Regression analysis showed that the stress manipulation influenced the level of anger and that change in anger predicted post-pain negative affectivity independently of the contribution of maximum pain (model R2 =.31), with 45% of the total model variance accounted for by change in anger and 17% of the total model variance accounted for by maximum pain intensity.
In the nonstress condition only level of pain intensity was an independent predictor of negative affectivity (model R2 =.16), with 69% of the total model variance accounted for by maximum pain intensity.
Test of F in the ANOVA analysis is used for evaluating model variance in residual variance.
Collinearity diagnostics did not suggest a problem with multicollinearity in this model (variance inflation factor <1.5 for all included variables).
Under an assumption of independence of model errors across species, the model error variance for a group or all species combined are computed as the sum of the variances of individual species.
In particular, we modeled the variance for the above three mean equations using GARCH, APARCH, EGARCH, TGARCH, and IGARCH models.
Between-model variances increase for the future time series compared to the historical ones indicating larger uncertainties in climate change projections.
Tests of variance inflation indicated no serious multicollinearity issues in the models (variance inflation factor <5 for all models).
For both models, variance components were estimated with restricted maximum likelihood methodology using the ASREML software package [ 7].
IT-based model averaging allows the estimation of confidence intervals that incorporate model estimates of the sampling variance for each model and of the model selection uncertainty.
To test this hypothesis, we fitted a model that allowed different variances for the plate makes, and compared it to a model with equal variance for all plate makes, using a likelihood ratio test.
