|
|
Planned comparisons of slopes |
|
Message/Author |
|
Wu wenfeng posted on Sunday, December 12, 2010 - 4:24 am
|
|
|
Dear all, I did a HLM of longitudinal data, and discovered a three way interaction among gender, cognitive style and stress to predict depression, so I rerun the model(without gender) in both male and female, and estimate the slope of regression line of "high cognition high stress","high cognition low stress", and so on, I got the estimate values of slopes, their standard error and free degree(df). Now I want to do a planned comparison of the slopes, unfortunately,I don't know how to do, does any kind person to help me? thank you! |
|
|
You can use chi-square or loglikelihood difference testing of the model where the slopes are free versus the model where the slopes are constrained to be equal or you can use MODEL TEST. |
|
Wu wenfeng posted on Sunday, December 12, 2010 - 4:44 pm
|
|
|
Thanks for your advice! I run the Model in male and female respectively,if used chi-square or loglikelihood difference testing of the model, I just can compare the slopes difference in male or female.I hope to compare the slopes difference between male and female, how to do this? thanka again! |
|
|
You need to do a multiple group analysis where gender is the grouping variable. |
|
Wu wenfeng posted on Sunday, December 12, 2010 - 7:41 pm
|
|
|
thanks |
|
Wu wenfeng posted on Monday, December 13, 2010 - 4:09 am
|
|
|
Sorry to disturb you again! If I know the estimate value of slopes of regression linear, their standard error and the free degree(df), can I do significant difference test of slopes? how to do this? thanks for your any advice! |
|
|
If you want to do this by hand, you can do a z-test. You can find the formula in any basic statistics book. |
|
Back to top |
|
|