This preview has intentionally blurred sections. Sign up to view the full version.
View Full DocumentThis preview has intentionally blurred sections. Sign up to view the full version.
View Full DocumentThis preview has intentionally blurred sections. Sign up to view the full version.
View Full Document
Unformatted text preview: Public Health 6450 Fall 2011 Lynn Eberly and Andy Mugglin Division of Biostatistics School of Public Health University of Minnesota [email protected] Part 18 Review Summarizing Relationships Between Quantitative Variables Where are we going? Previously: I one, two, and paired sample comparisons for binary data I one, two, and paired sample comparisons for quantitative data Current topics: I Overview of Analysis Methods (so far, and a few yet to come) I Describing Relationships Between Quantitative Variables (Ch. 2): Scatterplots, Correlation Next topic: I Simple Linear Regression (Ch. 2) Eberly and Mugglin PubH 6450 Fall 2011 Part 18 2 / 36 Review Summarizing Relationships Between Quantitative Variables Identify the variables of interest In scientific research, we are often interested in the relationship between two variables. I Comparing the effectiveness of two blood pressure lowering drugs: drug type and blood pressure. I Evaluate the effect smoking on lung cancer: smoking status and the presence (or not) of lung cancer. I Education level ( ≤ high school vs. ≥ college) and income. Most often, one of these variables is an outcome of interest (what we have often called a ‘disease’ variable) and the other is a variable we think may explain that outcome (an ‘exposure’ variable). Eberly and Mugglin PubH 6450 Fall 2011 Part 18 3 / 36 Review Summarizing Relationships Between Quantitative Variables Twosample ttest When the ‘exposure’ variable is binary (drug 1 vs drug 2; ≤ high school vs. ≥ college education) and the ‘disease’ variable is quantitative (blood pressure; yearly income), we can: I Randomize participants to receive drug 1 or 2 and compare their mean blood pressure after some period of followup. I Identify the two exposure populations ( ≤ HS vs. ≥ college), draw a random sample from each, and compare their mean incomes. Analysis method (when sample size is large enough) I Estimate: difference of the sample means, with CI. I Hypothesis test: twosample ttest. Eberly and Mugglin PubH 6450 Fall 2011 Part 18 4 / 36 Review Summarizing Relationships Between Quantitative Variables TwobyTwo (or Larger) Tables: Cohort Study or Clinical Trial When both variables are binary (smoking yes/no and lung cancer yes/no), we can: I Identify the exposure populations (smokers/nonsmokers), draw a random sample from each, and compare the proportions of lung cancer patients. Analysis method (when sample size is large enough) I Estimate: conditional probabilities (given exposure status) in the sample, risk difference, RR, or OR, with CI. I Hypothesis test: ztest of difference in proportions, chisquare test for independence, or chisquare test of homogeneity of proportions....
View
Full
Document
This note was uploaded on 11/21/2011 for the course PUBH 6450 taught by Professor Andymugglin during the Fall '10 term at Minnesota.
 Fall '10
 AndyMugglin

Click to edit the document details