Survey on attitudes toward immigration and immigrants in Spain (EASIE survey)

[Description of methods used for data collection] The EASIE survey was directed to a vast sample of Spanish nationals residing in Spain (N=2.344); data were collected by a combination of self-administered online questionnaires (CAWI; N=1.965) and telephone interviews (CATI; N=379); the latter were employed for sociodemographic profiles with low Internet use. Fieldwork was conducted by IMOP Insights, a specialized private-sector entity. The survey was approved by the Spanish Research Council’ Ethics Committee (reference nº 127/2020). The CAWI sample was random-selected among members of E-MOP who met required profile characteristics (age group, sex, etc.). E-MOP is a probability-based internet panel run by IMOP Insights. With a view to obtaining the best possible match between sample and the population of study, the CATI sample employed a dual (landline & mobile) sampling frame that sought to compensate internet penetration rates in each age group. Landline participants were selected in two steps: random selection of households based on landline listings, and subsequent quota-based selection of individuals (age & sex), always giving preference to the household member whose profile was worst represented in the sample at the time of call and postponing the interview if that person was temporarily unavailable. Mobile lines were randomly generated from the ranges assigned by the Spanish phone regulator to each mobile operator, automatically discarding inactive lines; the database was checked prior to fieldwork. For the CAWI sample, an average of 3 reminders were sent (maximum 5 contacts, i.e., initial invitation plus 4 reminders). For CATI, selected numbers not contacted at the first attempt were called two more times on average. Average questionnaire duration was 13.6 minutes for CAWI and 19.7 minutes for CATI. For illustrative purposes, the nominal margin of sampling error under the assumption of simple probability sampling for the whole sample (not applicable in this case) would be ±2.1 for p=q=50% and a significance level of 95%.