Checking assumptions in regression _Ch 12 13_

# Checking assumptions in regression _Ch 12 13_ - series data...

This preview shows pages 1–2. Sign up to view the full content.

Checking assumptions in regression situations 2. Checking for constant variability Why? To make sure that when you get confidence intervals to predict Y that they are accurate. How? Check a plot of residuals against the predicted values and see if the pattern looks random. If you see the variability get larger as the predicted values get larger (a wedge shape), you have evidence that the constant variability assumption is not true. 3. Checking for normality Why? To make sure that the confidence intervals and p-values that you obtain are accurate. How? Check a normal probability plot of residuals and see if the pattern looks like a straight line. If you see a curved line, you have evidence that the normality assumption is not true. You can also check by getting skewness and kurtosis statistics and seeing if they both fall within + 1. The skewness statistic is especially important. ASSUMPTION? Why do we need it? How do we check it? 4. Checking for independence (only for time

This preview has intentionally blurred sections. Sign up to view the full version.

View Full Document
This is the end of the preview. Sign up to access the rest of the document.

Unformatted text preview: series data) Why? To make sure that when you predict that future errors do not depend upon past errors. How? Check a plot of residuals against the time sequence and see if the pattern looks random. If you see a snake like pattern, you have evidence that the independence assumption is not true. You may also check the D-W statistic to see if it meets the guideline that it exceeds 1.3. Any value below 1.3 indicates that the independence assumption may not be true. 1. Checking for linearity Why? To make sure that a straight line is the type of equation that fits best. How? Check a plot of residuals against an independent variable and see if the pattern looks random. If you see a rainbow or smile shape, you have evidence that the linearity assumption is not true. To confirm, click on a point and add a polynomial trendline and check the R-Sq box. If R-sq < 0.20, the assumption is OK....
View Full Document

## This note was uploaded on 04/08/2011 for the course BUS 310 taught by Professor Abou-sayf,f during the Spring '08 term at University of Hawaii, Manoa.

### Page1 / 2

Checking assumptions in regression _Ch 12 13_ - series data...

This preview shows document pages 1 - 2. Sign up to view the full document.

View Full Document
Ask a homework question - tutors are online