RESEARCH QUESTIONS, METHODOLOGY, DATA

RESEARCH QUESTIONS, METHODOLOGY, DATA, AND

ANALYTIC STRATEGY

Research Questions

The research questions central to this research involved the application of the

theoretical framework of Institutional Anomie Theory to the criminal phenomenon of

organized crime in order to gain further knowledge of the socio-cultural factors affecting

organized crime development, levels, and movements across and within developed and

transitioning countries in Europe. As this was an exploratory endeavor designed to

improve the measurement of key elements in the theory, investigate the applicability of

Institutional Anomie Theory in different settings, and employ a new measure of serious

crime, the research questions did not have associated hypotheses. Each question is

derived from the literatures reviewed.

First Set of Research Questions:

The first research questions, utilizing measurements of both ÒAnomic cultureÓ

and Òsocial institutions,Ó addressed how the theory operated in the fourteen countries

in Europe, but also how it operated between and within six country-clusters over time.

As such:

RQ 1: The first research question considered whether countries in Europe with higher

rates of Institutional Anomie also have correspondingly higher rates of

organized crime activity.

RQ 1a: This research also sought to assess whether variations in levels of Institutional

Anomie over time correspondingly impacted levels of organized crime activity in

European countries.

RQ 1b: This research also assessed whether grouping countries into country-clusters

significantly impacts Institutional Anomie Theory’s ability to predict high levels

of organized crime.

While there are no associated hypotheses with this exploratory study, based on past

successful endeavors (i.e., Cullen, Parboteeah, and Hoegel, 2004), Institutional Anomie

Theory may be expected to be able to predict high rates of serious crime while

considering Anomic culture as a separate element from social institutions. However, it

must be noted that the two empirical studies that did attempt to measure ÒAnomic

cultureÓ did so only partially; they did not include a measurement representing a lack

of legitimate means. As such, it is unclear how the added operationalization of this

additional Anomic cultural element will impact the study.

It is unclear how considering changes over time will impact the theoryÕs

explanatory ability. While Messner and Rosenfeld (2009) encourage future research to

include this element, others argue that ÒculturesÓ change too slowly over time to

capture any significant changes. However, proponents of Institutional Anomie argue

that full-scale changes in culture and institutions are not needed to impact levels of

serious crime; changes in some elements of institutions and/or culture are enough to

effectively influence levels of Institutional Anomie and crime rates (Messner and

Rosenfeld, 2009; Scott, 2008).

It is also unclear how considering country-cluster differences will impact

Institutional Anomie Theory. Messner and Rosenfeld (1994) would argue that the theory

should work best in the country clusters containing developed countries; as such, this

research has similar expectations. This study will also be able to further discern whether

or not it is appropriate to lump or combine data from large samples of fifty

countries, assuming the theory operates the same in each.

The impacts of using a new measurement of Òserious crimeÓ on the explanatory

power of Institutional Anomie Theory are further unknown. However, what is known

about Òorganized crimeÓ and the documented movements and patterns of this criminal

phenomenon indicate that it may be well suited for testing using a socio-cultural theory

such as Institutional Anomie. Also, Europe has been identified as a global demand Òhot

spotÓ for illicit goods and services provided by organized crime syndicates, which most

prominently include drugs, and it is also evident that some regions within Europe

experience elevated levels of organized crime activity (Council of Europe, 2005; OCTA,

2011; UNODC, 2010). Therefore, it is reasonable to assume that applying this dependent

variable in this particular geographic region should produce at least preliminary

indications of what Anomic cultural and institutional factors may impact organized crime

in Europe.

Second Research Question:

The second research question further challenges the notion that Institutional

Anomie Theory was designed to operate best in Òmarket capitalist,Ó developed

countries (Messner and Rosenfeld, 2001). As such:

RQ 2: This research: examines how elements of culture and social institutions affect

levels of Institutional Anomie and organized crime in both

developed and transitioning countries.

As aforementioned, countries considered to be ÒtransitioningÓ in this research are those

that experienced the collapse of the USSR in 1989. In this research, this includes the

countries of Poland, Lithuania, and Slovakia. All other countries in this study (see Figure 9,

pg. 150) are included in the developed country group. Based on original arguments by

Messner and Rosenfeld (1994), it would be expected for the theory to operate best in the

developed country group. However, this research question, like RQ 1a, will also be able to

help discern whether lumping data together from fifty or sixty countries is appropriate. If the

theory operates the same in both developed and transitioning country groups, then perhaps

there is no cause for concern in combining large quantities country-data.

Third Set of Research Questions:

The third research question examined the dynamics of Anomic culture, social

structure, and Institutional Anomie in relation to organized crime activity in the

transitioning democratic country of Poland. This research question applied the

measurements of ÒAnomic cultureÓ and Òsocial institutionsÓ employed in the other

research questions to a country that Messner and Rosenfeld (1994) would not necessarily

expect the theory to operate well in, but it sought to address the new measure of

Òserious crimeÓ in a setting that had elevated concerns for this crime-type:

RQ 3: This proposed research seeks to address: how Institutional Anomie Theory

operates in the transitioning democracy of Poland.

RQ 3a: This research also considers: whether the culture-institutional configuration in

the transitioning democracy of Poland varies from other CEE countries.

RQ 3b: If so: how does it impact the levels of organized crime involvement?

As such, it is unclear first how the theory will operate in Poland or in the other CEE

country included in this study (i.e., Slovakia). However, because this region of Europe,

and Poland in particular, has been noted in the literature as experiencing high rates of

organized crime and drug trafficking, it can reasonably be assumed that the elevated

activity would allow for a signal of Institutional Anomie Theory to be detected, thus

allowing for a comparison between Poland and Slovakia. Moreover, as Messner and

Rosenfeld themselves noted in 2001 (along with other cultural and institutional

scholars like Dirk Enzmann), other cultural settings may result in different cultural-

institutional configurations that lead to high rates of crime. This third research question

hopes to add to this body of knowledge by considering which elements of Anomic

culture and social institutions most impact serious crime in Poland.

Methodology

A quantitative research design will be used to address all three sets of research

questions. The design will take an exploratory approach to determine whether countries

in Europe with higher rates of Institutional Anomie also have correspondingly higher

rates of organized crime activity, taking into account changes over time and the grouping

of European countries into clusters and developed vs. transitioning country groups,71

in addition to considering Poland and Slovakia separately.

More specifically, the quantitative research design used to address the three sets

of research questions took a three-step approach. First, the data were addressed through

visual inspection that included graphing the variables to determine the visual story that

the data were telling. Second, no study has considered how the theory operates across

time. Therefore, the years 1995 to 2009 were then empirically examined using

multivariate and pooled cross-sectional time series analysis techniques. These analyses

utilized explanatory variables representing the two elements of social organization:

Anomic culture (i.e., cultural pressures to succeed and a lack of legitimate means) and

social institutions (economic, political, familial, and educational) (see Figure 6 above,

for a list of variables).

The time period for the second stage of the quantitative research design has been

selected for multiple reasons. The time period beginning in 1995 is useful because it will

capture any changes or restructuring of the various elements of social organization

between developed and non-developed nations (due to the collapse of the USSR in 1989),

without risking contaminated or otherwise unreliable data from the years directly

following the collapse of the Soviet Union (Krajewski, 2003). The year 1995 is most

appropriate for these countries representing Central and Eastern Europe (i.e., Poland and

Slovakia) for two reasons. First, Poland did not draft a new constitution following the

collapse of communism until 1997, well after their liberation from the Soviet UnionÕs

sphere of influence. Second, the second country representing the Central and Eastern

European region, Slovakia, was not an independent country until the peaceful splitting of

Czechoslovakia in 1993. Thus, 1995 is an appropriate time period to begin collecting data

from these two countries; data are more reliable than in the years directly following the

collapse of the U.S.S.R., and beginning the time series in 1995 still captures data for the

most transitional stages of social change. Therefore, because of the necessity in capturing

this data for Poland and Slovakia, the same time period will be employed for collection of

data from all countries in this research.72

The third stage of the quantitative research design involved more sophisticated

time series analytic techniques that employed tests for Granger causality and

cointegration for the same time period. These added measures allowed this study to

determine whether there were any additional causal connections existing between the

variables, or if a common element was causing proxies of Anomic culture and social

institutions to move together over time with their dependent variables. These analyses

were important to include, because discovering Granger causal elements between

variables can more strongly support or refute the role the variable might play in

supporting Institutional Anomie Theory in each model. Additionally, finding the

existence of cointegration between pairs of variables has many policy implications, as it

indicates that impacting one variable also has the ability to equally impact the

cointegrated variable. In order to conduct these tests, however, it was necessary to first

run unit root tests to determine the stationarity of the variables before proceeding. The

same time period applies to this analytic stage; however, these techniques could not be

applied to the separate country analyses of Poland and Slovakia due to the lack of

observations. A larger than typical dataset is desirable for these two powerful tests

because of their specific requirements, discussed in greater detail below. However, this

does not take away from other methodological approaches used in this study to address

the operation of Institutional Anomie in these separate countries; it simply means that

this study was unable to address any underlying relationships among the variables in

these countries.

This quantitative approach allowed this exploratory study to address how the

theory operated in Europe as an entire region over time (combining data from all fourteen

countries), as well as within and between the six country clusters (breaking down the data

based on this country-grouping technique discussed below). Additionally, these methods

allowed for a comparison between developed and transitioning countries to uncover

whether any differences in configurations from Western nations were more conducive to

organized crime activity, and whether Institutional Anomie Theory could predict levels of

organized crime in transitioning nations. Finally, these methods were appropriate for

preliminarily investigating how the theory operated in the transitioning nation of Poland

in comparison to its CEE companion, Slovakia.

Dependent Variables and Measures

The dependent variable for this research is Òorganized crime.Ó As was noted in

Chapter 3, both scholars and practitioners alike have struggled to define organized crime.

Studies of this elusive criminal phenomenon have been confined to largely descriptive

accounts of ÒMafia-typeÓ groups (Arlacchi, 1988; Catanazaro, 1988; Dorn, et al., 2005;

La Spina, 2008; Paoli, 2003; 2008), theoretically informed work focusing on aspects of

illicit markets (Reuter and Haaga, 1989) and offender decision-making (Bullock et al.,

2010; Kleemans, et al., 2012; Levi and Maguire, 2004; Von Lampe, 2011), or describing

the characteristics of and differences in hierarchical structures and levels of organization

(Becucci, 2008; Hagan, 1983; Smith, 1975; 1978). As such, there exists little guidance

from criminology in operationalizing Òorganized crimeÓ in empirical studies.

One more recent development, however, is the ÒComposite Organized Crime

IndexÓ developed by Jan Van Dijk (2008). This composite index combines five

interrelated indicators that include the organized crime perception index and Òfour other

indicators of secondary manifestations of organized crime activityÓ (Van Dijk, 2012:

162). The other four are unsolved homicide rates, informal sector rates (i.e., shadow

economies or Òblack marketÓ economies), high-level corruption rates, and money

laundering rates. Van Dijk (2012) notes that unsolved homicide rates represent an

objective measure of organized crime, while the other four measures are based on

perceptional surveys (i.e., how do people perceive these problems of corruption, money

laundering, and informal sectors in their country).

These five indicators combine to form the Composite Organized Crime Index,

which Van Dijk (2012) notes, Òshould not be taken at face value. Scores on the index

and on the five source indicators should be used as a set of diagnostic toolsÓ to help

criminologists Òarrive at evidence-based crime diagnoses of [organized crime in]

individual countriesÓ (162). In other words, this index is useful Òin the search for

metrics on organized crimeÓ at certain points in time (i.e., the index does not consider

changes over time) (Van Dijk, 2012: 162; Van Dijk, 2008).

While the index has indeed proven valuable in providing a more comprehensive

and conceptually whole measure of Òorganized crimeÓ in individual countries and

regions of the world, the index suffers from a lack of data. For instance, Van Dijk (2012)

notes that Òin some cases, the index score is based on just one source, resulting in a

comparatively large margin of errorÓ (164), thus making cross-national comparisons of

index scores difficult. Despite these limitations, regional mean scores on the Composite

Organized Crime Index highlights the Caribbean, Central Asia, and Eastern Europe

as having consistently high scores (Van Dijk, 2012).

One problematic element of this approach is that the Composite Organized Crime

Index indicates that countries to include the United Kingdom, Germany, and Sweden, are

among the fifteen countries with the lowest composite scores on the world-ranking

system. This means that citizens of these countries perceive there to be little corruption or

organized crime operating in money laundering or in an underground economy, as well as

the country having a low unsolved homicide rank. This is problematic, because the

international organizations (e.g., UNODC, Council of Europe) tracking organized crime

activity have indicated that Europe is indeed the worldÕs main destination for illicit

goods and services trafficked by organized crime groups Ð a region of the world that

includes countries identified as low on Van DijkÕs composite score. This discrepancy

between the Composite Organized Crime Index results and evidence of heightened levels

of organized crime activity in Europe may result from failure on both parts to separate

organized crime activity based on Òsource countriesÓ and Òdestination countries.Ó The

Composite Organized Crime Index would therefore be most appropriate in identifying

source countries or regions of the world where organized crime syndicates rely on factors

of political corruption, high unsolved homicide levels, and informal sectors to operate,

cultivate and/or manufacture their illegal goods or services. However, factors such as

corruption are independent of organized crime activity in destination countries.73 Because

73 A good example of this is the case of drug trafficking organizations in North America. Mexico (a country

with a high Index score in van DijkÕs study) is a source country for illegal drugs such as marijuana, cocaine, and

black tar heroin, which are trafficked by organized crime groups into the U.S. (a country with a medium-low

Index score). These organized crime groups rely on high levels of corruption and informal markets in Mexico to

install items like hidden compartments in hundreds of thousands of automobiles in order to smuggle the drugs

into the U.S. This is done because there are not high levels of corruption in the United States that would allow

for such activities to occur within the destination country without alerting

this research examines the destination region of the world for organized crime activity

(UNODC, 2010) (see Figure 17, pg. 291), the Composite Organized Crime Index is not

an appropriate measure to utilize here for organized crime. However, the findings from

this study will be visually compared with the Index to examine any differences

between source and destination countries and/or regions.

In 1999, the Council of Europe (2000) suggested using drug seizure amounts (in

kilograms), namely seizures of heroin, cocaine, amphetamines, and cannabis, as

indicators of organized crime activity. This is largely because while organized criminal

groups were found to be involved in other crimes, the common denominator between

criminal organizations, particularly in Europe, was drug trafficking (Council of Europe,

2000; 2005; OCTA, 2011; UNODC, 2012). The Council of Europe warned against the

reliability of the reported numbers of drug seizures, but also acknowledged that it was

the most reported indicator of organized crime activity among European countries. Little

or no comprehensive data exists for the number of seizures, cases, arrests, or convictions

for vehicle theft, cybercrimes, illegal arms trafficking, or trafficking in human beings.

Minimal research has been done to confirm or refute drug seizures as an empirical

measure of organized criminal activity, but the international community agrees that drug

trafficking remains the common denominator for organized crime.

Additionally, the markets for various drugs are known to vary based on a number

of factors that include price, purity of the drugs, and regional (geographic) demand.

Organized crime syndicates have also been known to vary in the types of drugs they

law enforcement (though some corruption within the U.S. undoubtedly occurs). This is further supported by

the Composite Organized Crime Index itself, which reports that the U.S. has much lower levels of

perceived corruption than Mexico (Van Dijk, 2012). Therefore, high levels of corruption in the United

States are not needed for organized crime groups to infiltrate the country.

traffic. As such, this indicator of drug trafficking representing organized crime is most

appropriate when applied to cross-national or at least regional-level analyses in order to

allow for comparisons in these levels of drug trafficking to be considered.

Additionally, these characteristics of drug trafficking make this indicator suitable to

test against a socio-cultural criminological theory that was originally developed to

explain cross-national rates of crime.

This research will empirically operationalize organized crime through the separate

annual seizure amounts (in kilograms) of heroin, amphetamines, cocaine, and marijuana.

These measures will serve as representatives of organized crime for the three sets of

research questions, including the European country-level, cluster-level, and Polish-level

analyses for the time periods of 1995 to 2009. Even though it is recognized that there are

deficiencies in utilizing this measure of organized crime (e.g., drug seizure

amounts may also be indicative of police interdiction efforts,74 this measure does not

capture the multiple dimensions of organized crime as does the Composite Organized

Crime Index), this is the most appropriate measure for this research because drug

trafficking is such a pervasive element of organized crime groups worldwide, and has

been for decades (Council of Europe, 2000). Moreover, by using seizure amounts of

heroin, amphetamines, cocaine, and marijuana separately, meaning each drug-type

will be a dependent variable, this research will be able to account for variations in the

type and geographic region of demand.

74 Indeed, drug trafficking seizures are in fact indicative of police activity, since the seizure amounts

are reported by the police organizations themselves in each country. However, having personally worked

for the Drug Enforcement Administration in the U.S. for over three years, patterns of drug trafficking

organizationsÕ geographical movements and activities were in fact reflected in the amount of particular

types of drugs seized in each district or regional office over time. In countries such as Poland, Lithuania,

and Slovakia, problems of accurately reporting these seizure amounts becomes the more concerning

limitation Ð see further discussion of this in the concluding chapter of this study.

In order to test the additional proxy measure of serious crime using homicide data,

intentional/completed homicide rates will be measured as the intentional killing of a

person, including murder, manslaughter, euthanasia and infanticide (Eurostat, 2012d). As

has been previously noted in past research, criminological studies have long followed the

tradition of investigating serious forms of macro-level crime trends, particularly utilizing

homicide data (Maume and Lee, 2003; Messner and Rosenfeld, 1994; 1997b; 2009).

Homicide data are commonly used in large part based on the likelihood of an incident

being reported to police (due to the seriousness of the offense), and the continuity of

cross-national legal definitions of ÒhomicideÓ (LaFree and Drass, 2002; Messner and

Rosenfeld, 1994). Intentional (completed) homicides are used in this research as opposed

to ÒtotalÓ homicides, because total homicides rates include uncompleted homicidal

attempts (Eurostat, 2012d). Because ÒhomicideÓ was one form of serious crime that

Messner and Rosenfeld (1994) originally applied to Institutional Anomie Theory, this

research will utilize it as a comparative additional dependent variable to check against the

results of the organized crime dependent variable. This is important, as the findings from

models using this more reliable (and well-tested) measure of serious crime will act as a

gauge for soundness of the untested measures of organized crime. That is, if countries in

Europe with higher rates of Institutional Anomie do in fact experience comparatively

higher rates of both homicide and organized crime activity (meaning, the theory operates

the same for all dependent variables), it can be assumed that the measures of organized

crime are more reliable.

The independent variables to be used in this study are the following two concepts:

Anomic culture and social institutions. This section details each concept, providing support

for the measures chosen by this research to operationalize each construct. This was one of

the more difficult aspects of this study, as the model of Institutional Anomie Theory clearly

identifies culture as existing separately from social institutions, yet Messner and Rosenfeld

(1994) themselves concede that these concepts may be Òempirically inseparableÓ (55). If

they are empirically inseparable, Chamlin and Cochran (2007) point out that Institutional

Anomie Theory is then non-falsifiable, and a non-

falsifiable theory is arguably an un-scientific theory (Kuhn, 1996).75 However, in their

own arguments, Messner and Rosenfeld (2009) maintain that these two dimensions of

social organization must remain distinctive of each other, and they cannot be combined or

singularly emphasized. As such, this research chose to operationalize Anomic culture and

social institutions separate from each other to maintain continuity with the original model

of Institutional Anomie Theory. Multiple measures were used to represent each element

in the model to attempt to capture the multiple dimensions of the American Dream and

social institutions.

Anomic Culture

According to Messner and Rosenfeld (2012), ÒAnomic cultureÓ is composed of

three main elements (see Figure 3, pg. 41). The first element is the pressure to succeed,

measured by an emphasis on achievement, individualism, universalism, and pecuniary

materialism. The second element is the lack of emphasis on legitimate means to succeed.

These two elements combine to produce Anomie.

75 See Kuhn (1996) for further discussion of problems surrounding non-falsifiable theories.

The Anomic cultural element representing Òthe pressure to succeedÓ has proven

the most problematic for empirical studies, in that no criminological studies, save for a

few exceptions (i.e., Cullen, Parboteeah, and Hoegel (2004), Gross and Haussman

(2011)), have been able to empirically distinguish between the four elements (i.e.,

achievement, individualism, universalism, and pecuniary materialism).76 This is most

likely due to the ambiguous descriptions given by Messner and Rosenfeld (2012)

themselves. The below quote illustrates this ambiguity in the elusive description of

Òpressure to succeedÓ given by the authors:

These value commitments generate strong, relentless pressures for everyone to

succeed, understood in terms of an inherently elusive monetary goal. People

accordingly formulate wants and desires that are difficult, if not impossible to

satisfy within the confines of legally permissible behavior. This feature of the

American Dream helps explain criminal behavior with an instrumental character,

behavior that offers monetary rewards. (Messner and Rosenfeld, 2012: 88)

As previously mentioned, in their original conception of Institutional Anomie

Theory, the authors claim that cultural elements may not be empirically discernable from

institutional elements (Messner and Rosenfeld, 1994). One of the studies77 that was able

76 It should be noted that other studies published in non-English journals may have conducted such

studies (e.g., Hirtenlehner et al. 2010; Thome, 2003).

77 Cullen and colleagues (2004) first examined Òachievement,Ó which refers to the levels of

encouragement for members of society to Òmake something of themselvesÓ (Messner and Rosenfeld,

2012: 71). It is an Òassessment of personal worth on the basis of the outcome of effortsÓ (Cullen,

Parboteeah, and Hoegel, 2004). That is, the end goals (i.e., winning or losing) are more important than

Òhow you play the gameÓ (Messner and Rosenfeld, 2001: 63). Thus, the Òcultural pressures to achieve at

any cost are É very intenseÓ (Messner and Rosenfeld, 2012: 72).

Cullen, Parboteeah, and Hoegel (2004) operationalized achievement by creating an index

composed of three indicators: first two used percentages of people in each country who disagreed with

statements such as ÒThe respect a person gets is highly dependent on their family background,Ó while

the third indicator was an item from the World Values Survey represented the percentage of people

surveyed who agreed with the statement, ÒOne does not have the duty to respect and love parents who

have not earned it by their behavior and attitudesÓ (Cullen, Parboteeah and Hoegl, 2004: 415;

Trompenaars and Hampden-Turner, 1998).

Individualism Òencourages disengagement from the collective and, as a consequence, weakens bonds

of social controlÓ (Cullen, Parboteeah, and Hoegel, 2004: 413). ÒIn the pursuit of success, people are

encouraged to Ômake itÕ on their own. Fellow members of society thus become competitors and rivals in the

struggle to achieve social rewards and, ultimately, to validate personal worthÓ (Messner and Rosenfeld, 2012:

72). Cullen, Parboteeah, and Hoegel (2004) operationalized individualism by using three items from

to empirically separate the four cultural elements did not include any inter-item

correlation statistics, which would illustrate whether any correlation between the four

operationalized cultural elements existed. This is an important limitation because

Messner and RosenfeldÕs (1994; 2012) own ambiguity over whether or not cultural

elements can be empirically separated from each other would suggest the potential for

high correlations between any separate measures of Òpressures to succeed.Ó To address

this limitation and to ensure that a measure of this cultural element is present in this

study, this research seeks to employ existing composite measures (which take many

different variables into account, combining them into one ÒscoreÓ) to represent all

aspects of the cultural element of Òpressures to succeed.Ó Therefore, Figure 7 below

represents the adapted model of Institutional Anomie Theory that will be tested.

the World Values Survey that used the percentage of respondents in a nation making the individualism

choice on three issues. The issues and items were:

(1) Quality of life: ÒIt is obvious that if individuals have as much freedom as possible and the

maximum opportunity to develop themselves, the quality of their life will improve as a result;Ó (2)

typical job: ÒEveryone is allowed to work individually and individual credit can be received;Ó and

(3) negligence of a team member: ÒThe person causing the defect by negligence is the

one responsible.Ó (Cullen, Parboteeah, and Hoegel, 2004: 415).

For the national sample, the authors correlated this index with HofstedeÕs (2001) individualism measure.

Universalism Òpromotes equality of opportunity in that it creates expectations that all will be

judged on similar criteria rather than on particularistic relationshipsÓ (Cullen, Parboteeah, and Hoegel, 2004:

413). ÒWith few exceptions, everyone is encouraged to aspire to social ascent, and everyone is susceptible to

evaluation on the basis of individual achievementsÓ (Messner and Rosenfeld, 2012: 72). Similar to the cultural

element of achievement, universalistic cultural pressures may encourage an emphasis on outcomes Òat the

expense of the ethicality [or legality] of achieving these endsÓ (Cullen, Parboteeah, and Hoegel, 2004: 413).

Cullen, Parboteeah, and Hoegel (2004) were measured through two items: ÒOne dealt with testifying truthfully

regarding the driving speed of a friend involved in an accidentÉ The other asked whether a journalist should

write a positive review for a friendÕs restaurantÓ (416).

Pecuniary materialism (originally identified by Messner and Rosenfeld (1994) as fetishism of

money) represents a focus on monetary rewards. ÒTo the degree cultural values promote money as a

valued end independent of other material rewards, the desire for this end becomes insatiableÓ (Cullen,

Parboteeah, and Hoegel, 2004: 413; Messner and Rosenfeld, 2001). Personal monetary success becomes a

metric for comparison Òthat is not linked to group welfareÓ (Cullen, Parboteeah, and Hoegel, 2004: 413).

Thus, Cullen, Parboteeah, and Hoegel (2004) operationalize pecuniary materialism as a series of indicators

from multiple surveys (i.e., WVS and Inglehart (1997)). These items:

Écame from questions asking respondents to prioritize the following goals for their nation:

Òstable economyÓ and Òprogress toward a society where ideas count more than money.Ó To

improve reliability [the authors] added an indicatorÉ, the proportion of people in a nation

choosing Ògood payÓ as an Òimportant job.Ó (Cullen, Parboteeah, and Hoegel, 2004: 416).

Figure 7. Adapted Schema of Institutional Anomie Theory78

The composite index that was used to operationalize Anomic cultural pressures to

succeed is the World Index of Economic Freedom.79 This composite measure embodies

the Anomic cultural pressures to succeed as defined by Messner and Rosenfeld (1994;

2009; 2012);80 in other words, these scholars explicitly state that cultures that reinforce

economic gains and successes are more likely to be Anomic, which impacts serious rates

of crime. That is,

Cultural values that define success or social standing largely in economic terms and

extol the virtues of economic success for all members of society are likely to be

ÒanomicÓ to the extent that corresponding cultural emphasis is not placed on

78 Adapted from Messner and Rosenfeld (2012).

79 The Index of Economic Freedom is a composite measure that is ranked between 0 and 100,

with 0 representing the least amount of economic freedom, and 100 representing maximum

freedom.

80 Figure 9 below illustrates the model of Institutional Anomie with the assigned variables that will

be tested in this dissertation.

the normative status of the means for attaining success, and legitimate means are

distributed unequally across the social structure. (Messner and Rosenfeld, 2009:

214).

In other words, this statement by Messner and Rosenfeld (2009) underlines that the

more problematic Anomic cultural pressures to succeed are those that define and stress

ÒsuccessÓ for its members in economic and utilitarian terms. Cultures that do not

emphasize this definition of success should therefore not experience the same levels of

Anomie and crime. This is why this research utilized the World Index of Economic

Freedom;81 this index embodies the economic cultural values (as explicated by Messner

and Rosenfeld (1994)) that define ÒsuccessÓ in economic, utilitarian terms. This is in

full acknowledgement of past criticisms generated by this research leveled at the largely

economic focus in past studies of organized crime. However, this largely survey-based

index is only used to represent this element of Anomic culture in Messner and

RosenfeldÕs model of Institutional Anomie Theory, and is not used to measure

organized crime or any separate institutions. This research sought to test Institutional

Anomie Theory as closely as possible, and therefore must operationalize the variables as

the theory intended.

81 Figure 8 below illustrates the model of Institutional Anomie with the assigned variables that will

be tested in this dissertation.

Figure 8. Adapted Schema of Institutional Anomie Theory with Variables82

The World Index of Economic Freedom83 combines fifty economic indicators that

are grouped into 10 Òfreedoms,Ó including trade policy, fiscal burden of government,

government intervention in the economy, monetary policy, capital flows and foreign

investment, banking and finance, wages and prices, property rights, and black market

activity, in approximately 184 countries (Beach and OÔDriscoll, 2003; Heritage Foundation,

2012; OÕDriscoll, Holmes, and OÕGrady, 2003). The index is compiled from multiple

sources, to include data from social surveys and officially reported data. Taken separately,

these indicators are reflective of the economic institution, which would be problematic and

conflict with the empirical separation of institutions from culture in this

82 Adapted from Messner and Rosenfeld (2012).

83 Economic freedom is defined as Òthe absence of government coercion or constraint on the

production, distribution, or consumption of goods and services beyond the extent necessary for citizens

to protect and maintain liberty itselfÓ (Beach and OÕDriscoll, 2003: 50).

study. However, the composite index of these approximately 50 indicators is intended to

reflect gaps in what the Heritage Foundation labels Òeconomic freedom and prosperity.Ó

That is, the level of openness to free-market capitalism for each country is placed along a

continuum that reflects Òtraditional American valuesÓ (Heritage Foundation, 2012: 1).

This is directly in line with Messner and RosenfeldÕs (1994) conception of a culture that

is more susceptible to high rates of crime; these authors label this type of problematic

culture as the ÒAmerican Dream,Ó and claim any country with a cultural configuration

most closely matching the American Dream will in turn experience high rates of Anomie.

Therefore, in this research if a country scores low on the World Index of Economic

Freedom, it can be assumed to be less in line with Òtraditional American valuesÓ that

are problematic in generating higher rates of Anomie. Therefore, this Index is the best

operationalization of Messner and RosenfeldÕs (1994) utilitarian conceptualization of

cultural pressures to succeed.

In recognition that multiple measures are often preferred to single or uni-dimensional

indicators, this research will also consider the Gross Household Savings Rate to represent

Messner and RosenfeldÕs (1994) concept of cultural pressures to succeed. That is, they view

money as the ÒmetricÓ of success. As Orru (1990) states, ÒMoney is literally, in this

context, a currency for measuring achievementÓ (235; emphasis in original). The Gross

Household Savings Rate is calculated as the gross savings amount divided by the gross

disposable income, with the latter being adjusted for the change in the net equity of

households in pension funds reserve (Eurostat, 2012e). Gross savings is illustrative of the

amount of the gross disposable income that is not spent as consumption on necessities.

Therefore, countries with higher savings rates can be assumed to reinforce, or at least be

compatible with, Messner and RosenfeldÕs (1994) elements of individualism,

achievement, pecuniary materialism, and universalism. This indicator, together with the

Index of Economic Freedom, should present more comprehensive proxy representing

(utilitarian) cultural pressures to succeed.

The second element in Messner and RosenfeldÕs cultural schema (see Figure 8,

pg. 137) is the lack of emphasis on legitimate means to succeed. To-date, no empirical

studies have addressed this as a separate cultural element. However, Messner and

Rosenfeld (2012: 89) (in line with MertonÕs Anomie Theory) imply it to be separate from

the four cultural elements:

At the same time, the American Dream does not contain within it strong

injunctions against substituting more effective, illegitimate means for less

effective, legitimate means in the pursuit of monetary success. To the contrary,

the distinctive cultural message accompanying the monetary success goal in the

American Dream is the devaluation of all but the most technically efficient

means. This anomic orientation leads not simply to high levels of serious crime

in general but to especially violent forms of economic crime.

The above statement implies that Anomie, then, is the product of cultural pressures to

succeed coupled with a lack of emphasis on legitimate means to succeed (Merton, 1938).

Here, lack of legitimate means to succeed moderates the cultural pressures to succeed.

That is, if cultural pressures to succeed were high in a society, but the societal members

had access to and an emphasis on legitimate means to succeed, Anomie levels should be

lower than in countries without emphasis or access to legitimate means (Messner and

Rosenfeld, 1994). Again, Messner and Rosenfeld (2012) assign a utilitarian emphasis to

this cultural element. Because of this, this proposed research will measure a lack of

emphasis on legitimate means to succeed as the Corruption Perceptions Index (CPI) and

the unemployment rate.

The Corruption Perceptions Index (CPI) was first launched in 1995 by

Transparency International, an organization dedicated to collecting and disseminating

statistics, policies, and knowledge surrounding corruption in the public sector in over 100

nations around the world (Transparency International, 2012a). The Index itself is a

composite measure that combines information from a number of polls and surveys with

corruption-related data collected by a variety of institutions and official governments.

ÒThe CPI reflects the views of observers from around the world, including experts living

and working in the countries/territories evaluatedÓ and ranks countries Òbased on how

corrupt their public sector is perceived to beÓ (Transparency International, 2012b: 1).

Perceptions, capturing perceived measures of both administrative and political

aspects of corruption, Òare used because corruption Ð whether frequency or amount Ð is to a

great extent a hidden activity that is difficult to measureÓ (Transparency International,

2010: 4).84 To build this composite measure, the CPI includes survey questions

addressing Òthe misuse of public office for private (or political party) gain: including

corruption in public procurement, misuse of public funds, corruption in public service,

and prosecution of public officialsÓ (Transparency International, 2010: 17).

Questions also address the level of transparency regarding corruption (public

awareness), the likelihood of encountering corrupt officials, and the perceived levels

of undocumented payments or bribes connected with exports and imports, public

utilities, tax collection, public contracts, and judicial decisions.

84 To calculate the CPI, the data are first standardized using a matching percentiles technique that takes

the ranks of countries reported by each individual score. ÒThis method is useful for combining sources

that have different distributionsÓ (Transparency International, 2010: 15). The bounds of the CPI are

between 0 and 10 Ð with 0 being completely corrupt and 10 being the least corrupt.

The CPI is an appropriate measure to represent MertonÕs and Institutional

Anomie notion of Òlack of emphasis on legitimate means to succeedÓ because varying

levels of corruption in societies undermines the ability of its members to find and

maintain legitimate jobs, it risks the stability of financial markets, and corruption

negatively impacts the perceived legitimacy of the criminal justice system and the

political process (Transparency International, 2010). That is, high rates of perceived

corruption in the public sectors indicate that legitimate avenues may be threatened for

citizens to obtain even the basic necessities. Consequently, countries that score high on

the perception index (indicating very low levels of perceived corruption) are

representative of a cultural emphasis on legitimate means to succeed. Countries with a

low CPI (indicative of high levels of perceived corruption) represent a general lack of

emphasis on legitimate means to succeed.

Unemployment rates were also used as a proxy representing a lack of legitimate

means to succeed. The rate of unemployment, defined by the World Bank (2012) as Òthe

share of the labor force that is without work but available for and seeking employment,Ó is a

logical indicator to include, as the inability to find work significantly limits the ability to

achieve success goals through legal means. Therefore, the higher a country scores on the

unemployment rate, it can be assumed the country has a correspondingly higher lack

of emphasis on legitimate means to succeed.85

85 A good point to make, however, is that Messner and Rosenfeld (2009) acknowledge that cultural and

institutional elements are known to change slowly over time. Unemployment rates are known to fluctuate at

faster rates than perhaps would be reflective of an actual Anomic cultural (or institutional) shift; all

attempts at operationalizing ÒAnomic cultureÓ and social institutions should be wary of this point.

Therefore, these Anomic cultural variables considered together are seen as

representative of Anomie.86 That is, as countries experience intense cultural pressures to

succeed (reinforced by increased levels of economic freedom and increased household

savings), Institutional Anomie Theory predicts that these societies will experience higher

levels of Anomie when they also experience higher unemployment rates and elevated

levels of perceived corruption (resulting in a lack of emphasis on the legitimate means to

succeed).

Social Institutions

The other half of Messner and RosenfeldÕs (2012) model of Institutional Anomie

Theory (see Figure 7, pg. 135) are the four social institutions87 that represent an institutional

balance of power. This institutional balance of power is composed of the economic, political,

familial, and educational institutions. As the theory states, Anomic societies are typically

characterized by an economic dominance over other non-economic institutions (Gross and

Grossmann, 2011). However, to-date, no study has been able to

create an empirical measure of an Òinstitutional balance of powerÓ.88 The vast majority

of past research endeavors have simply tested the four institutions separately in

multivariate regression analyses and concluded a stronger economic dominance exists

when the coefficients for economic indicators are larger and more significant than non-

economic indicators. Recently, Gross and Haussmann (2011), utilized separate indicators

and inferred that lower scores on variables such as welfare spending was indicative of a

86 This will improve on past operationalizations of Anomie, as existing studies have largely

considered Anomie to be a representative of a strong economic institution (e.g., Bjerregaard and

Cochran, 2008a; 2008b).

87 No studies, aside from Cullen and colleagues (2004) and in part, Gross and Haussmann (2011), have

ever empirically considered variations in ÒcultureÓ apart from Òsocial institutions.Ó

88 However, scholars of Institutional Anomie Theory, to include Dirk Enzmann, are currently

examining ways of empirically measuring this balance.

dominant economy: ÒLow rates of welfare spending can be interpreted as an indicator of

pronounced commodification (commercialization of all areas of societyÉ) and thus

economic dominanceÓ (308). Without guidance on the formation of an empirical index

system to represent measure of the four institutions in one measure of an Òinstitutional

balance,Ó this research will employ the logic proposed by Gross and Hausmann (2011)

by including separate measures that can be inferred to represent an economic imbalance

of institutions.

The economic institution Òconsists of activities organized around the

production and distribution of goods and services. It functions to satisfy the basic

material requirements for human existence, such as the need for food, clothing, or

shelterÓ (Messner and Rosenfeld, 2012: 75). A commonly used indicator in past

research to represent the strength of the economy is the Gross Domestic Product (GDP)

per capita (Bjerregaard and Cochran, 2008a; 2008b). The GDP per capita is a basic

measure of a countryÕs overall economic health.

As an aggregate measure of production, GDP is equal to the sum of the gross

value added of all residentÉ industries engaged in production, plus any taxes, and

minus any subsidies, on products not included in the value of their outputs. Gross

value added is the difference between output and intermediate consumption.

(Eurostat, 2012b).

Thus, this research also utilized GDP per capita to represent the economic strength of each

country. However, instead of using just the GDP per capita, this research calculated the

annual percentage growth rate of GDP per capita in order to illustrate changes from year to

year, thus providing a more appropriate measure of economic strength. This is because the

annual percentage growth rate of GDP illustrates changes within the same country from

year to year, allowing the researcher to compare changes in percentage

growth within and between countries. That is, one can compare a 15% growth rate in

2003 in Poland to a 3% growth rate in the U.K. at the same time. GDP per capita rates do

not allow for this level of comparison; comparing per capita does not provide the context

to compare rates between countries (to say that AustriaÕs GDP per capita in 2003 was

38,967 versus BelgiumÕs GDP per capita of 43,849 is less meaningful).89

The next institution is the polity, which is designed to Òmobilize and

distribute power to attain collective goalsÓ (Messner and Rosenfeld, 2012: 75).

Messner and Rosenfeld (2012: 75) note:

One collective purpose of special importance [for the political institution] is

the maintenance of public safety... As part of the polity, agencies of the civil

and criminal justice systems have major responsibility for crime control and

the lawful resolution of conflicts.

However, to date studies of Institutional Anomie have only operationalized this

institution through voter turnout (with one exception). This measure is problematic,90 and

does not embody Messner and RosenfeldÕs (2012) definition of the social institution. The

one study to define the polity as Messner and Rosenfeld (1994) originally intended was

Gross and Hausmann (2011), who utilized survey data91 to capture trust in the police, the

overall government, and the legal system. Therefore, this research remained true to the spirit

of the theory and operationalized the polity using three measures that more closely

represents Messner and RosenfeldÕs (2012) above description of the political institution:

the total spending on social protection, a Rule of Law measure, and a Political Stability

89 The percentage growth rate of GDP per capita will only be used in the first stage of the analyses; that

is, the visual inspection. This is due to the restrictions of the use of negative numbers in the multivariate

time series analyses.

90 This measure is only appropriate for more fully developed democracies that actually hold true (i.e.,

non or less corrupt) elections, thereby excluding any transitioning or non-democratic nations from the

study. Additionally, there is Òconsiderable variation across the democracies in voter-turnout ratesÓ

(Jackman, 1987).

91 This data was gathered from the European Social Survey (ESS).

and Absence of Violence measure. The social protection expenditure is defined as the

amount of government expenditures (represented as per capita) on law courts, policing,

fire-protection services, and prisons (Eurostat, 2012). Therefore, countries ranking high

on this indicator are implied to be more politically driven, and hold a greater emphasis for

the political institution because more emphasis is placed on ensuring funding for these

services.

The Rule of Law measure and the Political Stability and Absence of Violence

measure are two indicators out of a total of six dimensions of governance compiled by

the Worldwide Governance Indicators (WGI) project, headed by the World Bank

(2013a). The WGI project reports aggregate and individual governance indicators for 215

nations from the mid-1990s until 2011.

Governance consists of the traditions and institutions by which authority in a

country is exercised. This includes the process by which governments are

selected, monitored and replaced; the capacity of the government to effectively

formulate and implement sound policies; and the respect of citizens and the

state for the institutions that governÉsocial interactions among them. (World

Bank, 2013b)

Each dimension of governance is based on a series of aggregate indicators that include

officially reported data as well as survey data. ÒThe six composite WGI measures are

useful as a tool for broad cross-country comparisons [of governance] and for evaluating

É trends over timeÓ (World Bank, 2013b). The Rule of Law measure and the Political

Stability and Absence of Violence measure were chosen out of the six dimensions for

their relevance to Messner and RosenfeldÕs (1994) conception of the polity.

The Rule of Law dimension92 Òcaptures perceptions of the extent to which agents

have confidence in and abide by the rules of society, and in particular the quality of contract

enforcement, property rights, the police, and the courts, as well as the likelihood of crime

and violenceÓ (World Bank, 2013c: 1). This dimension includes factors such as the

confidence in the police force, confidence in the judicial system, reliability of police

services, and effectiveness of the judicial system. This is very closely related to the

requirement of the polity outlined by Messner and Rosenfeld (1994); that is, the

maintenance of public safety. Here, the Rule of Law represents the extent to which the

people of each nation feel that their country is indeed maintaining public safety.

The Political Stability and Absence of Violence93 dimension Òmeasures

perceptions of the likelihood that the government will be destabilized or overthrown by

unconstitutional or violent means, including politically-motivated violence and

terrorismÓ (World Bank, 2013d). This measure includes factors such as the threat of civil

unrest, government stability, violent social conflicts, and violent demonstrations. This

governance dimension also closely follows Messner and RosenfeldÕs (1994) definition

of the polity, as it addresses levels of actual and perceived civil unrest within each nation.

The educational institution Òaims to enhance personal adjustment, facilitate the

development of individual human potential, and advance the general knowledge base of the

cultureÓ (Messner and Rosenfeld, 2012: 75). Moreover, Òschools are given responsibility

for transmitting basic cultural standards to new generationsÓ (Messner and Rosenfeld, 2012:

75). To represent the importance placed on the educational institution in

92 The Rule of Law dimension is measured in a percentile rank, ranging from 0 to 100. 0 represents

the lowest level of governance, and 100 represents the highest.

93 This dimension is also measured in a percentile rank, ranging from 0 to 100. 0 represents the lowest

level of governance, and 100 represents the highest.

each country, past studies have operationalized this construct as the educational

expenditures as a percent of GDP (Bjerregaard and Cochran, 2008a; Maume and Lee,

2003). Educational expenditures generally refers to Òdirect expenditure on educational

institutions: bearing directly the current and capital expenses of educational institutionsÓ

(EUROSTAT, 2012c). This measure has demonstrated the ability to capture the

emphasis placed on the amount of educational institutions available within each country,

thereby facilitating the advancement of knowledge within the society; this research also

utilized this measure with the slight modification of calculating it as per capita instead of

as a percent of GDP to ensure that the measure is not dependent on the level of GDP in

each country, but rather on the population.

Additionally, to include a proxy for the strength of the educational institution in

each country, this research also included a measure of the educational attainment level.

The primary education level is defined as the percentage of the labor force having

completed at least primary schooling (World Bank, 2012). This was an important

indicator to include in this study in order to separate aspects of the educational institution

from its inherent economic ties, thus reducing the likelihood for high correlation among

the variables.

Finally, the familial institution Òbears primary responsibility for the regulation of

sexual activity and for the replacement of members of societyÓ (Messner and Rosenfeld,

2012: 75). ÒOne of the most consistently utilized measures of the weakening of the family

units is the divorce rateÓ (Bjerregaard and Cochran, 2008a: 188), and additionally has been

one of the most consistent measures of the familial institution in empirical tests of

Institutional Anomie Theory (Bjerregaard and Cochran, 2008a; 2008b; Chamlin and

Cochran, 1995; Piquero and Piquero, 1998; Maume and Lee, 2003). High divorce rates

represent family disruption, thus Òindicating a breakdown of the traditional nuclear

familyÓ (Bjerregaard and Cochran, 2008a: 188). However, changes in marriage rates

over time must be accounted for. Therefore, this research utilized the ratio of divorce

rates to marriage rates in each nation as representative of the familial institution.

Divorce-to-marriage ratios will be computed as the number of marriages divided by the

number of divorces per 100,000 population, which is in line with past research (Cullen,

Parboteeah, and Hoegl, 2004).

Additionally, the Total Public Social Expenditures (per capita) was added as another

proxy representing the familial institution. The Social Expenditure Database (SOCX) was

developed by the OECD Òin order to serve a growing need for indicators of social policy. It

includes reliable and internationally comparable statistics on public policy and (mandatory

and voluntary) private social expenditure at program levelÓ (OECD, 2012). This indicator is

similar to the one used by Gross and Haussmann (2011) to represent welfare spending in

each nation. As Gross and Hausmann (2011) surmise, the amount of welfare spending in

each country has been interpreted as an indicator of the degree of commodification present in

each society. Countries with social policies less oriented towards existing needs are believed

to experience Òelevated commodification of labor (increased market dependency)Ó (Gross

and Hausmann, 2011: 308). This represents an imbalance of the emphasis on the importance

of the family as represented through social programs that strengthen the family unit

(healthcare, community programs, after school programs, day care, etc.) and the economic

institution. Here, parents are (culturally) encouraged and sometimes forced to work overtime

or work multiple positions instead of attending their childÕs soccer game or staying home

to help them with their homework (Messner and Rosenfeld, 2012).

To ensure that other factors are not significantly influencing these institutions, the

following variables were controlled for: the age distribution and the percent of population

that is male. Even though Institutional Anomie Theory does not address individual

differences, Cullen, Parboteeah and Hoegl (2004) suggest using these Òindividual-level

control variablesÓ largely because criminological research Òhas shown relationships

[exist] between most forms of crimeÉ with age [and] genderÓ (416).

Population and Sampling Selection Criteria

As international organizations have illustrated (e.g., Council of Europe, 2000;

2005; OCTA, 2011), there exists much substantial variation of organized crime activity

between and within Europe in terms of the activities, the consumer countries, and the

countries utilized for their trafficking routes. Based on this geographical representation

within Europe, the following countries have been selected to represent the Western and

Central European Region in this research (UNODC, 2010): Austria, Belgium, France,

Germany, Greece, Ireland, Italy, Lithuania, Poland, Portugal, Spain, Slovakia, Sweden,

and the United Kingdom. This selection was based on two main criteria: identifying

countries that provided geographical representation across all parts of Europe (to

include transitioning nations), and countries that had reliably reported indicators from

1995 to 2009 with the least amount of missing data.

Moreover, because Òit is very difficult to provide detailed descriptions of so many

nations simultaneously and to make sense out of itÓ (Marshall and Summers, 2012: 42),

scholars have found it helpful to organize these large numbers of countries into country

groupings or clusters.94 The classification system used in this research was one based on

the work of Esping-Andersen (1990) and Lappi-Seppala (2007), and Òhas a strong

conceptual foundation that takes into account a number of unifying and separating

factors like social welfare investment, income inequality, geography, political traditions

and orientations, and history and culture traditionÓ (Marshall and Summers, 2012: 42).

Thus, the Western and Central European region was broken down into the following six

country clusters (Enzmann, et al., 2010; Esping-Andersen, 1990; Lappi-Seppala, 2007;

Saint-Arnaud and Bernard, 2003):

Representation from each country cluster will allow for the three sets of research

questions to be considered from multiple comparative frameworks.95

Data Sources

All of the dependent and independent variables used to address the research

questions and their related hypotheses were collected from several official sources. The

dependent and independent variables used to empirically address regional variation in

organized crime activity were collected from official reports for the following European

countries: Austria, Belgium, France, Germany, Greece, Ireland, Italy, Lithuania,

Poland, Portugal, Slovakia, Spain, Sweden, and the United Kingdom. The data are

available online from Eurostat (2012), OECD (2012), Transparency International

(2012a), The Herritage Foundation (2012), and the World Bank (2012).

The data for empirical analyses of Poland were collected from a number of

national and international sources and databases. Specific drug-related data are annually

reported figures collected by the National Bureau for Drug Prevention (NBDP) and the

Institute of Psychiatry and Neurology in Warsaw under the Ministry of Health in Poland.

The NBDP receives data from the police, court systems, and government-run facilities,

and the data are made electronically available on a yearly basis through the National

Reports. The other dependent and independent variables are readily available online from

95 This research naturally lends itself to a comparative framework. The methods behind this framework,

identified first by Durkheim (1962/1895), include 1) the analysis of variations within one given society at

one point in time; 2) the comparison of similar societies but differing in certain aspects; and 3) the

comparison of dissimilar societies that share some feature(s). The first of these applications is intra-

societal in nature, meaning cultural and institutional elements are examined against some phenomenon to

be explained within that society. This application has historically been the most popular method employed

by comparative criminologists.

Eurostat (2012a), OECD (2012), Transparency International (2012), The

Heritage Foundation (2012), and the World Bank.

Analytic Strategies

The quantitative data analysis plan was able to explore whether countries in

Europe with higher rates of Institutional Anomie also had correspondingly higher rates of

organized crime activity, taking into account changes over time and the grouping of

European countries into clusters and into developed and transitioning country groups.

These same strategies were then employed to address how Institutional Anomie Theory

operated within the transitioning democracy of Poland. This analytic strategy involved

multiple stages that were most appropriate for the data at hand (see Figure 10 below).

First and foremost, this research graphed all of the variables included in this study

(see Figure 6, pg. 123) and calculated descriptive statistics, allowing for a visual

inspection of the data (Maltz, 2010). This involved inspecting the data for the two

concepts embodying Institutional Anomie (i.e., Anomic culture and social institutions),

the measures of organized crime (heroin, cocaine, amphetamine, and marijuana seizures)

and intentional homicide rates, as collected for each European country (Austria, Belgium,

France, Germany, Greece, Ireland, Italy, Lithuania, Poland, Portugal, Slovakia, Spain,

Sweden, and the United Kingdom). Once the descriptive statistics were run on all the

variables, any outliers were identified. Any outliers that were identified were noted, but

not removed from the study since the nature of the outliers remains unclear and these data

may in fact turn out to be valuable indicators.

96 The sources for this section of Chapter 4 that discuss time series analysis and pooled cross-sectional

time series analysis include the following: Dadkhah (2007), Maddala and Kim (1998), Hamilton (1994),

Enders (1995), and Campbell et al. (1997).

Full Model, Country-Cluster, and Developed v. Transitioning Group Analyses

The next stage in the data analysis addressed the full model (i.e., all fourteen

countries) variation in organized crime activity (see Figure 10 below). These methods

employed multivariate time series analyses (described below) for the full model from

1995 to 200997 across all five dependent variables.98 Next, unit root tests were conducted

to determine the stationarity of the variables, followed by Granger causality tests, and

(depending on the non-stationarity of the variables) tests for cointegration. Each

multivariate regression and Granger causality model indicated the coefficient, standard

error, and significance of the explanatory variables, which in turn indicated the Anomic

cultural and social institutional proxies with the greatest impact on each of the four types

of drug seizures in Europe. The results also illustrated whether countries in Europe with

higher rates of Institutional Anomie also had correspondingly higher rates of organized

crime activity (as measured through drug seizures) over time. The results from the

cointegration models yielded a t statistic, indicating the presence of cointegration

through the stationarity and significance of the t statistic. Diagnostics were run to assure

the absence of multicollinearity and highlight any potential serial correlation.

Then, pooled cross-sectional time series analysis was employed to address

differences between country clusters.99 The same analyses were conducted as used in the

full country model. That is, the independent variables were first tested using pooled cross-

sectional multivariate regressions. Next, following unit root analysis to determine

stationarity among the country clusters, Granger causality and cointegration models were

run. These results were compared with the full-country analyses to determine whether

grouping countries into country-clusters significantly impacted Institutional Anomie

TheoryÕs ability to predict high levels of organized crime. These results were also

compared with the full-country analyses to determine whether grouping countries into

country-clusters significantly impacts Institutional Anomie TheoryÕs ability to

predict high levels of organized crime.

This study next separated and pooled the countries in this sample into two groups:

developed countries (Sweden, Ireland, United Kingdom, Austria, Belgium, France,

Germany, Greece, Italy, Portugal, Spain) and transitioning countries (Poland, Slovakia,

Lithuania).100 This research first ran multivariate regression analyses on the pooled

data.101 Then, comparisons between the strength and statistical significance of the

coefficients for culture and social institutions were made between the two groups to

investigate any variations over time. Additionally, to uncover further relationships

existing among and between the proxies of Anomic culture and social institutions, tests

for Granger causality and cointegration were conducted.102 Diagnostics were again run

to assure the absence of multicollinearity and highlight any potential serial correlation.

Polish Analyses

A similar analytic strategy was applied to the country-level analyses that allowed

for comparisons to be made between the compositions of PolandÕs culture-institutional

configuration with the other CEE cluster country represented in this research (Slovakia) for

the same time period (1995 to 2009). This involved time series regression analyses.

However, due to the lack of observations for the separate country analyses between

100 The differences between ÒdevelopedÓ and ÒtransitioningÓ relate to the age of the democratic

governments. Poland, Slovakia and Lithuania are the youngest, following the collapse of the USSR in 1989.

Thus, these governments are still considered to be Òin transitionÓ as opposed to more stable, established

governments.

101 The developed country group had 165 observations, and the transitioning country group had 45

observations. Only slight concerns surrounded the results in the transitioning country group attributable to the n size.

Poland and Slovakia, Granger causality and cointegration tests were not calculated.103

Because of this limitation, the country-level analyses between Poland and Slovakia

represented preliminary endeavors to uncover the influence of social organizational

elements in these countries. However, the limitation of data was acknowledged and as

this research was exploratory, the analytic techniques applied here allowed for an initial

signal of Institutional Anomie Theory to be detected. These techniques were also able to

detect how these variables impacted proxies for organized crime.

The following elements of this section detail the econometric techniques used to

address the three sets of research questions (see Figure 10 above) by describing the

technical language behind multivariate and pooled cross-sectional time series analyses, as

well as the Granger causality and cointegration tests.

Time Series Analysis

Time series analysis gained prominence in econometric analysis during the 1970s

and 1980s. A time series is a collection of random variables ordered in time {Xt}:

Xt = Tt + St + Ct(1)

Where Tt is a long-term trend component; St is seasonality; and Ct is the

remaining short-term component of cyclical residual. The properties of time series became

hugely popular in macroeconomics, as this type of analysis Òprovide[s] valuable

benchmarks to assess the effect of policy changesÓ (Dadkhah, 2007: 143). To understand

103 As explained in sections below, these countries had non-stationary independent and dependent variables

that further limited the number of observations, because the variables would have had to be first-differenced and then

taken the lag of the first difference before inclusion in the Granger causality or cointegration models. Non-stationarity

was less of a concern for the Northern and Baltic country clusters, which is why these two statistical tests were

conducted tentatively for each cluster but not for the individual country analyses.

the basic elements of the time series analyses undertaken in this research, a few terms

and concepts first need to be defined.

A time series is said to be stationary if the data do show some sort of trend (as

opposed to non-stationary). Stationary time series illustrate mean-reverting processes,

while non-stationary time series are absent a mean and are either increasing (with a

positive trend) or declining (with a negative trend) towards infinity. In other words, the

mean and variance of a stationary process are constant over time. The covariance of a

stationary process between any two points in time depends only on the time difference

between these two points. A stationary series yt has the following properties:

(2) Most social science time

series are not stationary. ÒWhile we can consider each

observation [in historical data] as a sample or realization of a distribution with constant

mean and variance, the whole series cannot be assumed to have a constant mean and

varianceÓ (Dadkhah, 2007: 20). Therefore, the trend must be modeled. If a time series is

stationary, then it does not have a trend and contains no unit root (see footnote 107).

This means that any shocks to the time series are necessarily temporary and the effects

will dissipate over time as the series reverts to its long-run mean. A non-stationary time

series with a unit root has a stochastic trend. Therefore, any shock to the system will be

permanent. A non-stationary time series may have a deterministic trend; in this case,

shocks to the system affect the variable in only one time period and have no permanent

effect. A non-stationary time series may contain a unit root process as identified through

the Levi-Lin-Chu (2002) formal test for the presence of a unit root.104 Importantly, the

assumptions of the classical regression model105 require that both the {yt} (dependent

variable) and {zt} (independent variable) sequences be stationary (Enders, 1995).

However, most time series containing social science indicators (to include crime

statistics) show some sort of upwards or downwards trend, which is problematic for

statistical analysis. Thus, the time series must be first-differenced (known as second-

order stationary) before it can be further manipulated.

The field has since experienced many developments over the past three decades,

to include the causality test proposed by Clive Granger and cointegration analysis

developed by Engle and Granger. The issue of causality is one of particular interest to

criminologists and other social scientists alike. It must be remembered, though, that

104 The first step in testing for causality or cointegration between the variables of interest is to determine

if they are stationary or else the degree of their integration. The reason is that testing for causality between stationary

and non-stationary variables is different. Indeed it is easier to turn non-stationary variables into stationary by first

differencing than to test for causality among non-stationary variables. The danger is to fall into the trap of spurious

correlation. A requirement of cointegration is that at least two variables in the set be integrated of order one.

The widely used Levi-Lin-Chu (2002) test of stationary is for the existence of stochastic trend. Consider

the process:

Where is a stationary process with mean zero and variance of one. If then has unit root and is

integrated of order one. Thus, in order to test for a variable being stationary we consider the equation:

the null and alternative hypotheses are:

Rejecting the null hypothesis establishes that the process is stationary. Rewriting the equation as:

Since trends in social and economic variables are generally positive, it will be tested to determine if the

coefficient of is greater than or equal to zero. Rejecting this null hypothesis establishes that is

stationary. The test statistics is the t-statistics of the coefficient of . Its distribution, however, is not

the t-distribution. Indeed it is the ratio of two random variables. The distribution is available via different

statistical packages including Stata.

It is noteworthy that the trend may not be necessarily stochastic. Tests are available to

distinguish between stochastic and deterministic trends. These tests shall be used, but the power of tests in

distinguishing between the two types of trends is low.

105 yt = a0 + a1 z1 + et

correlation does not imply causality, and only true experimental studies have the ability

to successfully isolate the effect of a dependent variable. Therefore, for time series:

Clive Granger noted that a necessary, but not sufficient, condition for causality is

that knowledge of past values of x should improve the forecast of y. Surely if we

can reject causality in the Granger sense then definitely we can reject causality

in the more strict experimental sense. But failure to reject causality in the

Granger sense does not mean causality in the strict sense cannot be rejected.

(Dadkhah, 2007: 150)

The test of causality involves estimating the following equations:

(3)

The equations suggest that if one cannot reject the null hypothesis, it must be concluded

that x does not cause y. However, the Granger test for causality was developed in the field

of economics, and thus requires a large number of observations to be confident in the

results. Thus this test for causality has had limited application in the field of criminology

due to the limitation of existing data.106

Finally, Engle and Granger developed a test to see if non-stationary time series

move or ÒtrackÓ together Ð known as cointegration. For instance, consider demand and

supply of some licit good. If demand exceeds supply, an error correction mechanism is

set in motion to increase supply and reduce demand. These two time series are said to be

cointegrated; they share a common trend.

Technically, when a linear combination of two or more variables is stationary,

we say that they are cointegrated. Note, however, any combination of two or

more stationary variables would also be stationary. Therefore, the existence of

equilibrium or cointegration among variables cannot be statistically tested unless

at least two of the variables involved exhibit non-stationarity. (Dadkhah, 2007:

154)

106 It has been suggested (Enders, 1995) that approximately 50+ observations are optimal for achieving

sound empirical results. Quite often in criminological research, this equates to at least fifty years worth of data,

which is often difficult to obtain.

After performing a Levi-Lin-Chu (2002) unit root test and establishing the time series as

stationary upon first-differencing the data (known as integrated of order one), the

following regression can be run to test for cointegration:

(4)

If cointegration is present, the residual from the regression will be stationary. This

will indicate that an error-correction relationship (such as the one between the

example of demand and supply of a licit good) exists between them:

(5)

These tests (cointegration and causality) have important implications for the field

of criminology. As previously discussed in Chapter 3, much previous work on organized

crime has focused on market forces of drugs and drug trafficking (e.g., Reuter and

Haaga, 1989). If a relationship is found between two time series (exhibiting institutional

change and crime rates for Spain and Poland, for instance), any shocks to one system will

directly affect the system that shares a common trend. Elements of forecastability are

then present, meaning if a causal relationship is found, the impact of future institutional

changes in Spain may behave predictably with changes in Poland. As already mentioned,

the results are tempered when a less than desirable number of observations are used.

Pooled Cross-Sectional Time Series Analysis

Pooled cross-sectional time series analysis combines time series for several cross-

sections.107 Pooled data Òare characterized by having repeated observations (most

frequently years) on fixed units (most frequently states and nations. This means that pooled

arrays of data are one that combines cross-sectional data on N spatial units and T

107 It should be noted here that Òpooled cross-sectional time series analysisÓ is sometimes referred to as

a Òpanel analysis,Ó but this is often confused with panel research in social survey studies.

time periods to produce a data set of N x T observationsÓ (Podesta, 2002: 6). Given

this explanation, this section of the quantitative analyses utilizes fourteen spatial units

over fifteen years, or 14 x 15. The generic pooled linear regression model estimated by

Ordinary Least Squares (OLS) procedure is given in equation 1:

(6)

Here, i = 1, 2, É; N refers to a cross-sectional unit; t = 1, 2, É; T refers to a time

period and k = 1, 2, É; K refers to a specific explanatory variable. Therefore, yit and xit

refer to dependent and independent variables for unit i and time t; eit is a random error

and §1 and §k refer to the intercept and slope parameters.

Pooling data in this fashion is beneficial to social science research for two reasons.

First, in social science research, limitations of available data often result in small sample

sizes (which is also a concern shared by this current study). By rule of thumb, time series

analysis should utilize at least fifty observations, but this is not often possible in

criminological research. Reliable observations for fifty consecutive years (months, weeks)

are not readily available for multiple countries. Thus, pooled cross-sectional time series

analysis accounts for these limitations by combining limited number of spatial units and

limited number of observations over time. ÒThis allows us to test the impact of a large

number of predictors of the level and change in the dependent variable within the framework

of a multivariate analysisÓ (Podesta, 2002: 7). Second, pooled cross-sectional analyses allow

for the combination of time and space to be considered simultaneously. This is in

comparison to either cross-sectional studies that may look at multiple countries at one point

in time, or a time series model that looks at one country for multiple years of data. A pooled

model tests for multiple countries across multiple points in time.

As with any analysis of time series data, special problems of high correlation may

exist due to trends in the variables involved. Serial correlation might also exist among the

errors in the regression model. Therefore, diagnostics108 were run to assure the absence

of multicollinearity and any serial correlation that might be present among the variables.

Limitations of Data

Within the available collected data there remain some methodological limitations

that may impact the strength of the empirical tests. Officially collected statistics have

well documented problems of underreporting, and validity and reliability concerns

(Plywaczewski, 2004), particularly when the study employs officially reported data from

other countries for any length of time. This is primarily a concern in the three countries

considered to be in transition in this study (Lithuania, Poland, and Slovakia) particularly

for the early years of the data captured (i.e., 1995 to 2000). Following the collapse of

communism in 1989 these countries experienced high rates of institutional conflict and

restructuring. Officially reported statistics are problematic even through the late 1990s as

these countries transitioned (and continue to transition) to democracy. However, the year

1995 was chosen as the starting date for data collection to try and limit these

concerns.109 This also does not mean these data should be ignored. However, since there

is no way to correct or supplement this limitation, the data must be utilized as it was

reported and published.

108 Variance Inflation Factors (VIF) scores were calculated to ensure the absence of multicollinearity. A

VIF score of more than 5 is generally accepted as an indicator of multicollinearity in the field. Durbin-Watson test

statistics will also be calculated to ensure the absence of serial correlation. The DW statistic will be between 0 and

(+/-) 4. A value near (+/-) 2 indicates no first-order serial correlation, but values less than 2 indicate the presence of

positive serial correlation.

109The rationale behind this date selection is discussed earlier in this chapter.

Additionally, composite measures used in this research, such as the World Index

of Economic Freedom (Heritage Foundation, 2012), the Corruption Perception Index

(Transparency International, 2012), and the Gini coefficient, are subject to further

methodological concerns. All composite indices, most particularly the Index of

Economic Freedom, are based on a composition of multiple separate indicators that are

all individually prone to problems of underreporting, validity, and reliability of the

reported national statistics for each country. Additionally, the Corruption Perception

Index faces methodological issues related to international survey instruments (e.g.,

translation of questions between languages, nationally representative samples for each

country, response rates).

As documented in Chapter 2, scholars have criticized Institutional Anomie

Theory for failing to provide helpful guidelines for operationalizing key concepts within

the theory. Institutional Anomie Theory contains major sociological concepts of

historical importance, yet within the theory these remain latent constructs. Thus, any

measure of ÒAnomic cultureÓ or Òsocial institutionsÓ is not a direct measure, but rather

an indirect measure, measuring some aspect of the construct. Therefore, the measures of

these two latent constructs for this research are subject to validity and reliability issues.

However, many of the indicators chosen have been previously used in past research (e.g.,

Bjerregaard and Cochran, 2008a; 2008b; Cullen, Parboteeah, and Hoegel, 2004; Maume

and Lee, 2003; Messner and Rosenfeld, 1997c), and thus have demonstrated reliability in

these studies.

Further, concerns surround empirically separating ÒAnomic cultureÓ from Òsocial

institutions.Ó While Messner and Rosenfeld (1994) originally expressed the desire (in line

with Talcott Parsons (1951)) to keep these two dimensions of social organization distinct

from each other, they also noted Òthey are not ÔthingsÕ that can be neatly separatedÓ

(55). Indeed, some scholars argue that ÒAnomic cultureÓ (or ÒcultureÓ more generally)

is too abstract of a concept to be separated from social institutions at all (Messner and

Rosenfeld, 2001). However, the original model of Institutional Anomie Theory clearly

separates these two constructs from each other, outlining the components that constitute

each. To maintain continuity with Messner and RosenfeldÕs (1994) original conception

of the theory, and to add new measurements of Anomic culture to previous work done by

Cullen and colleagues (2004) and Gross and Haussman (2011), this research empirically

represented Anomic culture and social institutions separately from each other. The

possibility remains, however, for ÒAnomic cultureÓ to be empirically inseparable from

social institutions. Yet, if this is the case, this puts the theory of Institutional Anomie in

danger of being non-falsifiable (Chamlin and Cochran, 2007).

As documented in Chapter 3, there does not currently exist a universal definition

for what constitutes Òorganized crime.Ó Like the underlying concepts of Institutional

Anomie Theory, Òorganized crimeÓ is itself a latent construct. Thus, any proposed

operationalization of organized crime will measure an indirect aspect of the criminal

phenomenon. To help increase reliability and validity of any measure of organized crime,

the European Council (2000) proposed to operationalize the construct through annual

drug seizure amounts. This is largely because the United Nations and Council of Europe

have recognized that drug trafficking is the common denominator between the vast

majority of criminal organizations (Council of Europe, 2000; 2005; OCTA, 2011;

UNODC, 2010; 2012). These syndicates are largely responsible for the transportation and

manufacturing of illicit drugs, and most drugs are seized from these groups at points of

entry and exit from their respective countries (OCTA, 2011; UNODC, 2010; 2012).

Further, data on drug seizure amounts is captured and reported more regularly and for

longer periods of time than other forms of organized crime activity (e.g., human

smuggling and trafficking, cybercrime, weapons trafficking, trafficking in commodities

and prescription medicine) (Council of Europe, 2000). However, it must be noted that a

limitation of this particular operationalization of organized crime is also a measure of

police activity (i.e., drugs are seized by the police in each country), which might skew the

empirical results and in some cases, these drug seizure rates may be less representative of

patterns of organized crime movement. Together, these issues illustrate that measurement

is one important issue for this research.

Limitations of Analytic Strategies

Some difficulties were encountered with the econometric analytic strategies used in

this research that, as previously mentioned, relate to the number of observations available in

some of the empirical models. The power of a statistical test Òis equal to the probability of

rejecting a false null hypothesis,Ó and unit root tests such as the Levi-Lin-Chu (2002) test

have been shown to demonstrate low power (Enders, 1995: 251). Therefore, it is less likely

that the tests will reject the presence of a unit root existing in the time-series data.

Additionally, tests for Granger causality and cointegration require larger numbers of

observations to account for the loss of degrees of freedom when first differencing and/or

lagging the variables necessary to perform these additional tests. This does not present

particular limitations for the full model or the developed country group,

but it is a concern in varying degrees for some of the European country clusters, the

transitioning country group, and the separate country analyses. As such, the results from the

models facing the greatest limitation of observations (i.e., the Baltic and Northern clusters,

Poland, Slovakia) were tempered. However, in this exploratory study the numbers of

observations were large enough for the tests to be able to obtain an initial signal of the

proxies of Institutional Anomie Theory impacting their dependent variables.

Limitations of time series analysis also included making sure a spurious

relationship did not exist among the series. Two randomly generated, independent time

series with no connection could potentially have a high degree of correlation (Granger

and Newbold, 1974). Therefore, a very low Durban-Watson statistic and a low R2

indicate that the time series are correlated. When conducting tests for cointegration, two

time series may appear cointegrated but actually be spurious. This often occurs if non-

stationary time series are first differenced and then included in the model. However, a

spurious relationship will not contain a common trend, whereas two cointegrated series

will share a common trend. Correlation does not imply causation, and any potential serial

correlation encountered in this research was addressed.

Students also viewed