Variables and Measurement Scales, Normal Distribution, and Confidence Intervals
The dilemma between sensitivity and specificity noted by the choice of threshold arises from the intrinsic variability of biologic phenomena, both at the cellular level and the organ level in the human body in the presence and absence of pathologic processes. When one measures the same hallmark in a sample of individuals there always appears a spectrum of values which is a more-or-less wide numerical set characterizing that sample for the measured hallmark. It is not by chance that in Figure 1.6 we used a bell-shaped curve to represent the set of possible values of the measured variable. Such curves indicate that the variable may take all the values within them, and that the most frequently observed values correspond to the central part of the curves.
KeywordsAbdominal Aorta Central Limit Theorem Measurement Scale Standard Normal Distribution Central Tendency
Unable to display preview. Download preview PDF.
- American College of Radiology (2003) ACR Breast imaging reporting and data system: Breast imaging atlas, 4th edn. Reston, VAGoogle Scholar
- Altman DG (1991) Practical statistics for medical research. London: Chapman & Hall Gardner MJ, Altman DG (1990) Gli intervalli di confidenza. Rome: Il pensiero scientifico editoreGoogle Scholar
- North American Symptomatic Carotid Endarterectomy Trial Collaborators (1991) Beneficial effect of carotid endarterectomy in symptomatic patients with high-grade carotid stenosis. N Eng J Med 325:445–453Google Scholar
- Siegel S, Castellan NJ Jr (1992) Statistica non parametrica, 2nd edn. Milan, McGraw-HillGoogle Scholar
- Soliani L (2007) Statistica applicata alla ricerca e alle professioni scientifiche. Manuale di statistica univariata e bivariata. Parma: Uninova-Gruppo Pegaso; 2:61–97 (http://www.dsa.unipr.it/soliani/soliani.html).Google Scholar
- UICC (2002) TNM classification of malignant tumours, 6th edn. Geneva: UICCGoogle Scholar