|
|
Intercept with slope is not significant |
|
Message/Author |
|
Iris Au posted on Monday, April 22, 2013 - 6:48 pm
|
|
|
hi, in my analysis, i with s is not significant. slope mean and variance are not significant as well. however, some predictors predict slope significantly. is this possible? is this still useful analysis? looking forward to hearing from you. thanks, |
|
|
Yes and yes. For instance, the slope may be different for males and females, where gender is a predictor of the slope. But there is no variation in the slope within gender - that would imply a zero residual variance. Significance of a slope variance is often not found without covariates, while the slope still varies as a function of covariates when they are added. |
|
Iris Au posted on Tuesday, April 23, 2013 - 8:53 am
|
|
|
Thank you so much! |
|
Iris Au posted on Monday, April 29, 2013 - 4:43 am
|
|
|
Hi, I would like to analyze trajectory of elderly depressive symptoms over 6 years. Add to this, I would like to test if the trajectory influences on suicide ideation at 6th wave. In my data, suicide ideation is binary variable which was measured with yes or no question. In this case, do I have to apply logistic regression analysis in order to add suicide ideation variable as the outcome of depressive symptom trajectory? Is there any method for it? Looking forward to hearing from you. Thank you very much in advance. |
|
|
You can regress the binary variable on the growth factors. With weighted least squares, it is a probit regression. With maximum likelihood, it is a logistic regression. |
|
Iris Au posted on Tuesday, April 30, 2013 - 7:30 am
|
|
|
Thank you for your help! |
|
Back to top |
|
|