*2.5. Statistics*

Data are presented as means ± standard deviations and relative (absolute) frequencies, respectively. All statistical analyses were conducted using SPSS Statistics version 26 (IBM, New York, NY, USA). The first step of the analysis consisted of tests on differences in potentially associated factors between the four health groups group with very good perceived health (VH), good perceived health (GH), moderate perceived health (MH), and poor/very poor perceived health (PH). Since the group sizes were unequal (n = 34 for the smallest and n = 482 for the largest group), Kruskal–Wallis H-tests were calculated for continuous factors (age, BMI, mean daily electronic media use, mean weekly leisure time sports activity, and mean physical fitness). For categorical factors (sex, weight status, electronic media in bedroom, sports club participation, and prevalence of smoking and drinking alcohol, as well as of suffering from back pain), Pearson's chi-square tests were used to evaluate differences between the four health groups. In the case of a significant result, additional post hoc tests with Bonferroni correction were performed.

The second analysis step consisted of a multiple multinominal logistic regression analysis with health status as the dependent variable. A multinominal logistic regression model was chosen since proportional odds between categories of health status were not assumed. The reference level was set to the group with very good perceived health (VH). All variables with significant differences between categories were included as predictor variables to the multiple model, except for weight status to avoid redundancy with BMI. Although not significant in the simple analysis, age was included in the multiple model to account for age differences in the predictor variables (e.g., percentage of smokers, percentage of alcohol consumers, BMI). Odds ratios (OR), including 95% confidence intervals (95% CI), were calculated for all predictor variables.

All *p*-values were two-tailed, and values less than 0.05 were considered to indicate statistical significance.
