Preface to First Edition - lib

136 LOGISTIC REGRESSION AND GENERALISED LINEAR MODELSR> print(backpain_glm)Call:clogit(I(status == "case") ~ driver + suburban + strata(ID),data = backpain)coef exp(coef) se(coef) z pdriveryes 0.658 1.93 0.294 2.24 0.025suburbanyes 0.255 1.29 0.226 1.13 0.260Likelihood ratio test=9.55 on 2 df, p=0.00846 n= 434Figure 7.11R output of the print method for the conditional logistic regressionmodel fitted to the backpain data.Conditional on residence we can say that the risk of a herniated disc occurringin a driver is about twice that of a nondriver. There is no evidence that wherea person lives affects the risk of lower back pain.7.4 SummaryGeneralised linear models provide a very powerful and flexible framework forthe application of regression models to a variety of non-normal response variables,for example, logistic regression to binary responses and Poisson regressionto count data.ExercisesEx. 7.1 Construct a perspective plot of the fitted values from a logistic regressionmodel fitted to the plasma data in which both fibrinogen and gammaglobulin are included as explanatory variables.Ex. 7.2 Collett (2003) argues that two outliers need to be removed from theplasma data. Try to identify those two unusual observations by means of ascatterplot.Ex. 7.3 The data shown in Table 7.5 arise from 31 male patients who havebeen treated for superficial bladder cancer (see Seeber, 1998), and give thenumber of recurrent tumours during a particular time after the removal ofthe primary tumour, along with the size of the original tumour (whethersmaller or larger than 3 cm). Use Poisson regression to estimate the effectof size of tumour on the number of recurrent tumours.© 2010 by Taylor and Francis Group, LLC