Surveying migrant populations – methodological considerations: An example from Germany

Data available from the recruitment process for a population-based survey on psychological well-being in women of Turkish origin aged 18–75 years was subjected to post-hoc analysis to examine effects of different recruitment approaches and reasons for non-response. A random sample of 80,000 women was drawn from the Berlin population registry and 3884 women of Turkish origin identified by a name-based algorithm. The women were assigned randomly to three waves for further contact. The protocol consisted of an invitation letter, telephone contact and door-to-door visits. Door-to-door visits were only performed in one wave, to assess the effect of an intensified recruitment approach. Response rates were 11.9% (95% CI: 7.2–16.6) in the first wave, 3.9% (95% CI: 1.7–6.2) in the second wave and 1.3% (95% CI: 0.9–1.8) in the third wave. The contact rate was highest in the first wave (n=126; 68.1%), which included door-to-door visits. Participants were more often married, had more children, higher educational level, a higher net household income and more often a history of mental disorder. Participation was associated with door-to-door visits (OR=4.22; 95% CI: 1.89–9.45), age (for 35–54 years; OR=2.01; 95% CI: 1.03–3.91) and reply by telephone (OR=4.32; 95% CI: 2.18–8.60). Through the intensified recruitment approach the contact rate could be increased. Despite this we found a relatively low response rate in all recruitment waves.