This investigation focuses on the question whether assessments of the development of internalizing behavior from childhood to adolescence are affected by the kind of research design (longitudinal versus cross-sectional). Two longitudinal samples of 432 second-graders and 366 fourth graders participated in a longitudinal study with subsequent measurements taken 1, 2, and 3 years later. A third sample consisting of 849 children covering the same range of grades participated in a cross-sectional study. The results show that the development of internalizing symptoms in girls – but not in boys – varies systematically with the research design. In girls, there is a decrease of internalizing symptoms (especially between the first two timepoints) in the longitudinal assessment, which may reflect, for example, the influence of strain during the first testing situation. Both longitudinal trajectories converge to a common trajectory from grade 2 to grade 7 when controlling for this “novelty-distress effect.” Moreover, when we control this effect, the slight but significant decrease characterizing the common trajectory becomes similar to the one obtained in the cross-sectional study. Therefore, trajectories based on longitudinal assessments may suggest more changes with regard to internalizing symptoms over time than actually take place, while trajectories based on cross-sectional data may be characterized by an increased level of internalizing symptoms. Theoretical and practical implications of these results are discussed.
