Ecological Systems and Devices Annotation << Back SINGLE MEASURE OF EFFECT IN LINEAR REGRESSION MODELS IN MEDICAL AND ECOLOGICAL RESEARCH A.N. Varaksin, V.G. Panov, T.A. Maslakova, Yu.V. Shalaumova Versions of definition of single measure of effect (one-factor effects) for predictors of regression models are considered. Two types of one-factor effects are distinguished: unconditional and conditional ones. Unconditional one-factor effects show a measure of the impact of a single (each) predictor on the dependent variable of the regression model, regardless of the values of all other predictors. The analysis of various approaches for defi nition of unconditional effects shows that they are expressed by one relationship in both simple and multiple linear regression models. Conditional one-factor effects show a measure of the impact of each predictor on the dependent variable, which (the measure) depends on the values of other predictors. Conditional one-factor effects can be calculated by using multiple linear regression models with cross terms (which is a product of predictors, so we have models that are nonlinear in predictors). Regression models with cross terms are often more adequate to experimental data than models that are linear in predictors, so conditional one-factor effects are more informative than unconditional effects. Nevertheless, unconditional one-factor effects also have the right to exist, since they expresses certain aspects of the relationships between predictors and response in regression models. Keywords: Medical and environmental studies; regression models; cross terms; conditional and unconditional single measure of effects. DOI: 10.25791/esip.1.2022.1276 Pp. 03-09.