Image of


Immigrants’ social network and the transmission of health-related behaviors and outcomes

Principal Investigators: Alessia Melegaro, Carlo Devillanova
Researchers and Collaborators: Emanuele Del Fava, Dondena Centre, Bocconi University; Marco Bonetti, Dondena Centre, Bocconi University; Ted Mouw, University of North Carolina at Chapel Hill, USA; Giovanna Merli, Duke University, USA.
Funding Body: Fondazione Cariplo
Period: 2017-2018


The project NetHealth aims to sample from the population of the recent documented and undocumented immigrants in Milan, in order to (i) reconstruct the network structure of their social contacts and (ii) assess the role of such network in shaping immigrants’ health behaviours and outcomes.

The above objectives will be addressed by running a survey on the population of both documented and undocumented Chinese and African immigrants in the metropolitan city of Milan. Considering that immigration is often a network-driven process and that immigrants (mostly those who are undocumented) can be considered a hidden population, we need a sampling strategy that overcomes the issue of the lack of a sampling frame by exploiting the social network of the individuals to obtain the desired sample. At this purpose, we opt for an innovative sampling strategy called Network Sampling with Memory (NSM), which has shown to address some of the drawbacks of the Respondent-Driven Sampling (RDS) technique, by now a standard strategy to sample from the social network. RDS requires current respondents to recruit the next wave of respondents by giving coupons to 1-5 “friends”. However, recent literature shows that this sampling technique often provides estimates with large bias and low precision, mainly because the researcher is not able to control the recruitment process (which is totally in the hands of the recruiters) and the sampling can often get stuck in highly-structured networks. Conversely, NSM requires the current wave of respondents to nominate k “friends” and provide their contact information, basic demographics, and partially-identifying information to reconstruct ties between people. All nominated individuals are added to a list that, interview after interview, forms the sampling frame from which the next respondents are sampled. The selection of the next candidate respondent is performed by an algorithm that (i) tries first to explore the network by sampling individuals who can likely bridge to unexplored areas of the network (e.g., those who have been nominated few times), and (ii), once the network has extensively been explored, tries to sample individuals from the list of the nominated nodes with an adjusted sampling probability which balances between nodes entered at different waves in the list.

In addition to the sample of recruited respondents, NSM allows to collect a sample of their social network, with associated basic socio-demographic information. Given this unique feature of NSM, we aim to understand how respondents “activate” their social contacts depending on the information they are seeking Through our questionnaire, we will collect information on (i) immigrants’ access to health care in Milan (and presence of barriers to this access), and on (ii) their access to the labour market. We will also collect the information on the “friends” who mediated this access. Finally, we will gather data on conversational social contacts on a given day, both with members of the social network and with casually encountered people. This information may help us to understand how airborne infectious diseases might spread in the community of the immigrants in Milan, in case an infection outbreak occurred, and how to control it.