Results of the duiker carcass backward stepwise multiple regression model.

<p>For each model, independent covariates with the highest inclusion probability for each of the Bayesian predictive models (16 total–four interventions per market group) were included in the analysis. Dummy variables were created for each month to account for seasonal behaviors in response variables. Five separate multiple regression models were created for each of the previously determined five market periods. Independent covariates were removed stepwise using AIC values to determine the model of best fit. Coefficient estimates, standard errors, and p-values for models from each of the five periods are shown here.</p> <p>(PDF)</p>