Network Analysis: Methods for Solving Real-life Problems
From 24th to 28th July, Moscow hosted the Eighth International summer school 'Theory and Methods of Network Analysis' for students and researchers, held by the HSE International Laboratory for Applied Network Research. Its academic supervisor Stanley Wasserman from Indiana University Bloomington took part in the summer school's work.
‘Network analysis differs from conventional statistics in that it does not assume independence between observations. On the contrary, where conventional statistics are powerless, network analysis seeks to find relationships, as a minimum, between pairs of interrelated observations,’ explained Valentina Kuskova, Head of the International Laboratory for Applied Network Research (ANR-Lab). ‘In addition to examining units for which statistical data is available, we also look for connections between them. Network-based methods can be more effective than conventional analysis, since a relational approach to data analysis may lead to a more comprehensive picture of the analysed phenomena, be it bloggers' influence on their subscribers or society's perception of a new government initiative.’
The summer school programme included theoretical (network theory), methodological (network analysis methods and software) and applied components. The students were able to apply their knowledge of theory and methodology immediately in designing research projects in their respective areas of academic interest. For instance,Tatiana Nikolaeva, Senior Lecturer of the Faculty of Economics, HSE Campus in Nizhny Novgorod, intends to apply network analysis to tourism research, while Anastasia Eremina, postgraduate student of the same faculty, expects network analysis to be instrumental in her study of public and municipal procurement.
Stanley Wasserman, Academic Supervisor of HSE International Laboratory for Applied Network Research
‘School participants are working on different projects that involve network measurements,’ says Stanley Wasserman. ‘For instance, one student is interested in recruitment analysis in terms of network influences – if you know people that can help you get jobs. Another student is studying whether there is an influence through network processes on pregnant women in Africa and their nutrition behaviour. Will the women eat properly and take vitamins in order to have healthy babies as a result of the social influence that their friends have on them or not? Network information is all relation work.’
‘This is the fourth summer school focusing on network analysis which I have attended,’ says Dmitry Zaytsev, Associate Professor of HSE Faculty of Social Sciences. ‘I have applied methods taught here to political studies. Unlike conventional statistical methods, such as regression, correlation and some others, network analysis does not reduce the complexity of social phenomena. In certain cases, using this approach is the only opportunity for political and social scientists and some other researchers to confirm or disprove a hypothesis.’
‘I have been conducting a study on media effects, and now I’m thinking of using network analysis in my study – something I have never done before,’ says Aigul Mavletova, Deputy Dean of the HSE Faculty of Social Sciences. ‘The summer school gave me an idea as to how it can be done. The purpose of my study is to examine how mass media influence students' political attitudes and behaviour not only directly, but also via social networks. Network analysis is essential here.’
For the first time this year, the International Laboratory for Applied Network Research is admitting students to an English-language Master's programme 'Applied Statistics with Social Network Analysis' modelled after the programme in applied statistics taught at Indiana University, Bloomington. Students will have the opportunity to develop comprehensive knowledge and skills in statistical data and network analysis. The programme is suitable for specialists in different fields, such as sociology, management, political science and many others, who wish to learn how to apply data analysis, in particular network analysis, to problem-solving.
‘Many Bachelor's programmes in non-mathematical disciplines do not provide statistical tools other than a general understanding of statistics, which is often insufficient for practical purposes,’ according to Kuskova. ‘Mathematicians may also find the programme relevant, but perhaps not all mathematicians: those who wish to build their machine learning skills might choose the Faculty of Computer Science with its excellent programmes on the subject. But those mathematicians who would like to learn how to define problems for finding suitable solutions, e.g. in consulting, will do well here, since most courses in mathematics do not teach how to formulate research questions and we can fill this gap.’
‘Our new Master's programme is broader than just networks,’ says Wasserman. ‘It focuses on statistics and welcomes people of any background who want to study statistics further for finding solutions to real-life problems. This programme will help them get better jobs upon graduating in a variety of areas, as the demand for quantitative specialists keeps growing.’