The methods can be briefly summarised in three groups:

1. Survival (life course) methods
   The (L-ANS) and the (L-ANS-ALM) databases will be suitable to apply statistical models for retrospective cohort settings, such as multilevel models, and multistate models in the presence of competing risks. In order to gain more insight into student history, we will explore the possibility of applying extended Cox Models and eventually methods with unobservable variables. Another approach will take into consideration the possible effect in the long run of early-life conditions and events. This approach has its classical application in the study of health and mortality. It seems innovative in educational studies and it is grounded on the idea that the possible determinants of the choices toward university mobility might be traced in early-life conditions of the student career.
   
   
2. Survey data
   Data collection on students will be conducted via a mixed-mode survey, considering CAPI, CATI and CAWI mode to manage the questionnaires. Recently, these kinds of surveys have been used more frequently, both by international private research institutes and the Italian Statistical Institute. This kind of approach allows one to save time and substantially reduce the cost of the survey. Of course, in the case of mixed mode data collection special attention will be paid on questionnaire construction, data comparability among the modes, error sources and sizes.
   
   
3. Social Network Analysis (SNA) of student Flows
   The micro-data (L-ANS) allow to infer the network of student mobility from high school to university and between Italian universities. SNA will be used to analyse the roles played by each province or university in attracting undergraduate/graduate students through network indexes and the global structure of the network through clustering and block-modelling techniques for one-mode and two mode network data.
   
   
4. Systemic Level Network Analysis
   We intend to investigate Italian University in a systemic perspective as a time-varying multidimensional network of institutions interconnected at various levels. The first level is given by student flows, another one is the university faculty mobility, plus, of course, geographical or transportation proximity. This picture will be enriched by including vertex attributes that measure intrinsic factors and context variables. We will measure the vertex centrality of institutions, both in absolute and relative terms, and the degree of similarity between them. As well, similarity (as another kind of proximity) will provide a different “geography” of University taking into account degree courses, governance, performance, and context. Community detection algorithms will be used to identify groups of similar, or similarly performing institutions. The network’s time evolution and standard qualitative and quantitative methods will allow to measure the effects of quasi-market dynamics in reformed University subject to performance budgeting, on the universities’ strategies, for increasing performance and attractiveness, and on the divide between institutions.