piatok 29. októbra 2010

advanced graphics with R

advanced graphics with R slides

basic R, statistics

basic R slides - some useful tips, from UCLA

learning statistics R via video

R via videos , learning statistics via video

research presentation, software, "impressive", video/slides presentation

http://impressive.sourceforge.net/index.php

http://4dpiecharts.com/2010/10/29/adapting-graphs-for-presentations/


Ten (10) Easy Steps to Create an Online Program!
vcasmo.com

random numbers, R, sampling, bootstrapping

RANDOM NUMBERS - concise tutorial in R

rnorm
runif
sample


useful handout on random numbers, simulating dice roll, coin toss.
loops

cf wiki book on Statistical analysis with R

http://www.astrostatistics.psu.edu/datasets/R/Random.html

štvrtok 28. októbra 2010

multicollinearity, mediation, suppression

 multicollinearity - excellent tutorial on causes, consequences of multicollinearity in multivariate linear regression - Uni Pensylvania

“LINEAR REGRESSION collinearity suppression xxxXX,” http://www.psych.umn.edu/faculty/waller/classes/mult10/readings/wall05.pdf. by Friedman, Wall,  Graphical Views of Suppression and Multicollinearity in
  1. Multiple Linear Regression
confounding, collinearity  - examples in R XXX mcgill uni.

multicollinearity - concise introduction

wiki - multicollinearity

CAUTION WITH RULES OF THUMB - factors affecting collinearity (variance of regression coefs)
- O Brien

short one-page on multicollinearity measures: tolerance, variance inflation factor (1/tolerance) and its management

orthogonalization , factor analysis for collinearity - ppt - example from functional MRI

mediation, cf MLR, sobel test, structural equation modelling, pathway analysis....


structural collinearity  Reducing structural multicollinearity  XXX

A caution regarding rules of thumb for variance inflation factors XXX




treating multicollinearity

... one way is combining of correlated variables into one, if it makes sense  Leech eta al 2005 SPSS for intermediate statistics, p 96

cf centering

suppression  c Am. Statistician
effect of variable that is not correlated with outcome, but correlated with other predictors - it increases variance of estimates of other predictors. (Graphical views of suppression and multicollinearity in multiple linear regression.)