MacLin's Advanced Statistics Blog

This is a site for our advanced (graduate) statistics class...

Tuesday, November 07, 2006

HomeworkDue Friday

Using the data from exercise 15.4 conduct a regression analysis first usign the enter, then stepwise, and then backwards regression. Based on your reading of section 15.10 (page 520) discuss the different results and how the different methods might have lead up to those results.

Using the data from exercise 15.16 use stepwise regression to analyse the variables. Report and discuss yoru results. Next use the regression equation to calculate Y hat. Using the vales obtained from Y hat, calculate the residuals for the for the regression line. Next, repeat this using the enter method. Compare your results from the two methods, indicate which one produced a better fitting line.

Using the data from 15.16, run a stepwise and then enter method analysis request SPSS to conduct the diagnostics. Report the statistics from diagnostic test and compare them with the results from your previous analyese above (where you calculated your own residuals).

5 Comments:

At 7:46 PM, Anonymous Anonymous said...

Hi Dr. MacLin,

The 15.16 data does not produce significant results using the "enter" regression analysis and the stepwise method can't be conducted for some reason either. I did the backward procedure, and that worked, so I wrote up my results that way and did the regression equation and residuals using the output from the backward data, but I don't know what else you want us to do for the homework questions involving y_hat and residuals since the other two methods don't produce significant results.

Also, what would you like us to do for the third part of the homework since the stepwise can't be conducted? The diagnostics for the enter method can be done, but they may not be correct since the results are not significant.

Just curious where you want us to go from here! Thanks!!! :)

 
At 7:09 AM, Blogger maclin_uni said...

Do the best you can with this. It is important for you to know under what conditions one method might work and another might not. Using the backwards regression was a good choice under these circumstances if it works.
I think you have taken a good approach with this.

--Dr. M

 
At 10:41 AM, Anonymous Anonymous said...

When I try to run the stepwise analysis for the 15.16 data nothing happens, my output just says "Regression" and "Variable Entered/Removed, Dependent Variable:Y" I have all the X variables entered as independent but I get no data. Any ideas?

 
At 11:08 AM, Blogger maclin_uni said...

I think it is because the variables are not significant when you run the stepwise (see comments above). That will happen sometimes so in real life you will have to deal with what the data give you. For the homework, try doing a backwards regression and see if it is significant. It is really not a bad learning situation to have non significant data - although for homework it can be frustrating. Just do the best you can, worst that can happen is that Jen will give you feedback to make corrections. Thanks,

--Dr. M

 
At 11:22 AM, Anonymous Anonymous said...

If you want to do stepwise regression analysis when you have unsignificant results, just go to options withing the regression window and you will see two values, one is entry (which I'm not sure what that is) and the other says removal which I assume is the significance level at which the variables will be removed from the analysis. If you change the significance level here it will let you run the analysis.

 

Post a Comment

<< Home