1 We fitted the following models to the data nhanes ## (Dispersion parameter for binomial family taken to be 1) ## ## Null deviance: 1195.5 on 1459 degrees of freedom ## Residual deviance: 1073.5 on 1447 degrees of freedom ## AIC: 1099 ## ## Number of Fisher Scoring iterations: 5 c Which model will have the largest i maximised loglikelihood? ii deviance? [5 marks] [5 marks] d The following code conducts a hypothesis test. State the null hypothesis being tested and what you conclude from the output anova (model3, model4, test = "Chi") ## Analysis of Deviance Table ## [10 marks] # # Model 1: hadcvd RIDAGEEX + RIAGENDR + BPXSAR + BPXDAR + DMDEDUC + LBXGH + ## LBXTC + factor (smoking) # # Model 2: hadcvd RIDAGEEX + RIAGENDR+ BPXSAR + BPXDAR + DMDEDUC + LBXGH + LBXTC + factor (smoking) + factor(pfoa4) ## Resid. Df Resid. Dev Df Deviance Pr(>Chi) ## ## 1 ## 2 1450 1447 ## --- 1085 1073 3 11.9 0.0079 ** ## Signif. codes: # # 3 0 '***' 0.001 '**' 0.01 * 0.05 0.11 e What can we conclude about goodness of fit by comparing the residual de- viance and the residual degrees of freedom for model4? Why? [10 marks]