Non-experimental research is particularly important in the investigation of causal connections concerning the state of health of forests, because noticeable foliage loss occurs mainly in trees over 60 years old whereas experiments are mostly limited to young plants. The extrapolation of results which are valid for seedlings, saplings, or single twigs to long-lived trees is especially risky in this case. Therefore hypotheses on causal connections must be tested by forest surveys. This paper focuses on how differences in crown condition, stability, or increment between sample plots may be related to possible causes. Statistical relationships may directly lead to causal connections. If a regression model, including all known influences, is applied, the residuals, plotted on a map, often will display unknown and yet unexplained local pecularities. Sometimes explanations may be found immediately; otherwise, clusters of plots in especially good or especially poor state of health may be extremely well-suited, paired samples for further studies.