Sampling of alternatives in migration aspiration models

The use of discrete choice models (DCMs) is a regular approach to investigating migration aspirations concerning destination choices. However, given the complex substitution patterns between destinations, more advanced model specifications than the multinomial logit (MNL) and nested logit (NL) models which are commonly found in the literature are required. The cross-nested logit (CNL) model allows for a more sophisticated representation of the stochastic structure of destination choices, through the use of overlapping nests while it is also addressing deviations from the property of independence of irrelevant alternatives. However, the shift towards CNL does not come without a cost; these models can be computationally expensive to estimate, especially as the number of observations increases. The estimation speed can be mitigated though via sampling of alternatives i.e. reducing the number of alternatives in the model specification. This method has been previously used mostly in the context of residential choice location. In the current work, we implement sampling of alternatives on migration aspiration choices using the Gallup World Poll data. We examine the impact of stratification and number of alternatives on the CNL model estimates. Moreover, we consider additional MNL and NL specifications to further understand the implications of sampling on DCMs used for modelling migration aspirations.