Page 257 - Like me, or else... - Michelle Achterberg

Summary and general discussion
255
Methodological Considerations
The studies discussed in this thesis make an important contribution to the literature on the development of social emotion regulation and point to childhood as a possible window of opportunity. Apart from these theoretical implications, there are four methodological considerations that arise from these studies, which are reviewed below.
Two of a kind: Generalizability of twins to singletons
The classical twin design is sometimes referred to as “the perfect natural experiment”, as it provides the unique opportunity to tease apart genetic components from environmental influences. Using a twin design can provide important insights in the underlying mechanisms of a psychological construct. An important assumption of these studies is that findings can be generalized to the general (non-twin) population (Moilanen et al., 1999). Although several studies have shown that this is true when it comes to general physical characteristics (i.e., blood pressure or height, (Andrew et al., 2001)), twin- singleton comparisons on psychological constructs are limited. A large longitudinal study in middle and late childhood showed no significant differences between the developmental trajectories of externalizing problems of twins and singletons (Robbers et al., 2010), suggesting that twin findings on behavioral control or emotion regulation might be generalizable. However, when investigating social emotion regulation, it is important to keep in mind the unique social buffer that twin-hood might provide (Branje et al., 2004). It has been hypothesized that twins may have a favorable social environment due to interactions with, and social support of the co-twin (Pulkkinen et al., 2003). In order to test whether the findings of this thesis are generalizable to non-twin children, it is important to compare the results on aggression regulation following social evaluation with a sample of non-twins. Recently, several other research facilities have started to use the Social Network Aggression Task, and combining these samples will enable such direct comparisons.
Multiple samples vs. Massive samples
A twin study provides the additional possibility to test a specific psychological construct in two similar samples (one co-twin in each), thereby replicating findings within a study. Replication designs are very useful for testing the robustness and reproducibility of results (Schmidt, 2009; Open Science, 2015). Examples of multiple samples within one study are provided in this thesis in chapter 2 (pilot- test- replication design), chapter 5 (ROI selection in independent sample) and chapter 7 (functional connectivity in two independent samples). The findings of thesis also showed that high statistical power is needed to detect
9