Is intent to migrate irregularly responsive to recent German policy adjustments?

We conduct a single choice experiment in which participants are presented with 3 migration policy profiles, containing 5 attributes---asylum chances, waiting time until asylum decision, application location, payment mode of government transfers during the application process, and waiting time until receipt of government benefits equivalent to those received by locals---each with two randomly varying values. The 5 attributes and 2 values result in 32 hypothetical combinations of policy profiles.

The participant is presented with a profile of randomly selected policy values and then asked to state their intent to migrate given the policy profile. In total, the participant is presented with three profiles. For a better understanding of the policy profiles they are visualized through icons shown on enumerators' tablets and on printouts. After the choices were made participants receive a short debrief. The procedure is:

(1) Introduction: "We will now show you three sets of policies that this European country could have in place. Given each set of policies, please indicate, how interested you would be in trying to migrate irregularly (traveling without prior approval) to this country? [0 = Not at all. 10 = Very.]"
(2) Profile 1, then: "Given this set of policies, how interested would you be in trying to migrate irregularly to this country? [0 = Not at all. 10 = Very.]''
(3) Profile 2, then: "Given this set of policies, how interested would you be in trying to migrate irregularly to this country? [0 = Not at all. 10 = Very.]"
(4) Profile 3, then: "Given this set of policies, how interested would you be in trying to migrate irregularly to this country? [0 = Not at all. 10 = Very.]"
(5) Debrief: "The scenarios presented are hypothetical and are based on current political measures and discussions in Europe. We recommend that you carefully inform yourself about the actual numbers and processes if you consider migrating."

2023-11-26

2023-12-15

Intent to migrate irregularly

Our outcome is the response to the question "Given this set of policies, how interested would you be in trying to migrate irregularly to this country?" measured on a scale from 0 (not at all) to 10 (very). The observed response values will inform us about subjects' intent to migrate in absolute terms. In regression analyses we will standardize the outcome measure to a mean of 0 and a standard deviation of 1 to ease interpretation and comparability.

None

NA

The study is conducted with a total sample size of 1000 individuals in four cities of Senegal: Dakar, Kaolack, St.Louis and Ziguinchor. In Dakar, St. Louis and Ziguinchor male participants aged 18 to 40 of a previous data collection in these locations conducted in 2022 are followed up, with 334, 222 and 222 individuals respectively. These individuals were randomly sampled, within a random selection of quartiers/neighborhoods as primary sampling units. In Kaoloack a new sample of 222 males aged 18 to 40 will be created using a random walk selection procedure. Again, quartiers/neighborhoods serve as primary sampling units, within which a randomly selected set of individuals will be approached and asked for their participation. We focus on men aged 18--40 because this constitutes the demographic group in Senegal most likely to migrate, and we focus on these four cities because they are national and regional centers of migration.

See "Intervention" for details on the choice experiment. See "Randomization" for the choice experiments random assignment to attribute values.

Not available

We randomly assign three treatment profiles to each individual, with each profile consisting of five attributes that take on one of two possible values. Assignment is stratified by city. We use a re-randomization algorithm to ensure balance in subjects' age across all treatment profiles. Random assignment of the fully articulated set of treatment profiles ensures random assignment of values for a particular attribute that is exactly balanced across all other attributes, i.e., the randomly assigned values are uncorrelated across attributes.

Each treatment profile is assigned at most once to each individual, i.e., subjects do not encounter the exact same profile multiple times. Overall and across subjects, each of the 32 treatment profiles is assigned the same (or nearly the same) number of times. This also means that each attribute value appears with the same frequency.

Individual

Yes

1000 individuals.

Each of the 1000 participants is presented with three profiles, which means our sample for analysis consists of 3000 observations.

500 observations.

Each of the 1000 participants is presented with three profiles, which means our sample for analysis consists of 3000 observations, with participants as clusters and a cluster size of 3. The minimum detectable effect size given these cluster sizes, number of clusters, power of 80%, a significance level of 5%, and an intracluster correlation coefficient of 0.1 is 0.11 standard deviations. Assuming an intracluster correlation coefficient of 0.5 implies a minimum detectable effect of 0.14 standard deviations.

PAP