The social embedding of immigrants is of crucial importance for their structural integration. This issue is debated heavily in integration research: While some scholars reason that ethnic networks constrain the advancement of young immigrants, others argue that ties to co-ethnics can compensate for structural disadvantages. To make progress in this debate it is necessary to specify the mechanism more precisely. Therefore, it seems particularly promising to rely on social capital (SC) theory. However, empirical evidence is rather weak and far from conclusive – at least for the case of Germany.
In part, inconclusive findings follow from very different SC measures. Often proxy measures are applied to catch at least some kind of SC. However, such “proxies” are not able to test mechanisms comparatively. More direct empirical tests are needed. This is a challenging task because precise SC measures are hard to reach in large-scale studies.
A whole set of severe methodological challenges and measurement pitfalls centre on the operationalisation of SC. Particularly, causal interpretations of SC effects are questioned fundamentally because of ‘causal reversals’. In this regard, esp. the non-random formation of social networks (Mouw 2006) and difficulties to account for endogeneity when estimating SC effects (Kalter 2010) have been considered. To account for these challenges, complex panel data are required that include repeated measurements.
The general design of the National Educational Panel Study in Germany (NEPS) along with its specific SC instruments will help to avoid major pitfalls. A comprehensive SC instrument has been developed (for further details see, Kristen et al. 2011; Stocké et al. 2011). It is composed of five modules: a) a position generator (Lin et al. 2001), b) a resource generator logs what kind of support is available (Van der Gaag and Snijders 2004, 2005), c) a shortened Burt generator captures characteristics of strong ties (Burt 1982), d) a measure of normative expectations accounts for reference group effects, e) a measurement of global network characteristics provides information on network composition. These modules are supplemented with measures of ethnic network composition. Altogether, this offers a unique opportunity to differentiate SC mechanisms.
In this paper, major methodological challenges will be discussed and the NEPS SC instrument and measurement design will be introduced. Against this background, data of the NEPS stage 8 “Adult Education and Lifelong Learning” that have been conducted in 2010 are analyzed. For main immigrant groups unemployment durations are investigated as an indicator of structural integration. Since unemployment can occur repeatedly, competing risk Cox regression modelling is modified to meet the peculiarities of multiple spells. Results show that ethnic networks do not hinder immigrants’ job search per se, but are generally rather harmful because they give access to fewer social capital. By means of this example, I will show how NEPS data can help to differentiate SC mechanisms, esp. in integration research.