To test the hypotheses that resource base, competition and life history are associated with trophic level and C source, we created multiple regression models for all variables with δN discrimination factor (δN A. araneoides - δNbaseline) and δC discrimination factor (δC A. araneoides - δCbaseline) as dependent variables.
bAdjusted odds based on separate multiple logistic regression models for socio-demographic variables and for behavioural variables.
Finally in order to control the confounding effect of co-variables all comparisons between males and females have been conducted using multivariate models (logistic regression for binary variables and multiple lineal regression models for continuous variables).
Multiple logistic regression models for all variables and smokers demonstrated observed differences between BOP, the number of cigarettes per day, and duration of smoking, while serum cotinine, haptoglobin and alpha-1 antitrypsin levels showed no significant differences.
Box-Cox procedures were used to resolve problems encountered with the hypotheses of normality and/or homoscedasticity in the multiple regression models for the dependent variables SOS and SI, which were subsequently log-transformed.
Candidate models for multiple-variable regression were those in which individual factors demonstrated an association that was statistically significant or approaching statistical significance, and the individual factors were not highly collinear.
Variables that remained statistically significant after separate initial multiple logistic regression models for behavioral and sociodemographic variables with outcome of HSV-2 and syphilis are combined into final models for men (Table 2) and women (Table 3) separately.
Results reveal that the SEM for the estimation of health and socioeconomic status can easily be transformed into a bivariate multiple regression model for these variables, which, in the SEM framework, is also called the reduced form of the SEM.
The multiple linear regression models for each polymorphism and quantitative variable are presented in Table 4.
Multiple effects and interactions of predictors were analysed using a linear regression model for continuous dependent variables, and a logistic regression model for dichotomous outcome variables.
Multinomial logistic regression simultaneously estimates multiple binary logistic regression models for outcomes of a nominal variable using one base category.
