Anne Chan posted on Tuesday, April 17, 2012 - 12:56 am
Hello! I run a linear latent growth curve model on a repeated measure, with gender as a covariate (female = 1, male = 0). The slope is insignificant (slope = 0.3, p = n.s.), but there is a significant gender effect on the slope (standardized estimate = 0.21, p<0.05).
May I ask how should I interpret this finding? Or it does not make any sense to interpret a significant predictor effect on a insignificant slope? Thanks!
In a model with a covariate, it is the residual variance of the slope growth factor that is not significant. A variance is estimated only when covariates are not present. So the variance may be significant. Even if it is not, you may have more power in a model with a covariate.