What Tears Couples Apart: A Machine Learning Analysis of Union Dissolution in Germany

A data-driven approach shows that individuals' characteristics are more important than their networks in predicting fertility preferences

People's networks are considered key in explaining fertility outcomes-whether people want and have children. Existing research on social influences on fertility is limited because data often come from small networks or highly selective samples, only few network variables are considered, and the strength of network effects is not properly assessed. We use data from a representative sample of Dutch women reporting on over 18 000 relationships. A data-driven approach including many network characteristics accounted for 0 to 40% of the out-of-sample variation in different outcomes related to fertility preferences. Individual characteristics were more important for all outcomes than network variables. Network composition was also important, particularly those people in the network desiring children or those choosing to be childfree. Structural network characteristics, which feature prominently in social influence theories and are based on the relations between people in the networks, hardly mattered. We discuss to what extent our results provide support for different mechanisms of social influence, and the advantages and disadvantages of our data-driven approach in comparison to traditional approaches.

Knowledge Discovery: Methods from data mining and machine learning

The interdisciplinary field of knowledge discovery and data mining emerged from a necessity of big data requiring new analytical methods beyond the traditional statistical approaches to discover new knowledge from the data mine. This emergent approach is a dialectic research process that is both deductive and inductive. The data mining approach automatically or semi-automatically considers a larger number of joint, interactive, and independent predictors to address causal heterogeneity and improve prediction. Instead of challenging the conventional model-building approach, it plays an important complementary role in improving model goodness of fit, revealing valid and significant hidden patterns in data, identifying nonlinear and non-additive effects, providing insights into data developments, methods, and theory, and enriching scientific discovery. Machine learning builds models and algorithms by learning and improving from data when the explicit model structure is unclear and algorithms with good performance are difficult to attain. The most recent development is to incorporate this new paradigm of predictive modeling with the classical approach of parameter estimation regressions to produce improved models that combine explanation and prediction.

Prosocial activity in later life: are informal help and care associated with volunteering and charity?

Older adults' engagement in various prosocial activities is a salient question in present-day societies that aim to promote active ageing. However, there are only a few studies focusing on associations between several types of prosocial activities, and they have rarely considered help to relatives and friends separately. Moreover, there is lack of studies considering informal monetary help and charity donations when analysing multiple prosocial activities. Using population-based data of older Finns (N = 2,184), we examined whether providing informal help (i.e. practical help, financial support or personal care) to relatives and friends is associated with participation in volunteering and charity, respectively. Overall, 5 per cent of the participants provided all examined forms of informal help and volunteered, 16 per cent provided two types of help and volunteered, and 23 per cent provided one type of help and volunteered. In addition, 9 per cent of the participants provided all types of informal help and made charitable donations, 33 per cent provided two types of help and made charitable donations, and 54 per cent provided one type of help and made charitable donations. Practical help and care channelled outside the household were associated with an increased probability of volunteering, although they were not associated with the probability of making charitable donations. Practical help, financial support and personal care provided to friends were particularly important predictors of volunteering and charity. These results are discussed in the context of the role overload and role extension hypotheses.