This paper focuses on phone conversations wiretapped by the police. It discusses issues of validity and reliability of this type of data and it proposes the use of a combination of data analysis techniques. In order to utilize wiretapped conversations in a valid manner, individuals under surveillance must talk freely on the phone, the coverage of the group must be reasonably wide, and a large enough sample of conversations must be available. As for the analysis, we propose the use of a set of techniques: content analysis, correspondence analysis, descriptive network analysis and longitudinal stochastic actor-oriented models. Each technique highlights a different aspect of the criminal network. Systematic analysis of phone conversations can yield valid inferences on the nature and activities of criminal groups and enrich the understanding of the ties within a criminal network. If followed, the procedures discussed here should facilitate comparisons across groups.
This is a preview of subscription content, log in to check access.
Buy single article
Instant access to the full article PDF.
Price includes VAT for USA
Subscribe to journal
Immediate online access to all issues from 2019. Subscription will auto renew annually.
This is the net price. Taxes to be calculated in checkout.
The popularity of the SNA is also reflected in the increasing amount of definitions of ‘Organized Crime’ that contain the term ‘network’ (Varese 2010: 7–8).
In several jurisdictions (e.g. the United States, Canada, Germany, France, Italy, Holland and Sweden), wiretaps can be introduced as evidence in trials. Since they are to be used as evidence, the conversations are transcribed by officers and made available to the prosecutor. On the contrary, in the UK, wiretaps cannot be used as evidence in court, thus scholars have no access to them. Still, this information is used extensively as part of police investigations: 2,243 warrants to intercept communications had been issued in the UK between January and March 2006. Interestingly, police forces in the UK do not transcribe the conversations, leading to several mistakes made by investigators, as reported by a Government review (Reuters News Agency 20/02/2007).
It can also be argued that when criminals do not talk openly, they might talk even more freely since they do not expect the police to be able to decrypt their conversations.
Mistaken interpretations by the police may occur. The extent to which this happens depends on the quality of training the police forces get. In this regard, there is little that a researcher can do apart from disregarding any set of conversations where the wiretapping activity is clearly invalidated by lack of preparation on the police side (this should emerge during the trial). Yet, based on our and other authors’ experience (see for example Morselli 2009: 43), this is seldom the case with large and medium scale police investigations, which are often conducted by special units. Furthermore, the impact of such inaccurate interpretations on the overall results can be lessened by applying a coding scheme based on broader topics.
If there is strong evidence that the police focused on individuals suspected of particular kinds of crime leaving out all the other members, it is better not to rely on that corpus of conversations (see also note 10).
It is difficult to determine a minimum threshold that has to be generally met, since this threshold is not only a function of the data analysis techniques to be used, but also of the number of network nodes, the duration of the intercept operation and the coding procedure.
It is not uncommon to find un-usable set of conversations in court archives. TM (1994) is such an example. The investigation focused on an Italian mafia group linked with an `Ndrangheta family who run various types of businesses in a small town near Milan. The conversations wiretapped were less than 30 and focused only on drug dealing, and only a few actors were put under phone surveillance. A quantitative analysis of the conversations would have led to a biased picture of the group.
Investigators selectively transcribe only the conversations that are related to a criminal activity, and therefore the only network that we are able to reconstruct is the criminal one. Yet criminals – as all the other social actors – may be part of a number of different networks at the same time, and some of these networks may also be particularly helpful in understanding the structure and evolution of the criminal one. Whilst some extra information may be collected from other sources and still integrated into the analysis, it is true that our purposive criterium does not allow us to reconstruct any other network but the criminal one.
It should be borne in mind that a criminal group may well be part of a bigger network of lawbreakers related to a specific criminal industry, e.g. drug trafficking, and that members of the criminal group may be in contact with other criminals active in the same ‘sector’. Only a substantive criterion, based for instance on the content of wiretapped conversations or other police/court files, may help the researcher to establish the boundaries of the criminal group under scrutiny at a specific point in time, and assess whether the boundaries identified by the police can be accepted without major concerns.
As Berelson (1952: 173) points out, there could be an underestimation of reliability when the latter arises ‘from the measurement of reliability on detailed categories which are later subsumed into more general categories’.
If one does choose this path, the unit of analysis becomes the single conversation, which would be akin an ‘item’ in the Berelson (1952) and Holsti (1969) classification of units. The use of different units of analysis within the same study is recognized also by Berelson (1952), when he stressed that “there is no reason […] why a particular study must use only one of the possible units of content analysis. The choice of the appropriate unit depends upon the problem and the content under investigation, and this may necessitate the use of different units within the same study” (1952: 143).
It may be the case that a theme is nested into deceptive preliminaries and digressions, making difficult to quantify the correct number of words. Also for this reason, it is not possible to undertake an automatic coding procedure. Thus, the manual coder must be highly trained in order to deal with such coding problems.
For an introduction to reliability coefficients see Krippendorff (2004, ch. 11).
To be more precise, correspondence analysis is a family of techniques based on the singular value decomposition (SVD) algorithm. Simple correspondence analysis (SCA), multiple correspondence analysis (MCA) and Homogeneity analysis (HOMALS) differ in respect to the type of input matrix: SCA uses a simple bivariate crosstab, MCA a Burt matrix containing a set of bivariate crosstabs while HOMALS is based on an Objects-by-Variables matrix (where the variables are dummy variables derived from K-polytomies).
Given the chi-square distance’s feature of increasing the relative contribute of the components with lower masses, it is better to be very careful when modalities with very low mass are included into the analysis.
Hierarchy could of course be defined differently, for instance as a structure embodying relations of authority and subordination. Such a concept could be operationalized by looking for items in a conversation that would suggest the relative status of the speakers (Natarajan 2000). Giving orders, expressing satisfaction, and requesting information would indicate a high position in the informal hierarchy of the group (Natarajan 2000). Provided one has such information from the conversation, a content analysis can be undertaken.
The notable exception is the so-called QAP procedure that regresses one or more independent matrices on a dependent matrix, and assesses the significance of the r-square and regression coefficients (the procedure is implemented in UCINET software. See Borgatti et al. 2002).
The actor-oriented models implemented in SIENA software have some limitations. For instance, these models do not produce a measure similar to the reproduced variance and it is not possible to compare statistics estimated by SIENA with statistics calculated via other statistical techniques (Snijders et al. 2007; Burk et al. 2007: 403).
Amorim D, Saul G, Barthélemy J-P, Ribeiro CC (1992) Clustering and clique partitioning: simulated annealing and tabu search approaches. J Classif 9(1):17–41
Baker WE, Faulkner RR (1993) The social organization of conspiracy: illegal networks in the heavy electrical equipment industry. Am Sociol Rev 58(6):837–860
Benzécri J-P (1973) L’Analyse des Données, Tome 2. Dunod, Paris
Berelson B (1952) Content analysis in communication research. Free Press, New York
Borgatti SP, Everett MG, Freeman LC (2002) Ucinet 6 for windows. Analytic Technologies, Harvard
Bruinsma G, Bernasco W (2004) Criminal groups and transnational illegal markets. A more detailed examination on the basis of Social Network Theory. Crime, Law Soc Chang 41(1):74–94
Burk WJ, Steglich CEG, Snijders TAB (2007) Beyond dyadic interdependence: actor-oriented models for co-evolving social networks and individual behaviours. Int J Behav Dev 31(4):397–404
Burt RS (1992) Structural holes: the social structure of competition. Harvard University Press, Cambridge
Campana P (2011). Eavesdropping on the Mob: the functional diversification of the Mafia activities across territories. Eur J Criminol 8(3):1–16
Coles N (2001) It’s not what you know—It’s who you know that counts. Analysing serious crime groups as social networks. Br J Criminol 41(4):580–594
de Nooy W, Mrvar A, Batagelj V (2005) Exploratory social network analysis with Pajek. Cambridge University Press, Cambridge
Dijkstra JK, Lindenberg S, Veenstra R, Steglich C, Isaacs J, Card NA, Hodges EVE (2010) Influence and selection processes in weapon carrying during adolescence: the roles of status, aggression, and vulnerability. Criminology 48:187–220
Finckenauer JO, Waring EJ (1998) Russian Mafia in America: immigration, culture and crime. Northeastern University Press, Boston
Freeman LC (1979) Centrality in social networks: conceptual clarification. Soc Netw 1:215–239
Gifi A (1990) Nonlinear multivariate analysis. Wiley, Chichester
Greenacre MJ (1984) Theory and application of correspondence analysis. Academic, London
Hanneman RA, Riddle M (2005) Introduction to social network methods. University of California, Riverside. At: http://faculty.ucr.edu/~hanneman/nettext/
Harcourt BE (2002) Measured interpretation: introducing the method of correspondence analysis to legal studies. Univ Ill Law Rev 979–1017
Hill MO (1974) Correspondence analysis: a neglected method. Appl Stat 23(3):340–354
Hirschfeld HO (1935) A connection between correlation and contingency. Proc Camb Philos Soc (Math Proc) 31:520–524
Holsti OR (1969) Content analysis for the social sciences and humanities. Addison-Wesley Publishing Company, Reading
Horst P (1935) Measuring complex attitudes. J Soc Psychol 6:369–374
Klerks P (2001) The network paradigm applied to criminal organisations: theoretical nitpicking or a relevant doctrine for investigators? Recent developments in the Netherlands. Connections 24(30):53–65
Knoke D, Rogers DL (1979) A blockmodel analysis of interorganizational networks. Sociol Soc Res 64:28–52
Krippendorff K (2004) Content analysis: an introduction to its methodology, 2nd edn. Sage, Beverly Hills
Lauman EO (1991) Comment on “The future of bureaucracy and hierarchy in organizational theory: a report from the field”. In: Bourdieu P, Coleman JS (eds) Social theory for a changing society. Westview, Boulder, pp 90–93
Lauman EO, Marsden PV, Prensky D (1983) The boundary specification problem in network analysis. In: Burt RS, Minor MJ (eds) Applied network analysis. Sage, Beverly Hills, pp 18–34
McNally D, Alston J (2006) Use of Social Network Analysis (SNA) in the examination of an outlaw motorcycle gang. J Gang Res 13(3):1–25
Mohr JW (1998) Measuring meaning structures. Annu Rev Sociology 24:345–370
Mohr JW, Duquenne V (1997) The duality of culture and practice: poverty relief in New York City, 1888–1917. Theory Soc 26:305–356
Morselli C (2005) Contacts, opportunities, and criminal enterprise. University of Toronto Press, Toronto
Morselli C (2009) Inside criminal networks. Springer, New York
Natarajan M (2000) Understanding the structure of a drug trafficking organization: a conversational analysis. In: Natarajan M, Hough M (eds) Illegal drug markets: from research to policy. Crime preventions studies, Vol. 11. Criminal Justice Press, Monsey, pp 273–298
Natarajan M (2006) Understanding the structure of a large heroin distribution network: a quantitative analysis of qualitative data. J Quant Criminol 22:171–192
Pattison PE, Breiger RL (2002) Lattices and dimensional representations: matrix decompositions and ordering structures. Soc Netw 24:423–444
Podolny JM, Page KL (1998) Network forms of organizations. Annu Rev Sociology 24:57–76
Reuter P (1994) Research on American Organized Crime. In: Kell R, Chin K, Schatzberg R (eds) Handbook of organized crime in the United States. Greenwood, Westport, pp 91–119
Richardson M, Kuder GF (1933) Making a rating scale that measures. Pers J 12:36–40
Roberts CW (ed) (1997) Text analysis for the social sciences: methods for drawing statistical inferences from texts and transcripts. Erlbaum, NJ
Robins G (2009) Understanding individual behaviours within covert networks: the interplay of individual qualities, psychological predispositions, and network effects. Trends Organ Crime 12(2):166–187
Robins G, Kashima Y (2008) Social psychology and social networks. Asian J Soc Psychol 11(1):1–12
Schlegel K (1984) Life Imitating Art: Interpreting Information from Electronic Surveillances. In: Fairbank JK (ed) Critical issues in criminal investigations. Criminal Justice Press, Cincinnati, pp 53–61
Scott J (2000) Social network analysis. A handbook, 2nd edn. Sage, London
Smith DC (1975) The Mafia mystique. Basic Books, New York
Snijders TAB (2001) The statistical evaluation of social network dynamics. Sociol Methodol 31:361–395
Snijders TAB, Steglich CEG, Schweinberger M, Huisman M (2007) Manual for SIENA version 3.1. University of Groningen: ICS / Department of Sociology and University of Oxford: Department of Statistics
TM(Tribunale di Monza) (1994) Atti del procedimento contro Sansalone Antonio (Proceedings against Sansalone Antonio), Monza Penal Court
Torgerson WS (1952) Multidimensional scaling: I. Theory and method. Psychometrika 17(4):401–19
Varese F (2006) The structure of a criminal network examined: the Russian Mafia in Rome. Oxford Legal Studies Research Paper 21. University of Oxford, Oxford
Varese F (2010) General introduction. What is organized crime? In: Varese F (ed) Organized crime. Routledge, London and New York, pp 1–33
Varese F (2011) Mafias on the move. Princeton University Press, Princeton
von Lampe K (2009) Human capital and social capital in criminal networks: introduction to the special issue on the 7th Blankensee Colloquium. Trends Organ Crime 12(2):93–100
Wasserman S, Faust K (1994) Social network analysis. Cambridge University Press, Cambridge
Weber RP (1990) Content analysis, 2nd edn. Sage, Beverly Hills
The authors are grateful to John Goldthorpe, Michelle Jackson, Luca Ricolfi, the two anonymous referees and the editor for their valuable comments and suggestions. They are also thankful to Liz David-Barrett and Morag Henderson for their help in editing the text, and their comments. This research was supported by a grant from Leverhulme Trust (F/01 532/B).
Authors are listed in alphabetical order.
About this article
Cite this article
Campana, P., Varese, F. Listening to the wire: criteria and techniques for the quantitative analysis of phone intercepts. Trends Organ Crim 15, 13–30 (2012). https://doi.org/10.1007/s12117-011-9131-3
- Criminal groups
- Wire tapped conversations
- Content analysis
- Social network analysis