EPID 301 Chapter Notes - Chapter 7: Observational Error, False Positives And False Negatives
11 Jul 2018
School
Department
Course
Professor
Document Summary
Measurement bias (a. k. a information bias): erroneous measurement leads to misclassification, the resulting type of bias. Results from systematic error in the measurement (from flawed disease-detection tests, diagnostic algorithms, classification protocols) The ruler measurement) generate errors that are not by chance (increase the sample size don"t help) Will not detect disease in all study subjects with disease. Prevalence of insensitive test: numerator be smaller / all members of the sample, regardless of their disease status or classification would be present in the denominator = **underestimation of prevalence. Mistakenly classify some disease-free subjects as diseased (false-positive) Misclassification would change the calculated frequency in the numerator. Denominator of the calculated parameters does not change. **prevalence toward values higher than the true population value. Tendency to underestimate due to low sensitivity + tendency to overestimate due to low specificity = cancel out (unbiased prevalence estimate) Most diseases study contains far fewer people with disease than without (many more false positive than false negatives)
