) but also the probabilities of false positive and negative results, taking into account typical imperfections in the data. We concentrate on a common source of imperfection in the data: non-differential misclassification of binary exposure classifier. In context of an unmatched case-control study, we

In quantitative proteomics, the false discovery rate (FDR) can be defined as the number of false positives within statistically significant changes in expression. False posi-tives accumulate during the simultaneous testing of expres-sion changes across hundreds or thousands of protein or peptide

ii This thesis addresses issues concerning Cicero’s prosecution of Verres over the course of three chapters. The first chapter examines the case and clarifies what actually happened at the trial, what form of defence Hortensius raised, and why Cicero was so successful at overcoming his opposition’s ploys. As Cicero had modified the trial’s format, and Verres had fled from Rome before it had concluded, modern scholarship has not established the details of the trial clearly. The second chapter analyses the case from the perspective of the defence, putting forward the most likely arguments Hortensius would have considered. These are formed from anecdotes in the Verrines, Cicero’s defence speeches, and the recommendations of the rhetorical handbooks. By taking each probable argument in turn, Cicero’s prosecution is shown to negate their effectiveness in advance. Although this chapter can extend no further than the hypothetical realm, it illustrates that Cicero carefully crafted his prosecution to counteract any defence. This implies that rhetorical brilliance was at work in his performance, a proposition at odds to scholarship’s tendency to attribute his success solely to his overwhelming evidentiary advantage. The final chapter utilises the scepticism in Cicero’s reliability that the first two chapters have developed, and focuses on the analyses of economic scholars. The De Frumento is a central source in studies of Roman grain production, yet has received little critical analysis from rhetorical scholars. Economic scholars therefore utilise its figures literally, which has resulted in a slight miscalculation as Cicero manipulated the statistics he presented in court. iii

mixed machine learning and knowledge engineering methods, making it difficult to draw conclusions about the performance of particular methods. In this paper we present empirical results on the performance of a Bayesian classifier and a decision tree learning algorithm on two text categorization data

1) Initial-based methods and failing to conduct disambiguation distorts network properties: • Underestimation of number of unique authors, average distance, and clustering coefficient • Overestimation of average degree. • These trends amplify over time. • Leading to false finding about network

of recognition memory performance assumes an underlying model. Second, even acceptable models can lead to different conclusions about patterns of learning and forgetting. Third, efforts to characterize and ameliorate abnormal memory should address both discrimination and bias deficits. A common method

this paper, we investigate this revi.-ed principle as applied to question answering. In particular the goals of the research described here are to: 1. characterize tractable cases in which the system as respondent (R) can anticipate the possibility of the user/questioner (Q} drawing false

We often reach conclusions partially on the basis that we do not have evidence that the conclusion is false. A newspaper story warning that the local water supply has been contaminated would prevent a person from drinking water from the tap in her home. This suggests that the absence

� Conclusion

of training set and not only that due to the test examples, as is often the case. This could lead to gross underestimation of the variance of the cross-validation estimator, and to the wrong conclusion that the new algorithm is signicantly better when it is not. We perform a theoretical investigation