Question 1: A clinical trial has been performed in a radiation oncology clinic treating a certain population of cancer patients with the same fractionation scheme but performing the treatment with or without adjuvant chemotherapy. At the conclusion of the follow up, the principal investigators want to express the probability of a specific adverse reaction occurring in the populations with chemo compared to those without. Which statistic would best describe this?

Reference:  Sistrom, C.L. and Garvan, C.W., 2004. Proportions, odds, and risk 1. Radiology, 230(1), pp.1219. 
Choice A:  pValue 
Choice B:  Relative risk 
Choice C:  Kaplan Meier Statistic 
Choice D:  Absolute risk ratio 
Question 2: A clinical trial radiopharmaceutical study is being performed with two arms, arm A and arm B to test a new drug. The two arms have different dosing levels, one at the standard of care and the other at a new, higher dose level. If this is a randomised doubleblind trial, which of the following is true?

Reference:  Schulz, K.F. and Grimes, D.A., 2002. Blinding in randomised trials: hiding who got what. The Lancet, 359(9307), pp.696700. 
Choice A:  Both the administering physician and the patients have no vision. 
Choice B:  The patient does not know which arm they are a part of, but the physician does. 
Choice C:  No one involved with any element of the clinical trial knows what arm the patient is in. 
Choice D:  Neither the administering physician nor the patient know which arm they are assigned to. 
Question 3: To compare the means of the same patient group before and after treatment we use a paired ttest. It is appropriate to use a parametric test because... 
Reference:  Tello, R and Crewson, P.E., 2003, Hypothesis Testing II: Means. Radiology, (227), pp. 14 
Choice A:  Parametric tests should always be used for patient data. 
Choice B:  The variance of the two samples will be different, even though they are linked. 
Choice C:  Then we don’t need a null hypothesis. 
Choice D:  The variance of the two samples can be assumed to be equal, because they are linked. 
Question 4: Correcting for multiple comparisons is necessary when... 
Reference:  Tello, R and Crewson, P.E., 2003, Hypothesis Testing II: Means. Radiology, (227), pp. 14

Choice A:  You are comparing more than 5 groups. 
Choice B:  You are comparing more than 2 groups. 
Choice C:  You planned on comparing 2 groups, but later changed your mind based on the data and are now comparing more than 2. 
Choice D:  You find multiple ttests have been performed comparing the same data sets. 
Choice E:  All of the above. 
Question 5: You have continuous independent and dependent variables (X and Y, respectively). You want to assess if Y increases as X increases without assuming any particular relationship. You should use: 
Reference:  p. 97 of Statistics, An Introduction using R by M. J. Crawley (1 ed; 2005; John Wiley and Sons) 
Choice A:  Pearson’s “ProductMoment” Correlation. 
Choice B:  Spearman’s “Rank” Correlation. 
Choice C:  Fleiss’ Kappa Correlation. 
Choice D:  BlandAltman Analysis. 
Question 6: What is the problem with using only two observers when performing ROC analysis?

Reference:  Shirashi J, Pesce LL, Metz CE, Doi K. Experimental Design and Data Analysis in Receiver Operating Characteristic Studies: Lessons Learned from Reports in Radiology from 1997 to 2006, Radiology 253(3): 2009.

Choice A:  The results are not generalizable to the larger population. 
Choice B:  There is not enough data to construct the ROC curve. 
Choice C:  The AUC could be negative because of conflicting data. 
Choice D:  Readers might agree exactly and thus an ROC curve could not be constructed. 