MAT 183 Study Guide - Midterm Guide: Confounding, Clinical Trial, Standard Deviation
17 Oct 2017
School
Department
Course
Professor
Document Summary
Observational studies can only find association, not causation. Cannot limit study to one variable to prove causation. Multiple ultrasounds are associated with low birth weighs, but they do not cause low birth weight. Association may point to causation, but need stronger evidence to claim causation: census poll/measure the entire population, cluster. Sample dividing the population into smaller (geographical) subgroups and randomly select certain number of the groups. Sample the entire population of the subgroup: confounding confounder: third variable that can affect the, controlled. Experiment outcome and is generally related to the variable of interest. X causes y, z also causes y, so x and z are related. A treatment is deliberately imposed on the individuals in order to observe a response/outcome. All participants are similar or as similar as possible. All but one variable in the experiment is fixed. Participants are randomly selected for treatment and control (a placebo may be administered to the control to help blind).
