Introduction

As recently noted by Pearson and Hobbs (2001), the vast bulk of research on drug dealing is concerned with retail or “street level” dealing. Ethnographers have assembled a considerable body of knowledge about how drug dealing at this level is organized and carried out, the kinds of individuals involved, the roles they play, how they avoid detection and prosecution, the extent to which they use drugs themselves, how they package and adulterate drugs, the street price of drugs and how much profit they make. In contrast, ethnographers have rarely been able to gain access to the earlier stages of the drug distribution process, known as middle or upper level dealing, and Pearson and Hobbs were only able to identify a handful of relevant studies (Adler 1985, 1992; Reuter and Haaga 1989; Dorn et al. 1992; Ovenden et al.1995, Natarajan and Belanger 1998; Dorn et al. 1998; DesRoches 1999; Natarajan 2000). Footnote 1 Consequently, knowledge of these levels is extremely limited. In particular, little is known about the topic of this paper—the structure and organization of groups operating at these levels. Following Pearson and Hobbs (2001) and Reuter (2004), the present state of this knowledge can be summarized as follows:

  1. 1.

    Little evidence supports the idea that upper and mid level dealing is controlled by tightly organized, complex and hierarchical organized crime groups that operate around the world. Even the Medellin and Cali “cartels” seem to be only loose syndicates of independent entrepreneurs, who sometimes collaborate but who also compete with each other and with other, smaller organizations.

  2. 2.

    Market power is elusive at these trafficking levels; it is only at the level of a few streets that drug dealing organizations have the power to exclude others or to set prices.

  3. 3.

    Considerable variety exists in upper and mid level drug dealing groups and the roles of supplier and buyer are generally interchangeable. Some regional or local “corporate” style groups exist, and there are also some fairly large “family” businesses or ones founded on ethnic ties among individuals with links to source countries for drug production and with access to smuggling routes.

  4. 4.

    However, most criminal networks involved in upper and middle market drug distribution are small, led by one or two individuals, who control the money and who have contacts with a small number of producers or wholesalers. These individuals employ small teams of runners, sometimes on a casual basis, who collect and deliver batches of drugs to customers.

The considerable variety of drug dealing organizations operating at mid and upper levels means that these conclusions, based on only a few studies, must be treated as provisional and incomplete. However, they gain some support from the similar findings of contemporary research on organized crime (e.g. Kleinknecht 1996; Klerks 2001; Moore 1987; Albanese 2004; Clarke and Brown, 2003; Guerette and Clarke 2005; Fijnaut et al. 1998; Finckenauer 2000; Kleemans and van de Bunt 1999; Zhang and Chin 2003). These studies have concluded that trafficking and smuggling, whether of humans, drugs or other commodities, are not generally committed by traditional organized crime groups such as the “Cosa Nostra”. Rather, they are committed by loose networks of small groups of entrepreneurs who are exposed to the opportunity to participate in these crimes by family and ethnic ties, and who engage in the crimes for personal profit not organizational gain (Zhang and Chin 2003). Footnote 2 The networks might form quickly to accomplish specific transactions and be dissolved equally quickly once these are complete. This is made possible by the fact that many of the individuals in the networks involved have legitimate ways to make money and engage in the crimes as a sideline. These networks provide many operational advantages for engaging in smuggling and trafficking: “Such small-sized groups can adapt well to ever-changing market constraints and uncertainties. They have little vertical bureaucratic structure, even though their operations may involve highly specialized tasks. They ally with only those with valuable resources to contribute to their objective of making a profit.” (Zhang and Chin 2003, p. 478). Furthermore, small groups have certain important security advantages because “the fewer the number of people involved in a smuggling ring, or knowledgeable about its activities, the greater the chance of keeping the operation concealed from law enforcement agencies.” (Zhang and Chin 2003, p. 478)

Given the consistency of findings from the organized crime and drug trafficking literatures, it could be that quite widespread changes in the nature of organized crime have been taking place in recent years, and that it might now be more often correct to refer to these crimes not as organized crime, but as “crimes that are organized” (Finckenauer and Waring 1998; Hobbs 2001; Potter 1994; Zabludoff 1997). This represents a new paradigm for understanding mid and upper level drug dealing, as well as organized crime in general (Brown and Clarke 2004; Hobbs 2001; Zhang and Chin 2003, 2004), and it also has far reaching implications for interdiction.

Before embracing this view, however, it would be important to undertake more in-depth studies of specific drug dealing organizations. There is a particular need to understand their structure, the roles of individual members, the network of contacts among them, the degree to which they specialize in particular tasks, the extent of communication within and between sub-groups, and the degree to which they are geographically concentrated. Traditionally, this type of understanding has been produced by ethnographic and interview methods, but Natarajan (2000) has shown that analysis of wiretap records collected in the course of prosecuting trafficking organizations can also yield detailed knowledge on these points. She studied a large, 28 member organization involved in importation and wholesale distribution of cocaine, which was prosecuted in New York City in 1996. The organization included a large number of individuals, with a clear division of labor and a recognizable hierarchy. It therefore met the criteria of a “corporate” style organization in Natarajan and Belanger’s (1998) classification of drug dealing organizations.

The present study adapts and extends Natarajan’s (2000) methodology in studying another, somewhat larger and more loosely structured heroin trafficking operation successfully prosecuted in a New York City court at the end of 1993 after an investigation lasting three years. The prosecution claimed that they had unmasked an international drug trafficking conspiracy, with links to mafia families, which had acquired $144 million in assets. The organization was said to be responsible for transporting, receiving and selling more than 200 hundred kilograms of heroin per year (approximately 193 kg were recovered during the investigation). As for its structure, 35 defendants were identified who controlled a series of 15 independent yet interrelated narcotics distribution groups involved in importing, processing, packaging and distributing narcotics for resale in wholesale and retail quantities. These reached from street level sellers of individual glassines of heroin and operators of heroin “mills” who sold small quantities by the “bundle”, through to middle level wholesalers of heroin and to high-level traffickers within the structure of traditional organized crime groups. The prosecution document (page 3 of exhibit C) claimed that:

“Due to the many different groups responsible for importing and distributing the narcotics in this case, the far flung enterprise uncovered during the course of our investigation is not always divisible into easily defined categories. Moreover, at any given time, a particular subsidiary within the overall criminal conglomerate may be fulfilling several roles in relation to other subcomponents within the organization. However, despite the unique characteristics associated with each particular group within the illicit enterprise, we have learned that what were first viewed as independent narcotics trafficking groups are now recognizable as interrelated segments of the overall trafficking mechanism.”

This portrayal fits the “communal business” type of drug dealing organization described by Natarajan and Belanger (1998). These organizations are composed of small cliques of people forming a larger entity. They have a relatively flat hierarchy and a flexible division of labor (some people taking multiple roles). Their operations tend not to be confined to one geographic location and they mostly deal in one drug type (heroin in this case). There are usually strong ethnic ties among the members of the organization (Italian in this case) and most members are older. The organization also fits the description of a “network organization” that has been described in the business literature as follows:

“...a network organization maintains permeable boundaries either internally among business units or externally with other firms. Management is less hierarchical, deriving its authority more from expertise than from rank... Communication is direct and point-to-point rather than “through channels”, while knowledge of emerging problems and opportunities may arrive via multiple loose associations or weak ties. Resources are specialized and customizable within a given product or service scope yet they are less vertically integrated than their hierarchical counterparts. Purposeful agents within the network may establish ties to other agents and organizations to wrest control for themselves or thwart competitors’ attempts to do the same Tasks are more project and less functionally driven, leading to shorter runs of more differentiated products. Networks reintegrate staff conception and line execution such that local concerns are more locally addressed. ” (Van Marshall 1997, p. 3)

Despite the apparent fit with the literature, a major difficulty with the prosecution’s portrayal of the organization is that no previous study of drug trafficking organizations has identified one with such a broad span of operations. In particular, Natarajan and Belanger’s (1998) study of 39 large drug trafficking organizations prosecuted in the Federal courts in New York City between 1984 and 1997 found none that was engaged in all levels of trafficking (smuggling, wholesale distribution, regional distribution and retail) and, indeed the tasks performed by the majority (24) fell into just one of these levels. This does not mean of course that no organizations exist with such a broad span of operations, but if they do, they are rare.

A second potential difficulty with the prosecution’s portrayal is that it might have been influenced by the intention to secure convictions under conspiracy law. This has several advantages for the prosecution including that conspiracy charges attract additional penalties, they encourage and reward those in the organization who inform on others, and they make low level operators as culpable as the leaders. In these cases, the courts have most often treated drug trafficking enterprises as “chain” conspiracies that involve large numbers of persons divided into layers or groups of personnel. The prosecution must show direct connections between all the members of the conspiracy and must establish that they are knowingly involved in a single enterprise to purchase and distribute illegal drugs (see United States v. Sobamowo, 892 F.2d at 94). It can be assumed that participants understand that they are participating in a joint enterprise because success is dependent on the success of those from whom they buy and to whom they sell (Marcus 1995). Where two or more chains are connected to a hub by core conspirators the courts usually view the entirety as a single enterprise. Footnote 3 (See United States v Mallah, 503 F2nd at 984.) In other words, to successfully prosecute the organization under the laws of conspiracy, the prosecution must portray it as an enterprise with a single objective to distribute narcotics at different levels including importing, wholesale and retail operations.

Given these two difficulties with the prosecution’s portrayal of the organization, it is reasonable to ask whether the prosecution had in fact identified a discrete organization. While there is no reason to doubt that the individuals identified in the case were involved in the distribution and sale of heroin, they might not really have been linked together in a single organization. It could be that the investigation had identified a segment of the heroin market in New York City, not a definable organization, as a result of scooping into its net a multitude of unconnected deals made by various groups and individuals who had intermittent and tenuous contacts with a core group of individuals originally targeted in the investigation.

The study reported below seeks to shed light on this question by analysis of the wiretap records collected in the case: Has the prosecution identified a heroin-dealing organization or has it identified a segment of the heroin market in New York City? If it is a definable organization one might expect to see: evidence of some hierarchy, with a recognizable leader and some status differentials; regular contacts among the members of the organization, particularly among the core group; some division of labor and organizational and geographical division of responsibilities among lieutenants. If these expectations are not met, it might be concluded that the prosecution had identified a market segment, not an organization.

A second purpose of the research, pursued in tandem with the first, was to provide a detailed picture of the organizations—if that is what it really was. This picture would be a significant addition to the small literature on drug trafficking organization. The prosecution materials provide quite detailed information about the nature of the organization, about its day-to-day operations, and about its links to organized crime and to overseas suppliers. This information was needed to mount a successful case against the individuals in the higher reaches of the organization. The prosecution did not describe other features of the organization that were not directly relevant to its case, but which would round out a more scientific description. These include:

  1. 1.

    The approximate size of the entire organization.

  2. 2.

    The relative status of members and the degree of task specialization.

  3. 3.

    The cohesiveness of the organization; the size of the core group; the number and nature of contacts between this group and peripheral members; and the numbers of definable cliques within the organization.

  4. 4.

    The geographic extent of operations within New York City and the links between street selling and wholesale distribution.

The study reported below attempts to provide this and other information about the organization by undertaking a series of separate but linked analyses of the wiretap data gathered in the course of prosecuting the case.

Methodology

The data

The data for this study consisted of English translations of more than 2,000 pages of transcripts in electronic form of wiretap conversations gathered in the course of prosecuting the case. These transcripts were imported into a Folio Views database. Folio Views is a hypertext software program that can materially assist theoretical understanding of large and complex textual files. Every word in the infobase Footnote 4 is indexed and can be manipulated using Boolean operators and other search techniques to find instant answers within massive amounts of information.

For the purposes of the present study, each wiretap conversation was created as a separate record in the Folio Views database. A record is a unit of information (usually a paragraph, sometimes several paragraphs) in an infobase. Records are basic building blocks of the infobase, which may be grouped by topic, assigned levels to simplify formatting and organization, or modified to enhance or emphasize portions of the infobase.

The transcripts yielded a total of 2,408 conversations, ranging in length from one minute to ten minutes gathered from 21 wiretapped phones. To give a flavor of their content, the following is a conversation recorded between two “sellers”, X and FF, who are discussed later in the analysis:

  • Seller X: Hello.

  • Seller FF: Yeah, hello.

  • Seller X: What’s up bud? How you doing?

  • Seller FF: Alright. What’s happening?

  • Seller X: Yeah. Listen. I want to see you man.

  • Seller FF: Yeah.

  • Seller X: I want to give you that thing before they fucking steal it off of me, man. (Laughs) I’m getting a hair cut right now. I could see you in an hour.

  • Seller FF: Oh alright, cause “Y” thought he had an appointment with you at 4:00, but I’ll get a hold of him then.

  • Seller X: You know, alright, no, I won’t make in by 4:00.

  • Seller FF: So what time do you want?

  • Seller X: (Inaudible) Queens. Alright in ah hour, in exactly an hour from now. You want me to call you when I’m down there?

  • Seller FF: (Inaudible).

  • Seller X: It don’t take you but 10 min to get to me.

  • Seller FF: Yeah, yeah.

  • Seller X: Alright, so I’ll call you back.

  • Seller FF: Alright. You wanna talk to, you wanna ...

  • Seller X: I wanna talk to you man.

  • Seller FF: Alright, have you spoken to “him” at all?

  • Seller X: No, I beep him, he don’t answer me. I mean I talked to “V” last night but he don’t know what was going on.

  • Seller FF: Yeah.

  • Seller X: Let’s talk in person me and you. How about that? Huh?

  • Seller FF: Yeah alright, because I’m trying to arrange, I don’t know what’s the way to go. I want to explain to you what “he” told me and maybe you can make fucking sense out of it.

  • Seller X: Oh, OK. Listen, so I’ll call you as soon as I’m finished with my haircut.

  • Seller FF: OK.

  • Seller X: Later.

  • Seller FF: Bye

Table 1 provides the number of conversations and the number of identified individuals for each of the 21 phones Footnote 5 intercepted in the course of the investigation from 1991–1993. Footnote 6

Table 1 Wiretap conversations for 21 phones, 1991–1993
Full size table

The data summarized in Table 1 provide the basis for the analyses reported below about the organization and it is important to be clear about their limitations as a source of information. Many of these derive from the practical and legal constraints on gathering wiretap data for a criminal investigation. In particular, wiretap interceptions must be authorized by the relevant court upon a detailed showing of probable cause. Specifically, the investigating officer must provide a detailed affidavit showing there is probable cause to believe the phone is being used to facilitate a specific, serious, indictable crime. Undoubtedly, chance will help to determine which phones are targeted by the investigators and there is no guarantee that they will succeed in identifying all the phones involved, or even the most critical ones. Beyond demonstrating probable cause, very little information was provided in the present case about the reasons for selecting a particular phone.

At an average in 1992 of $46,492 per authorized wiretap (Delaney et al. 1993), costs also limit the number of wiretaps that can be placed. Footnote 7 This means the suspects targeted for wiretapping may not be representative of the organization as a whole. Their phones may be tapped for relatively brief and possibly unrepresentative periods. They often speak in a foreign language (as in the present case) and translation may result in loss of important information. They also sometimes speak in code that may not be easy to decipher.

The conversations included in Table 1 are only a small proportion of all calls intercepted in the case and comprised only the conversations judged “pertinent” by the prosecution. Under the “minimization” rules, the law enforcement officer performing the surveillance must turn off the intercept when the conversation is not pertinent to the investigation. In cases where the conversation is in code or in a foreign language then it can be recorded and minimization deferred until a qualified person is available to interpret the code or provide translation (Delaney et al. 1993). To illustrate the attrition of calls resulting from these procedures, phone 1 (Table 1) had a total of 3,087 calls during 1991–1992, of which 1,411 were not completed or were beeper calls. Of the 1,676 calls that involved conversations (868 incoming and 808 outgoing calls), only 174 calls were considered pertinent by the prosecution. These conversations involved only the individuals who were indicted and only involved matters relevant to the case. For all these reasons, it is apparent that the data analyzed below comprise a small and possibly unrepresentative sample of the telephone communications involving the members of the organization. This means that considerable caution must be exercised in reaching conclusions from the study.

Methods of analyses

Following Natarajan (2000), a series of separate but linked analyses were undertaken of the wiretap conversations to study different aspects of what was assumed to be an organization and of the roles of individual members within it. The analyses were as follows:

  1. 1.

    Contact analysis: For each identified individual in the database, counts were made of the number of other people with whom he or she had telephone contact. This led to the identification of 38 “core” individuals who had extended contacts and repeated transactions with others. The analyses listed below are mostly based on these 38 individuals.

  2. 2.

    Role analysis: The major tasks performed by each of the core individuals were identified through inspection of the content of conversations in which he or she was involved.

  3. 3.

    Status analysis: Using a coding guide developed from Natarajan’s (2000) study, the relative status of the core individuals was determined through an analysis of the conversations in which each was involved.

  4. 4.

    Network analysis: Links between core individuals and others were analyzed using a UCINET VI software program for social network analysis.

  5. 5.

    Geographic analysis: The addresses of outgoing telephone calls from the tapped phones were examined to see if this revealed any geographic specialization among the core group.

Each analysis was designed to build upon previous ones. Thus, the role analysis of core individuals in the organization builds upon the base line information provided by the contact analysis and was undertaken to clarify each core member’s organizational “niche”. Information provided by the status analysis was designed to improve understanding of the organizational structure (i.e. the degree to which it is hierarchical or egalitarian), the chain of command and the accountability of core members. The network analysis explores the nature of the connections between and among these core members, the existence of cliques, and the contacts between core and peripheral members of the organization. Finally, the geographic analysis explores whether there was any evidence that individual sellers confined their operations to a particular part of the city.

Findings

Contact analysis

In order to identify the core and peripheral members, a count was made using Folio Views’ query feature of the individuals participating in the 2,408 telephone conversations. In 557 of these conversations it was not possible to identify the people speaking. For the remaining 1,851 conversations, 294 individuals were identified by name. Of these 294, 208 (nearly 70%) were a mix of overseas contacts, street level buyers, and an assortment of friends, family and casual contacts, who spoke with only one other person in the database. The remaining 86 individuals spoke with 2 or more others (see Table 2). These 86 individuals were involved in more than half the conversations in the database (1,299) and spoke to 2–76 people.

Table 2 Telephone contacts for 294 identified individuals
Full size table

Table 2 suggests that the database falls quite naturally into two groups; one large group composed of individuals who spoke to only one other person in the organization and another smaller group with extended contacts. In fact, the smaller group proved to be quite heterogeneous when the number of conversations they were involved in was counted. On average each of these 86 individuals was involved in six conversations, but some were involved in many more than this, up to a maximum of 81. This suggested that many of the 86 individuals who had more than one telephone contact were in fact also peripheral members of the organization. It was decided to define the “core” members of the organization on the basis of a combination of number of contacts and number of conversations. On this basis, 38 core members were identified who had two or more contacts and were involved in five or more conversations.

There is only a moderate degree of overlap between these 38 core individuals and the 35 defendants in the criminal case: only 23 individuals both fell into the core group and were prosecuted. Thus, fifteen members of the core group were not prosecuted and of the 35 individuals prosecuted on criminal charges, nine were not in the core and another three were not in the database at all. Some of these differences were no doubt due to evidential requirements for prosecution. Thus, the prosecution might have been well aware of the importance of some of the core members not prosecuted, but might have been unable to muster the necessary evidence for successful prosecution. With respect to those prosecuted who were not in the core, the prosecution might have known about their importance in the organization from sources other than wiretap data—for example from direct surveillance or from informants.

Role analysis

The contact analysis provided a general description of the organization in terms of the number of people involved, the frequency of contacts and the size of the organization’s active “core” of 38 members. This next stage of the analysis explores the roles in the organization of these core members. In doing this, content analyses were undertaken of three sets of conversations: (i) those among the 38 core members (the 45 randomly selected conversations used in the “status” analysis reported below were used for this purpose); (ii) those between the core members and the other 48 members (i.e. 86 minus 38) with more than one contact (a random sample of 20 conversations; (iii) those between the core members and 208 members and 208 members with only one telephone contact (a further random sample of 20 conversations).

The content analysis revealed that the 38 core members performed one of four primary roles: sellers, brokers, retailers and secretaries/assistants. These are described in more detail below.

Sellers (N = 18)

This group consisted of individuals who were mostly involved in selling drugs in quite large quantities (a pound or more), but who were also involved to a more limited extent in brokering deals. Fourteen of the sellers were prosecuted. Some of the group had connections with overseas suppliers. Footnote 8 Sellers dealt primarily in heroin and were particular about the quality of drugs they dealt. They usually examined consignments before agreeing to purchase them and were sufficiently knowledgeable to make comparisons with products from other sources. Money transactions were very informal among the sellers despite the large sums often involved. These were referred to using code words. They frequently operated on credit and regularly lent each other supplies. Most were older and were friendly in their dealings. A few of the sellers sold drugs other than heroin in small quantities for retail sales to the 208 people with only one contact.

Retailers (N = 8)

Retailers bought exclusively from sellers. They were of all age groups and included some women. They knew the sellers on a friendly basis and there was relatively little bargaining on price. Code words were used in arranging deals. The price and amount quoted in conversations suggested that most deals involved between a quarter of an ounce to four ounces. Sometimes retailers asked for drugs by price rather than amount. Retailers were very particular about the quality of drugs. Many of them bought on a regular basis and tried to arrange deals with a variety of sellers as and when they obtained the orders. They generally made the pick-up arrangements in some specified location, say “69th on the West End”.

Brokers (N = 8)

Brokers were middlemen between sellers as well as between sellers and retailers. They set up meetings to inspect drugs and helped negotiate the price. Retailers sometimes complained to brokers about the quality of the drugs they purchased, which is why they took responsibility for assessing quality. Once an agreement was reached between retailers and sellers, the broker arranged a place for them to meet and conclude the deal. Brokers seemed to be relatively young with connections around the city, mainly among relatives. These connections often notified the broker of the arrival of supplies from out of town. The brokers passed on this message to sellers and made the arrangements for them to buy from the suppliers. Brokers were the lynch pins in the organization and played an important role in bringing people who did not know each other together to conduct business.

Secretary (N = 4)

These women were generally wives or girlfriends of the active members. They passed messages to buyers, sellers and brokers. They were well known to those involved in buying and selling.

Status analysis

In order to establish the relative status of these 38 core members, a content analysis of their conversations was undertaken using Natarajan’s (2000) six-point scoring scheme. This allows the relative status to be established of two individuals engaged in a single conversation, or conversational dyad. This coding scheme assumes that the higher status individual would generally:

  • (a) express satisfaction

  • (b) request information

  • (c) not provide information

  • (d) give orders

  • (e) not seek clarification of orders

  • (f) not use the word “sir” when talking to others.

For the individuals in each dyad, the numbers of segments were counted that contained instances of each of the six items above. A conversational “segment” is a continuous, uninterrupted utterance by an individual. This definition can be clarified by the following fictitious example with three segments:

  • Person 1: Hello! How are you?

  • Person 2: All right. And you?

  • Person 1: Can’t complain!

Reliability and validity of coding scheme

Before coding the conversations, the reliability of Natarajan’s guide had to be determined when used with the present data. For this purpose, a random sample of 20 conversations in which the 38 core members were involved was selected. Two coders then independently coded each segment of the 20 conversations. Correlation coefficients were calculated between the ratings made by each rater for both individuals involved in the 20 conversations. Each conversation thus had two correlation coefficients, one for each of the individuals speaking. The overall correlation between the two coders was 0.91, somewhat higher than in Natarajan’s original study (0.80), and indicating a satisfactory level of reliability.

If the coding scheme is a valid measure of status, its individual items ought be inter-correlated. In fact, in Natarajan’s original application of the coding scheme, five of the 15 correlations between items were significant. This was taken to indicate a reasonable degree of consistency in the measure. Because it could not be assumed that the measure would be equally valid when applied to a quite different type of organization, correlation coefficients were calculated for a random sample of conversations drawn from the 45 conversational dyads in which the 38 core members were involved.

The procedure was as follows. One conversation for each dyad was randomly selected Footnote 9 and each segment of the conversation was scored using the coding scheme. In fact, no conversational segment contained the item “Sir” Footnote 10 and this item was not included in the calculations. For the remaining five items, only two coefficients were found to be significant at the 0.5 level: between “request information” and “provide information” (0.23); and between “give orders” and “express satisfaction” (0.21). This was many fewer than in Natarajan’s (2000) study and one of these correlations (between “request information” and “provide information”) was in the opposite direction. In sum, “sir” was almost never used, few of the items were inter-correlated and there was a positive association between requesting and providing information.

This pattern of results suggests that the present organization is quite different from the highly structured corporate type organization, with a clear hierarchy of employees, which Natarajan studied earlier. Rather, the present organization seems to be much less hierarchical and more loosely structured. In such an organization the status scores of individual members are likely to show much less variation.

Results of the “status” counts

In fact, very little variation was found among the 38 core members of the organization in mean status scores. The range was 0–2 with an average of 0.97. This represents much less status variation than found by Natarajan (2000) in the “corporate” organization she studied, where the range was 1–5, with an average of 1.6. Of the 15 individuals who scored higher than 1.0, 13 were sellers (the remaining two were buyers), but none of these individuals scored so much higher than the remainder as to suggest he might be a boss, in charge of the entire group. In sum, though there was some level of status differentiation, with most of the sellers being of higher status than the remainder, the status analysis provides further evidence that the present organization has a flat, egalitarian structure with little differentiation between the core members.

Network analysis

The three previous analyses provided information about the nature and structure of the organization, about the relative status of individual members, and about the tasks they performed. However, the links among individual members are unclear and, in order to understand these, social network analysis was undertaken. Social network analysis is the mapping and measuring of relationships and flows between people (Borgatti et al. 2002; Borgatti and Everett 1992; Borgatti and Foster 2003; Berkowitz 1982; Breiger 2004; Burt 1976; Klerks 2001; Kilduff and Tsai 2003; Knoke and Kuklinski 1981; Moreno 1934; Scott 1991; Sparrow 1991; Wasserman and Faust 1994; Watts 1999; Wellman 1992). It considers primarily: the number of people; the proportion of people that are not connected to each other, the number of sub-groups or cliques, the strength of links and the degree to which power seems to be centralized on particular individuals.

Using the 2,408 conversations, sociomatrices were developed of contacts Footnote 11 among, (i) the 294 individuals in the organization, (ii) the 86 with more than one contact, and (iii) the 38 core members. These were analyzed using UCINET VI, a powerful software program (Borgatti and Everett 1999) to answer the following questions:

  1. 1.

    How cohesive is the organization?

  2. 2.

    Are there are any subsets or groups within the organization?

  3. 3.

    Are there any particular individuals who are more powerful than others?

Cohesiveness/density

Barnes’s set analysis was used to measure the degree of inter-relatedness or network “density” of contacts between individuals or teams (Ianni and Ianni 1990). This yielded the “percent density” values shown in Table 3.

Table 3 Density scores
Full size table

The percent density values for all three groups are well below 80%, which is the critical value suggested by Barnes (1972). This indicates that this is a sparse, loosely connected organization, with many isolated individuals and with relatively few contacts among the people in the organization. Sparsely knit networks provide people with considerable room to act autonomously and to switch between relationships.

Sub group/cliques

The software program can identify sub-groups, or cliques, with a minimum of three members, within the broader network. Because so many of the individuals in the network were isolates, sub-groups were sought only among the 38 core members.

Twenty-nine (29) cliques were identified when a clique was defined as having a minimum of three members; when defined as four members, eight cliques were identified; when defined as five members, no cliques were identified. This reveals that there were a number of small groups with considerable overlapping membership. In fact, 29 of the 38 core members belonged to at least one clique (when these were defined as having a minimum of three members). The nine individuals who did not belong to any clique were three retailers, two brokers, three sellers and one secretary.

The cliques fell into two broad groups, one consisting of 16 cliques associated with Mr. DD (a seller) and the other of 13 cliques associated with Mr. X, Mr. J, Mr. FF and Mr. F (all of whom were sellers). There were limited direct contacts between these groups, though both were separately connected to Mr. Y, which suggests that he also played a very important part in the organization. Indeed, he was recognized as an important figure by the prosecution. UCINET’s net draw procedure was used to see whether Mr. DD and Mr. X were “cut points”, Footnote 12 i.e. network members whose removal would cause the network to fall apart (see Bonacich 1987; Freeman 1979; Friedkin 1991; Wasserman and Faust 1994). However, when they were removed, the network structure did not completely fall apart but was maintained by links among the remaining members (Fig. 1).

Fig. 1
figure 1

Network structure of 38 core members

Full size image

The special case of sellers

Sellers were included in every one of the 29 cliques with a minimum of three members. In fact, 15 of the cliques were exclusively composed of sellers. There were no exclusive cliques of any of the other groups (broker, retailers or secretaries). While they do form cliques, Fig. 2 shows that the interconnections among the 17 sellers (excluding Mr. EE) and five of the eight brokers (not including Mr. R, Ms. Q and Mr. P) were quite extensive. Together with their common role, this fact suggests that the sellers and half the brokers might have comprised a separate, coherent organization that, despite the variety of business links with other members of the “core” group, was somewhat independent of it. However, the low density score (14.46) for this group of 22 sellers and brokers suggests that even if the organization existed, it was not very tightly knit and might be more accurately termed a network.

Fig. 2
figure 2

Sellers and brokers network (N = 26). N.B. Shaded circles denote brokers

Full size image

Power scores

Power in social networks may be viewed either as a micro property (i.e. it describes relations between actors) or as a macro property (i.e. one that describes the entire population). Bonacich’s power analysis (1987) takes into account any single individual’s connections in relation to the connections of the other individuals in the group. It connects the individuals to the structure of the network facilitating the micro-macro link (Markovsky et al.,1988, 1993). It assumes that power arises from connections to weak others, as opposed to strong others. According to Bonacich (2001), some actors have more transaction partners than they can possibly transact with; they must choose some and reject others. Power accrues from having potential partners who have no alternative choices.

The mean power score for the core members was 4.56, with fourteen of the 38 achieving scores higher than the mean. Mr. DD and Mr. X obtained the highest scores (respectively, 21 and 18) which was consistent with the sub groups analysis discussed above. Ms. CC (Mr. DDs secretary/girlfriend) also scored highly (10), as did Mr. Y and Mr. FF (both 8), Mr. BB (7) and Mr. J (6). With the one exception of Ms. CC, all these individuals were sellers and all were identified as important players in the earlier analyses.

Links between street level and mid level operations

The flow of drugs from mid level to the street can be illustrated by the personal (“egocentric”) network of Mr. M, one of the sellers (See Fig. 3). Mr. M. was based in Manhattan and obtained drugs through a variety of sources: from Mr. AA (who in turn obtained them from Mr. DD), and from Mr. F, Mr. J and Mr. HH, core members who sold directly to him. Mr. M was connected to 53 of the 208 peripheral individuals with only one contact. Inspection of a sample of these contacts found that these 53 individuals seemed to be a mix of acquaintances and street level dealers.

Fig. 3
figure 3

Links between street level and mid level operations

Full size image

Geographical analysis

The geographic analysis explores whether there was any evidence that individual sellers confined their deals to particular part of the city, as would be expected if they belonged to a single organization. If no evidence were found of any geographic specialization, this would suggest that the sellers operated independently of one another.

Only a few of the conversation transcripts (a total of 172 records) enabled the addresses of the numbers that were called to be identified. The great majority of these addresses were within the five boroughs comprising New York City. Two thirds of the calls were made to addresses in high/medium drug arrest areas (identified from precinct arrest data for 1993) in the respective boroughs—showing again the probable links between midlevel and street level operations. (One central player in the organization, Mr. Y, actually resided in a high drug arrest area). Footnote 13

The data permitted a limited geographical analysis of the phone contacts of the two most active sellers—Mr. DD and Mr. X. For the 284 phone calls made from Mr. DD’s tapped phone, 51 listed addresses and for the 250 phone calls made from Mr. X’s tapped phone, 26 listed addresses. Table 4 shows the geographic distribution of the addresses for Mr. DD and Mr. X’s phones.

Table 4 Geographic distribution of calls made from the phones of the two most active sellers
Full size table

It can be seen there is no significant difference in the distribution of the addresses called, at least for this crude geographic differentiation, and for the small perhaps unrepresentative sample. Nonetheless, it provides little reason to think that there was an agreement either formal or informal between these two key sellers to serve different parts of the city. Instead, it would appear that they found customers and made deals wherever they could and, generally, the analysis is consistent with the picture that has emerged from the other analyses reported above: The organization seems to have been very loosely structured, with few organizing principles, and with individual operators who formed ad hoc relationships with a handful of others as needed to conduct deals.

Summary of findings

The contact analysis found that most of the 294 identified individuals had very limited contacts with anyone else in the organization and that the active core was quite small, consisting of 38 individuals. The role analysis of this core found that they could be sorted into one of four groups: sellers (18); retailers (8); brokers (8) and secretaries (4). However, these individuals sometimes switched roles in the furtherance of particular deals. The status analysis showed that there was little status differentiation among the core members and the organization had a flat, egalitarian structure. The network analysis complemented this picture by characterizing the organization as sparse and loosely connected, with many isolated individuals and with relatively few contacts among those in the organization. Among the 38 core members, there were a number of small sub groups, or cliques, dominated by sellers, with considerable overlapping membership. These sub groups fell into two broader groupings, associated with two very active sellers (Mr. DD and Mr. X). There were few direct contacts between the two halves, but both were separately connected to another important seller (Mr. Y), who thus seems to have played a central role in the organization. In addition, the network analysis suggested that the sellers and half the brokers constituted a coherent, more tightly connected network of 22 individuals that was somewhat distinct from the rest of the core. Finally, though it was severely constrained by lack of data, the geographic analysis found little evidence of geographic separation in the operations of key dealers.

Discussion and conclusions

The analysis reported above using wiretap data collected in the investigation of a large heroin trafficking case in New York City was undertaken with two linked objectives. The first was to fill out the picture of what the prosecution had successfully argued was a large international drug trafficking conspiracy, dealing in more than 200 kg of heroin each year. It was claimed that the 35 criminal defendants had controlled fifteen narcotics distribution groups, reaching from street level sellers through to middle level wholesalers and high-level traffickers with close links to organized crime groups. In Natarajan and Belanger’s (1998) terms, the prosecution’s description loosely fits the “communal business” type of drug dealing organization, albeit a large one, composed of small cliques of people forming a larger entity, with a flat hierarchy and a flexible division of labor. On the other hand, no drug trafficking organization with such a large span of operations, whether “communal business” or not, has previously been described in the research literature and it is possible that, rather than having identified a discrete drug trafficking organization, the prosecution had identified a segment of the heroin market in New York City. The second objective of the analysis was therefore to see what evidence could be found to support either this interpretation or the prosecution’s view of the case.

It might have been expected that the picture of the organization provided by the analysis reported in this paper would differ considerably from the prosecution picture. The quantitative analysis was based only on the wiretap data presented (which suffers from the various limitations discussed above), while the prosecution drew on much wider sources of information. In many respects, however, such as the large extended operations, the relatively flat hierarchy, and the division of the organization into multiple drug dealing cells, the wiretap analysis supported the picture presented by the prosecution. The analysis also filled out the prosecution’s picture by providing considerably more information about certain aspects of the organization, including its large size (294 individuals were identified in the wiretap database); the relative lack of status differentials among members; the frequency with which they switched roles; the loose organizational structure; the many overlapping cliques within the organization; the number and nature of contacts between a relatively small core group and the large number of peripheral members; the wide geographic extent of operations within New York City; and the links between street selling and wholesale distribution. Footnote 14

However, there were some important differences between the two pictures, including only a moderate degree of overlap among the 35 individuals prosecuted and the 38 identified in this paper as falling into the “core” of the organization. Furthermore, the wiretap analysis provided no information concerning the involvement of organized crime groups, nor did it properly reveal the extent and nature of the contacts with overseas suppliers—features that would have been important for a successful prosecution. As explained, these differences between the prosecution picture and that revealed by the wiretap analysis could have resulted from evidential requirements for prosecution and from the additional information available to the prosecution about the organization.

It is impossible to make a definitive judgment from these results whether the prosecution had identified a large dug trafficking organization with a very wide span of operations or merely a segment of the drug market in New York City. Perhaps the most likely interpretation is that the 294 individuals identified in the wiretap analysis constitute a segment of the drug market. However, among the 38 “core” members was a smaller group of 22 sellers and brokers, who were interconnected in a loose network and who were somewhat independent of the wider core. Deals made among them were characterized by a high level of trust and it is tempting to argue that it is this group which comprised a “communal business” of the kind identified by Natarajan and Belanger (1998). Even though this group interacted frequently with some of the other members of the core, these latter might not have been true members of the communal business.

While some of the key players in the organization appear to have had links with organized crime families, the organization as a whole did not fit the traditional organized crime model, Instead, as explained, the picture that emerges is one of a loosely structured network of all groups or cliques, with little or no hierarchy. This reinforces the view advanced in recent publications (Pearson and Hobbs 2001; Reuter 2004) of drug dealing—and indeed of organized crime in general (see Introduction)—as a highly fragmented business, consisting of a large number of entrepreneurial groups separately engaged in exploiting the lucrative opportunities presented by the demand for drugs. As mentioned in the Introduction, these kinds of drug dealing enterprises would be described as “network organizations” (Van Marshall 1997) in the business literature. Burt (1992) characterizes large, low density/low hierarchy networks, such as the present one, as “entrepreneurial” organizations which are composed of entrepreneurs who bridge a number of diversified, unconnected groups. Footnote 15 According to transitivity theory (Granovetter, 1983, 1985, 1995) a large organization with weak ties, like the present one, provides individuals with greater business opportunities and access to resources. Weak ties permit change and mobility, both crucial for sustaining a drug dealing enterprise when core members are arrested. Even if law enforcement dismantles the core of a large network, the extended connections at the periphery with importers and with organized crime permit its regeneration with a new cast of core actors.

While this picture is consistent with a broad range of research, the findings reported here are also consistent with an alternative, untested model. It might be that the “organization” described in this paper had no real structural existence beyond that imposed on it by the actions of law enforcement. It may be that the individuals and cliques identified in the analysis were all independent operators, owing no particular allegiance to other individuals or cliques. They might simply have been exploiting what they regarded as opportunities that came their way in the ordinary course of business or opportunities that they created to make some money. While they might routinely have done business with some other individuals or groups in the “organization”, many of them might just as regularly have done business with other individuals or groups quite outside the borders of this particular organization. Indeed, the “organization” might have been defined only by the resource limitations of a law enforcement operation. This raises the intriguing possibility that, in reality, it might be incorrect to think of drug dealing, at least in a large city like New York, as being conducted exclusively or even primarily by a relatively small number of differently structured organizations. These have been identified in past research and they might continue to exist (there is no way of knowing this given the lack of contemporary studies). On the other hand, it might also be the case that drug dealing has been evolving in response to changing opportunity structures for this type of business brought about by diversification of supplies, proliferation of substances and expansion of trafficking routes and methods. These changes would make it easier for drug dealing to be conducted by large numbers of entrepreneurial individuals or small groups, who forge business deals with first one entity and then another as opportunity dictates. When law enforcement dips its ladle into this bubbling broth of entrepreneurial activity it might scoop out a random collection of small groups and individuals that, by its network of contacts, is defined as a criminal organization.

While this is mere speculation at this point, the implications for criminological understanding of drug trafficking and its prosecution are so important that it should be tested in further research. The view that drug trafficking is primarily conducted by small groups of entrepreneurs who almost randomly come together to conduct particular deals and then disperse, perhaps to reconvene at a later date to conduct a new deal with some other entity, presents law enforcement with a difficult challenge. It is much harder to hit a moving target than a static one, such as that presented by a large criminal organization. Furthermore, there are many advantages for law enforcement in portraying themselves as engaged in a fight against a powerful, highly organized enemy. This attracts resources and media attention. Fighting loose networks of opportunist entrepreneurs is not only more difficult, but is less glamorous and somehow less worthwhile.

Finally, this study supports Natarajan’s (2000) contentions about the utility of wiretap data, with all its limitations, for analysis of drug dealing operations. It shows how modern software programs make it possible to conduct rigorous quantitative analysis of textual data, and how wiretap data can be used to analyze the network structure of illegal organizations. It also confirms the generally robust nature of the methodology she described, despite the limited value of the status analysis in discriminating among core members of the organization which, as argued above, was probably the result of the flat egalitarian structure of the enterprise studied.