Skip to main content

Table 3 Alternative sampling methods in Italy and Spain

From: Surveying immigrants in Southern Europe: Spanish and Italian strategies in comparative perspective

Methodology Sampling design Advantages Problems and difficulties application
Snowball K stages Useful for pilot survey Choice of initial list
Attrition rates
The probability of inclusion is unknown
Full network connectivity assumption
Interuniversity survey on “Foreigner Immigration to Italy”
Respondent driven sampling Each municipality was divided into four areas, and one eligible immigrant was selected in each of them. First respondents in each area were not allowed to know each other to avoid overlapping networks; and each respondent could provide up to three contacts for other immigrants who had never lived in the same household as the respondent. The procedure was repeated until the final number of questionnaires was achieved.   over-representation of men, especially in the age group 30-39 SPA
Romanian Communities in Spain, 2008. Four municipalities in the region of Madrid. (Sandu, 2009)
Residential units method (unità abitative) 2 stages
List of apartments with at least a foreigner (1° level unit); random selection of one foreigner (2° level units).
The initial list is updated through snowball question.
Lower non-response rate Not suitable in big cities
High attrition rate (snowball component)
Casacchia & Strozza (1990)
Natale & Strozza (1997)
Quota sampling Quotas by age and gender in 4 Italian Regions.
Selection channels: from contacts obtained in Senegal, or public spaces, migrant associations, snowballing, interviewers’ contacts
  Multiple steps with high sample attrition and (non-random) selection
selection biases at origin
ITA & SPA (partial sample)
MAFE 2008.
Beauchemin & González-Ferrer (2011)
Random routes 3 regions with largest percentages of the targeted population
Districts where more than 10% of the population belonged to the target population
To select the final respondents, random routes were combined with a spatial sampling method (see below)a
age distribution and the average length of stay of the surveyed population was quite similar to the Padrón distribution women were seriously under-represented in most origin groups SPA
Remittances of Latin-American immigrants in Spain 2012. Adult immigrants from six Latin American countries
(Izaguirre Vizcaya, et al., 2016).
Spatial sampling (and random routes) Sampling in centres of aggregation: immigration information offices, public spaces in neighbourhoods and, above all, the queue outside the Moroccan consulate refusal rates were low – especially among the individuals who were queuing for long hours outside the consulate premises often the inability to locate anyone in certain public spaces required changing the intersection points SPA
Localmultidem 2006b
(Morales & Ros, 2013).
Centre sampling technique 2 stages
List of centre of aggregation (1° level unit); casual selection of one foreign (2° level units).
Non response rate due only to refusal
Suitable also to big cities
Cost-efficient method
Sample more representative and with weighted probabilities
Target population “present population”
Use a shorter questionnaire
Regional surveys:
Lombardy Regional survey every year since 2001
National surveys:
South survey aimed at studying effects of the 2003 regularization (Blangiardo & Farina, 2006)
Integrometro survey aimed at gathering information on the integration process (Cesareo & Blangiardo, 2009).
PER.La survey, to analyze the working trajectories of the migrant population (Ismu, Censis and Iprs, 2010).
International surveys:
Italian sample in the “Push and Pull factors” (Birindelli, et al., 2000) in the Italian, Portuguese, and Hungarian samples of the “Immigrant Citizens Survey” (Huddleston & Tjaden, 2012); in the sample of Milan for the Localmultidem survey (Morales & Pilati, 2014).
  1. a1) routes were designed in a way that included concentration points for the target population (surrounding areas of supermarkets, malls, banks, bus and metro stops, health care centers, call shops, remittances agencies, bars, restaurants, churches, pharmacies, parks, etc.); 2) interviewers chose respondents from the resident population from passersby on the “random” walk. Two rules had to be respected: a) no more than 10 contacts or interviews could be completed in the same point on the same day and, b) only one member of the same family could be interviewed (in case more than one eligible respondent was identified)
  2. bDue mostly to deficiencies in the sampling frame (Padrón prior to expiry procedure) and the impossibility of obtaining a new random sample from the register to complete the study on time, 267 interviews out of the 600 to complete the Moroccan subsample in the Localmultidem Survey in Madrid and Barcelona had to be obtained through spatial sampling