Part 2 of Report 4 covers three topics. The first is attempts to systematize, weighing and forming a hierarchy ("ranks") for the causal criteria. Two methodologies were used: referring to specific constructions of various researchers and assessment by determining the frequency of occurrence of a particular criterion in those works in which they were used as a way to establish the effect causality. Apart from the a priori leading criterion "Temporality," the first two places should be given to "Strength of association" and "Consistency of association" (their sequence may be reversed), that is, inductive points. The third place is taken by the "Counterfactual experiment" ("Natural experiment"), which, however, is rarely available in epidemiology and ecology. In its absence, the third place is again taken by the inductive criterion "Biological gradient" ("dose–effect" relationship). In the fields of ecology and ecotoxicology, where experimental or observational studies in humans are rarely available, the ranks of the criteria are specific, and "Biological plausibility" is in first place. Second, it presents a criticism of the causal approach in observational disciplines based on criteria of causality, starting from the lack of evidence of the inductive method itself. It points to the "mosaicism," multidisciplinarity, and subjectivity of the criteria, as well as their nonabsoluteness ("are saddled with reservations and exceptions"). It is noted, however, that such questions can be posed before almost any approach in the field of natural sciences. There is the opinion that the criteria of causality, as it were "confirming effects," should be left only for decision-making in the field of health care, and not for strict evidence in the field of "pure science" with its falsification of hypotheses. It is unlikely that this approach can be adopted for epidemiology. Third, models for assessing the causality of effects in epidemiology outside of causal criteria are considered. The most famous is the K.J. Rothman ("Causality Pie"), implying determinism of epidemiological effects, as opposed to the position of practical epidemiology and public health, in which causality is defined as probabilistic. The second model is called the "counterfactual" or "potential outcome model." The third is directed acyclic graphs or causal diagrams (DAG), and the fourth is the "Structural equation model." Another approach was discovered by C.V. Phillips, which quantifies errors (uncertainties) at all stages of an epidemiological study with Monte Carlo simulation. It is concluded that the named models turn out to be either qualitative and illustrative, or represent only a graphic or mathematical apparatus that still serves the same initial approach based on causal criteria. [ABSTRACT FROM AUTHOR]