György Csepeli and Gábor Csere Inequalities and Networks in Society 1. Paradigm Change in Quantitative Sociology

Transcription

1 György Csepeli and Gábor Csere Inequalities and Networks in Society 1. Paradigm Change in Quantitative Sociology The technological innovations of the Information Age have radically changed the functioning of society on both the life-world and systemic levels. The degree and pace of change clearly imply that change has enormous reserves. Over time no element of man s natural and socially constructed reality will be left unaffected by the radical effects of the technology for creating, transmitting and storing information. The culmination of these effects is the information society, which can be hardly understood through the exploration methods developed before the Information Age. Empirical sociology was created in the twentieth century in response to the ideological and sociopolitical challenges of the industrial society. The critical potential of sociology drew on the concept of a perfect society. Since the social reality experienced had never met this concept, knowledge gained through sociological theories and methods about society could serve as fertile soil for plans to enhance society and for concepts born under the sign of modernizing reforms. Whether politicians got to implementing these plans and reforms or dropped them, sociology had never been able to keep up with the processes occurring within society. The image which sociologists developed about society based on their research work could never be utilized. Quantitative sociological research relied on the collection and analysis of data through samples which were generated according to principles described by mathematical statistics. The different samples were based on a universe with varying status over time, making it difficult to account for the time dimension which is essential to understand the functioning of society. Moreover, data processing took a long time, and thus a research report could easily be published in a journal of history rather than a sociological review. Modernist presumptions implied normative and universal rules which were thought to be applicable to the largest possible populations, whose examination seemed to require only operationalizing and measuring a relatively small number of independent and dependent variables which were held a priori relevant. The sociologist created order in the universe of measured variables through applying multivariate means of mathematical statistics. Quantitative sociology responds properly to the challenges of social changes triggered by the information society if it attempts to utilize in its investigations the same technological changes which brought forth the information society. As information conditions are changing, no area within the operation of society will remain where generated information are not retained intactly. Data generated in the course of various social operations are continuously placed in data storages and can be retrieved for the purposes of comprehensive as well as sample-based analyses. The time-span between data generation and analysis will be considerably shorter, while changes occurring over time will also be retained and ready for analysis. All of the above changes are based on the possibilities offered by digital information management technology, which also allow researchers to have an immediate and interactive contact with the people who entrust a survey to them. When defining the paradigm of the new quantitative sociology, it is reasonable to start from the change in the nature of databases and our attitudes to these databases. However, a change 1

2 in the empirical foundations can result a complete paradigm change only if theoretical starting points are also altered. Access to digital databases for research purposes is aptly called data mining. Theoretically, there is no a priori truth, which the analyzer has to prove in order to justify his critical notions. The regularities sought can be applied to a great number of smaller populations. The relationship between dependent and independent variables develops dynamically. Order is a status of chaos, which is fragile and completely unintelligible. Consequently, the information provided by the new quantitative sociological paradigm is not aimed at politicians, intellectuals and popular leaders but decision-makers in the spheres of public policy and business, who are interested in running efficiently the organizations under their management. 2. A Methodological Experiment: Network Analysis through a Telecommunications Database In this article we will not attempt to describe the variety of effects that IT devices have on society. We aim only to apply the new paradigm of quantitative sociology, in the course of which we will try to describe a new empirical model of social stratification. In the context of the information society, the possession of information and network capital gained importance. They are latent forms of capital which can be examined only through their communication aspects rather than directly. Telecommunications companies, such as Matáv, possess an asset of enormous value: the matrix of their customers social network, which allows us to discover the real, multidimensional structure of society. The inputs for this new model of social stratification were drawn from call records archived in the data files for calls initiated and received by the whole population of Matáv s residential customers over its telephone network in March 2004, which the company provided for us retaining their anonymity. We tried to sort the immense pool of data generated from the traffic in March through a clustering procedure. A program was applied which could, without any intervention, group calls based on reciprocity into sets of customers who have strong ties with each other, while customer who typically did not call each other were excluded from these sets. In the course of this grouping process the program cut up the whole customer base at minimum points where ties were the weakest, and in turn divided the resulted sets again and again. The strength of a tie between two customers were measured through the sum of money spent on calls by the two customers, but the duration and number of calls could also be used as a measure. Clustering was carried out in a hierarchical fashion, from top down. Cutting up into sets was repeated until the number of customers in each did not exceed 100. The dimension number of the call-graph, which originally equaled the number of customers initiating or receiving calls, was reduced through the Singular Value Decomposition procedure into a 3-15 dimensional space, which was clustered through a simple K-means algorithm. Experience gained so far suggests that customers who call each other are extremely similar, i. e., they tend to have contacts with customers who are similar to them: they usually live close to each other, often have the same line type and subscription package, and spend more or less the same amount of money on telecommunications. When mapping this social network with the procedure above, we found homogeneous groups, which means that customers can be sorted into a set of homogeneous groups through the interpretation of their networks. 2

3 Line-based telephone is not an individual device, since measurable traffic reflects the combined activities of household members, and thus our database helps examine networks on the household rather than the individual level. Keeping contact through telephone lines is only one of several contact forms which develop between the members of a society. Therefore the database itself does not allow us to conclude how the social network explored through telephone use relates to the full or real network. However, we can get closer to this vigorously functioning real network within society, whose members communicate face-toface, via the Internet, mobile phones as well as telephones, if we also include other databases (of mobile phone, Internet and use) into our analysis. Obviously, the network outlined on the basis of usage data for a wide spectrum of communication devices will be more complex than the one we present below. Figure 1: Social Networks on Level 1 of the Clustering Routine Lakossági ügyfélkör A csoport B csoport C csoport [Residential Customers; Group A, Group B, Group C; Groups of Matáv s Residential Customers by Connection Distance (March 2004, sum spent on calls; 1,000, , ,000] First, the program divided the total of Matáv s residential customers into three groups. In the figure above the number of subscribers is indicated by the size of the pie charts which belong to various localities. Naturally, we had no information about areas outside the Matáv service domain (LTO), and thus no diagram is shown for such localities. In each locality, the proportion of customers who belong to the same communication network is marked by the same color. 3

4 There is no alternative to group membership: the proportion of those who potentially constitute a communication island (e. g., customers who call each other only) and are not linked to any of the three customer groups is insignificant. The sharpest difference from other groups was found for the group whose members live in the capital or nearby settlements. Of the data recorded by Matáv, Internet use has the greatest explaining power in terms of statistics: a customer using the Internet is likely to be the member of Group A, shown in red in Figure 2. This figure also implies that the new medium (so far) tends to increase rather than reduce existing social inequalities. Figure 2: The Social Network of People Living in the Capital or Nearby Settlements and Internet Penetration [Percentage of Internet Use (Dial-up and ADSL) by Matáv s Residential Customers] The boundaries of the two other groups are much less marked. The members of the two networks of people living in the countryside mingle. The strongest organizing principles of group membership are, at least on the first level of clustering, topographic proximity and Internet use (which is, in our assumption, one of the indicators of the level of economic, social and cultural development for a given settlement). Due to the hierarchical clustering routine, the program further divided the resultant groups in the subsequent steps. If we examine these groups, we can conclude that the organizing principles of group membership change on the deeper levels of clustering: topographic proximity is replaced by increasingly softer characteristics, such as settlement type, account size, family structure, line type, subscription package, etc. 4

5 In terms of account size, a great difference can be found between the three resultant networks on the first level of clustering. On he second level of this procedure it seems that increasingly homogeneous groups are produced by account size, which implies the communicative segregation of households depending on the account size (and, indirectly, socioeconomic status). Customers who spend much on telephoning tend to speak to each other, as customers who spend less do (only less frequently). Regional segregation cuts across status-related communicative segregation. It is also supported by the result that even when two households have the same account size, communication via telephone will occur between them only if they belong to the same greater cluster. Figure 3: Social Network and Account Size on Level 3 of the Clustering Routine [Capital Hungary, Large Cities, Small Towns, Rural Communities; Average Account Size (HUF), Number of Calls Between Residential Customers (March 2004); High Low; Level 1, Level 2, Level 3] Work on the deeper and more complex analysis of these data is currently in progress. However, it is already obvious at this stage that we are discovering a new world of organizing principles and rules which provide a foundation for the developed networks. This stage of analysis already implies that membership in major social groups is exclusively determined by sociodemographic variables, as it is assumed in classical theories of stratification. Although communication which serves as a basis for group formation is not entirely independent of social status, it also shows the signs of autonomy. It would be important to know the forces affecting the kind of communication which facilitates autonomous group formation because it would allow us to build a more accurate image of how the systemic operations of the economy, media, culture, politics and ideology penetrate or, in Habermas phrase, colonize the life-world. 5

6 3. Prospects Today we are still at the dawn of the Information Age. This statement is particularly true in the case of Hungary, where the spread of the new medium, the Internet, is in its second phase, when the followers of an experimenting avant-garde begin to discover this innovation. The diffusion of Internet use within society will inevitably accelerate. Particularly, broadband access constitutes a medium through which all earlier media can reach users. A new world of a communicating society evolves right before our eyes, which, using the platform of the Internet, permeates and transforms all segments of social life. The electronic dimension provides an opportunity for all traditional social functions education, government, trade, work, religion, sex, cultural activities, politics, etc. alike. Interpersonal communication will be the single link capable of updating group membership and identity. If we want to understand the new patterns of social organization, we have to extend our investigations to the databases which contain the logs of modern information technology. The study of society performed in this way opens new perspectives for the application of social research findings as well as social theory. Theory and practice will be able to converge eventually. Literature Castells, M The Rise of Network Society. Blackwell Publishers. Castells, M The Power of Identity (The Information Age). Blackwell Publishers Csepeli, Gy., and G. Prazsák, Paradigm Change in Sociology. Review of Sociology 10, No. 2 (2004): Dessewffy, T Bevezetés a jelenbe. Nemzeti Tankönyvkiadó. Habermas, J Theorie des kommunikativen Handelns. Bde Frankfurt am Main: Suhrkamp Verlag. Nyíri, K Enzyklopädisches Wissen im 21. Jahrhundert. In Vernetztes Wissen: Philosophie im Zeitalter des Internets. Wien: Passagen Verlag. 6