1. Explain the fundamental difference between correlation and regression.
 

2. Regression analysis distinguished between three types of variation. Define each type of variation and draw a diagram to illustrate your definitions.
 

3. The Los Angeles Catalyst Study (LASC) assessed the effectiveness of the catalytic converter . As part of this research, the Environmental Protection Agency established carbon monoxide (CO) monitoring stations along the San Diego Freeway. Hourly CO averages were recorded on summer weekdays at stations located 25 feet from the freeway. Researchers believe that there is a linear relationship between observations of CO concentration (parts per million) and traffic density (traffic counts). The data reside in the file Traffic.sav.

1. Conduct a regression analysis between traffic density and CO concentration (Note: follow steps 1 to 6 outlined in the lab manual and conduct the analysis in SPSS).
    
   
2. Use the line of best fit equation to estimate CO concentration for traffic densities of 95, 200, and 275 cars and calculate confidence intervals for each prediction. Summarize these results in a table.
    
   
3. Compare the predictions to observed values. Would you recommend using this model for future predictions? Explain why or why not.
    
   
4. Explain why the widths of confidence intervals for the predicted values vary in regression analysis.
    
   
5. Predict the CO concentration for a traffic density of 500 cars (no confidence interval). Is there likely to be a problem with this prediction? Explain why or why not.
    
   Question and data from the “Carbon monoxide from a Freeway” story at OzDASL (Australasian Data and Story Library)
    
   

4. A geographer conducted regression analysis for tree height = f(soil moisture) in a small local watershed. She notices some large outliers (positive and negative) in the residual plot. To investigate these outliers, she wants to know where they are located in the watershed. Luckily, she recorded the coordinates of her samples, so she prepares a map of the residuals. She compares the residual map to a map showing slope types (steep to flat).

Explain what the residual map indicates about the performance of her prediction equation. (i.e. where does the line of best fit equation over or under-predict tree height based on soil moisture?)

Show all your steps for conducting the analyses. For SPSS instructions, go to the SPSS Help manual and click to Statistical Tools Regression.

Marking Guide (Lab Total = 40)
Question		Mark
Q1		2
Q2		5
Q3	a	13
	b	6
	c	4
	d	4
	e	3
Q4		3

---

© University of Victoria 2000-2001     Geography 226 - Lab 9 Developed by S. Adams and M. Flaherty     Updated: 17 Nov2003