This preview shows pages 1–7. Sign up to view the full content.
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: PUBH 7430 Midterm review (Lecture 17) J. Wolfson Division of Biostatistics University of Minnesota School of Public Health November 1, 2011 Preliminaries: bivariate dependence The correlation coefficient Know the formula: ( X , Y ) = Cov ( X , Y ) p Var ( X ) p Var ( Y ) Intuition: Covariances depend on scale of random variables, correlation gives scaled version of covariance Key properties: [L5, p.6] Indicates strength of linear relationship May be affected by outliers Alternatives: Spearmans , Kendalls Preliminaries: multivariate dependence Notation Outcome vector Y i and Y Mean vectors i and Variancecovariance matrices i and Predictor/covariate vector x ij and matrices X i and X The linear predictor X Exploratory data analysis Wide format and long format for longitudinal data [L3, p. 5] Clusterinvariant vs. clustervarying covariates [L6, p. 14] Paired data: Twosample, onesample, and paired ttest [L3, p. 16] Exploratory data analysis Graphical summaries Spaghetti plots and ways to make them more readable [L4, p. 17] Withinperson residuals : Compare longitudinal patterns (trajectories) by eliminating differences in individual variability [L4, p. 20] Withintime residuals : Make crosssectional comparisons by eliminating time trends [L4, p. 22] Exploratory data analysis Smoothers Basic goal: Estimate the mean response curve nonparametrically Kernel smoothers: [L4, p. 27] Kernel determines how to average observations Bandwidth determines how much influence nearby vs....
View Full
Document
 Fall '04
 Prof.Eberly

Click to edit the document details