.
Hilbe, J. (2011). Negative binomial regression. Cambridge, UK: Cambridge University Press.
Inglehart, R., & Baker, W. (2000). Modernization, cultural change, and the presence of tradi-
tional values. American Sociological Review, 65, 19-51.
Karstedt, S. (2001). Comparing cultures, comparing crime: Challenges, prospects and problems
for a global criminology. Crime, Law and Social Change, 36, 285-308.
Kim, S., & Pridemore, W. (2005). Social change, institutional anomie and serious property
crime in transitional Russia. British Journal of Criminology, 45, 81-97.
Kornhauser, R. (1978). Social sources of delinquency. Chicago, IL: University of Chicago Press.
Krohn, M. (1976). Inequality, unemployment and crime. The Sociological Quarterly, 17,
303-313.
962 International Journal of Offender Therapy and Comparative Criminology 60(8)

LaFree, G. (1999). A summary and review of cross-national comparative studies of homicide.


In M. Smith, & M. Zahn (Eds.), Homicide: A sourcebook of social research. Thousand
Oaks, CA: Sage.
LaFree, G., & Drass, K. (2002). Counting crime booms among nations: Evidence for homicide
victimization rates, 1956 to 1998. Criminology, 40, 769-800.
Land, K., McCall, P., & Nagin, D. (1996). A comparison of Poisson, negative binomial, and
semiparametric mixed Poisson regression models with empirical applications to criminal
careers data. Sociological Methods & Research, 24, 387-442.
Lee, M., & Ousey, G. (2001). Size matters. The Sociological Quarterly, 42, 581-602.
Liu, J. (2005). Crime patterns during the market transition in China. British Journal of
Criminology, 45, 613-633.
Long, J. (1997). Advanced Quantitative Techniques in the Social Sciences: Vol. 7: Regression
models for categorical and limited dependent variables. Thousand Oaks, CA: Sage.
Maume, M., & Lee, M. (2003). Social institutions and violence: A sub-national test of institu-
tional anomie theory. Criminology, 41, 1137-1172.
McCullagh, P., & Nelder, J. (1989). Generalized linear models (2nd ed.). London, England:
Chapman and Hall.
Merton, R. (1938). Social structure and anomie. American Sociological Review, 3, 672-682.
Merton, R. (1949). Social theory and social structure. Glencoe, IL: The Free Press.
Merton, R. (1957). Social theory and social structure (Rev. enlarged ed.). New York, NY: The
Free Press.
Merton, R. (1964). Anomie, anomia, and social interaction: Contexts of deviant behavior. In M.
Clinard (Ed.), Anomie and deviant behavior (pp. 213-242). New York, NY: The Free Press.
Merton, R. (1968). Social theory and social structure (Rev. enlarged ed.). New York, NY: The
Free Press.
Messner, S. (1982). Poverty, inequality, and the urban homicide rate. Criminology, 20, 103-115.
Messner, S. (1986). Modernization, structural characteristics, and societal rates of crime: An
application of Blau’s macrosociological theory. The Sociological Quarterly, 27, 27-41.
Messner, S. (1988). Merton’s social structure and anomie: The road not taken. Deviant Behavior,
9, 33-53.
Messner, S., & Rosenfeld, R. (1994). Crime and the American dream. Belmont, CA: Wadsworth.
Messner, S., & Rosenfeld, R. (1997). Political restraint, market and levels of criminal homicide:
A cross-national application of institutional-anomie theory. Social Forces, 75, 1393-1416.
Messner, S., & Rosenfeld, R. (2007). Crime and the American dream (4th ed.). Belmont, CA:
Wadsworth.
Messner, S., & Sampson, R. (1991). The sex ratio, family disruption, and rates of violent crime:
The paradox of the demographic structure. Social Forces, 69, 693-713.
Messner, S., Thome, H., & Rosenfeld, R. (2008). Institutions, anomie, and violent crime:
Clarifying and elaborating institutional-anomie theory. International Journal of Conflict
and Violence, 2, 163-181.
Neapolitan, J. (1997). Cross-national crime: A research review and sourcebook. Westport, CO:
Greenwood Press.
Nivette, A. (2011). Cross-national predictors of crime: A meta-analysis. Homicide Studies, 15,
103-131.
O’Brien, R. (2007). A caution regarding rules of thumb for variance inflation factors. Quality
& Quantity, 41(5), 673-690.
Osgood, D. (2000). Poisson-based regression analysis of aggregate crime rates. Journal of
Quantitative Criminology, 16, 21-43.
Schaible and Altheimer 963

Osgood, D., & Chambers, J. (2000). Social disorganization outside the metropolis: An analysis
of rural youth violence. Criminology, 38, 81-116.
Pampel, F., & Gartner, R. (1995). Age structure, socio-political institutions, and national homi-
cide rates. European Sociological Review, 11, 243-260.
Pratt, T., & Cullen, F. (2005). Assessing the macro-level predictors and theories of crime: A
meta analysis. In M. Tonry (Ed.), Crime and justice: A review of research (vol. 32, pp. 373-
450). Chicago, IL: University of Chicago Press.
Pratt, T., & Godsey, T. (2003). Social support, inequality, and homicide: A cross-national test
of an integrated theoretical model. Criminology, 41, 611-643.
Pridemore, W. (2008). A methodological addition to the cross-national empirical literature on
social structure and homicide: A first test of the poverty-homicide thesis. Criminology, 46,
133-154.
Pridemore, W. (2011). Poverty matters: A reassessment of the inequality-homicide relationship
in cross-national studies. The British Journal of Criminology, 51, 739-772.
Savolainen, J. (2000). Inequality, welfare state, and homicide: Further support for the institu-
tional anomie theory. Criminology, 38, 1021-1042.
Schacht, R., Rauch, K., & Borgerhoff Mulder, M. (2014). Too many men: The violence prob-
lem? Trends in Ecology & Evolution, 29, 214-222.
Schaible, L., & Hughes, L. (2011). Crime, shame, reintegration and cross-national homicide:
A partial test of reintegrative shaming theory. The Sociological Quarterly, 52, 104-131.
South, S., & Messner, S. (1987). The sex ratio and women’s involvement in crime: A cross-
national analysis. The Sociological Quarterly, 28, 171-188.
Stamatel, J. (2009). Correlates of national-level homicide variation in post-communist East-
Central Europe. Social Forces, 87, 1423-1448.
Stults, B., & Baumer, P. (2008). Assessing the relevance of anomie theory for explaining spatial
variation in lethal criminal violence: An aggregate-level analysis of homicide within the
United States. International Journal of Conflict and Violence, 2, 215-247.
Thorlindsson, T., & Bernburg, J. (2004). Durkheim’s theory of social order and deviance: A
multi-level test. European Sociological Review, 20, 271-285.
Tukey, J. (1977). Exploratory data analysis. Reading, MA: Addison-Wesley
Wooldridge, J. (1997). Quasi-likelihood methods for count data. Handbook of Applied
Econometrics, 2, 352-406.
World Values Survey. (2009). World Value Survey 1981-2008 official aggregate (World Values
Survey Association, v.20090902). Madrid, Spain: ASEP/JDS Data Archive [Aggregate
File Producer]. Available from www.worldvaluessurvey.org
Zhao, L. (2008). Anomie theory and crime in transitional China. International Criminal Justice
Review, 18, 137-157.
Zhao, R., & Cao, L. (2010). Social change and anomie: A cross-national study. Social Forces,
88, 1209-1230.

Vous aimerez peut-être aussi