Multivariate ANOVA models found no effect of gender on the mean test difference for any angle.
The multivariate linear regression models found several factors that were significantly associated with the AQ-9 stereotypes of "pity", "dangerousness", "blame", "anger", "help", "segregation" and "avoidance" (Table 5).
Univariate and multivariate Cox regression models found the preoperative varus angle not to be a significant factor (p = 0.8 and p = 0.84, respectively).
Final multivariate models were found by stepwise regression.
We conducted Hosmer-Lemeshow goodness-of-fit tests on our multivariate models and found that the models fitted to a satisfactory extent.
In multivariate models, we found no evidence of an association between any of the measures of functional limitations and breast cancer-specific mortality in this group (table 2).
However, in our multivariate models we found considerable odds elevation of psychosomatic symptoms, musculoskeletal pain, and depression 30-3300%) in relation to neighbourhood social capital and general social trust.
In addition, when both MEHP and oxidative metabolites were considered in multivariate models, we found an inverse association between MEHP and free T4 along with evidence of a positive association between MEHHP and free T4 (Table 5, Figure 4).
In multivariate models, we found that older maternal age and earlier birth year were both associated with elevated cord blood PCB levels, indicating both cumulative exposure and temporal trend effects.
Similarly, although gonorrhea was a strong statistical predictor of recurrent SSTI in several preliminary multivariate models, we found that models including gonorrhea were overly influenced by the single subject with a history of gonorrhea during follow up with recurrent SSTI (out of two subjects with a history of gonorrhea during follow up).
