Testing complex causal hypothesesScientists as well as nonscientists generate and test hypotheses about causal relations. There are two kinds of causal hypotheses, simple ones that refer to single causal relations and complex ones that refer to causal structures. Research on simple hypotheses has shown that people use statistical covariation information for their judgments in a normative fashion. Little is known, however, about how complex causal hypotheses are evaluated. According to normative theories, hypotheses about causal models require the evaluation of the strength of the individually hypothesized causal links along with tests that address the adequacy of the assumed causal structure. In 3 experiments it was investigated how participants tested complex causal hypotheses. The results showed that they tended to evaluate the individual causal links but appeared not to have any explicit knowledge about how hypotheses on the structure of causal models should be tested.https://www.psych.uni-goettingen.de/de/cognition/publications/hagmayerwaldmann2001https://www.psych.uni-goettingen.de/@@site-logo/university-of-goettingen-logo.svg
York Hagmayer and Michael Waldmann (2001)
Testing complex causal hypotheses
In: False, ed. . Bern Studies in the History and Philosophy of Science
Scientists as well as nonscientists generate and test hypotheses about causal relations. There are two kinds of causal hypotheses, simple ones that refer to single causal relations and complex ones that refer to causal structures. Research on simple hypotheses has shown that people use statistical covariation information for their judgments in a normative fashion. Little is known, however, about how complex causal hypotheses are evaluated. According to normative theories, hypotheses about causal models require the evaluation of the strength of the individually hypothesized causal links along with tests that address the adequacy of the assumed causal structure. In 3 experiments it was investigated how participants tested complex causal hypotheses. The results showed that they tended to evaluate the individual causal links but appeared not to have any explicit knowledge about how hypotheses on the structure of causal models should be tested.
