Quantifying the influence of the cultivated plant species on the occurrence of carabid beetles within certain species using contingency table analysis

The influence of the cultivated plant stand, subject to the timing of sampling, on the occurrence of selected carabid species is investigated for three years running. Due to the nominal character of the factors (cultivar, timing, carabid species and year) we used log-linear models of four dimensional contingency tables under certain hypotheses of independence. A coefficient of contingency is calculated for characterizing the partial association between two factors respectively. The difference in G 2 's between two nested models is used to test if there are differences between the three years concerning the strength of partial association between the number of beetles in the selected species and the cultivar or the sampling time respectively.