PCL376H1 Study Guide - Confidence Interval, Statistical Population, Central Limit Theorem
28 Jan 2015
School
Department
Course
Professor
Document Summary
Pcl376 experimental design and data analysis in pharmacology and toxicology. Case study: effects of guancydine on central venous pressure. Internal vs. external validity: threats to internal validity: assignment bias, attrition, subject expectations, experimenter effects, time-related effects. Selection bias occurs when the sample is not representative of the population: selection vs. assignment, prospective vs. retrospective. Experimental error: random error (noise) cannot be eliminated, distorts results randomly. To reduce random error: increase sample size, homogeneity, repeated measure designs: constant/systematic error eliminated in a controlled experiment, can distort results in a particular direction. To reduce constant error: blocking, randomization, blinding. Experiments: natural experiment intact group designs, non-random assignment, draw correlations, controlled experiment random assignment, draw causation. 23-47: randomized control trials (rcts, cohort studies. Like cohort study, but one pre-existing group, and they are exposed to a condition: case-control studies. Dvs on pre-existing groups: cross-sectional surveys. Snapshot in time: before and after designs (repeated measures)
