Fig 2. Graphical presentation of the interaction between pain symptoms and parental education.

Models were adjusted for year and parental origin. Parental education is an effect modifier (interaction is present). The effect of pain symptoms, on school grades, is different when parents has high compared with low education. This is for example seen in the 10^th percentile, where the two lines for high respectively low parental education are not parallel. N_girls = 9889, N_boys = 959.