What procedure would best fit my longitudinal data?

#1
Hi everybody

I want to thank you in advance for reading my post! ....and I hope my question is appropriate!
I am in a bit of a pickle!
I have been conducting a longitudinal study for my PhD. I followed children born 2010 and 2011 for three years collecting data on various symptoms, such as ADHD, externalizing behavior problems, emotion dysregulation etc. I started my data collection when the children where finishing their last year at preschool and ended my data collection when they finished 2nd grade in elementary school. I used both online questionnaires which parents and teachers answered about the children and also diagnostic interview I administered to parents of children who where reported above the cut-off score on any of the online questionnaires.

Now I want to explore if emotion dysregulation at time 1 is predicting behavior problems at time 3. What do you think is appropriate analysis for that - I have been looking at, for example, Latent Growth Model, Partial Least Squares, Discriminant Analysis Functioning. Should I be looking at something else and should I focus on linear models or non-linear models?

Thank you for reading my looooong post!
 

Karabiner

TS Contributor
#2
How exactely was "emotion dysregulation at time 1" measured, and how is "behaviour problems at time 3" defined and measured ?
Is the whole baseline sample included into the analysis, or is the analysis restricted to a subsample? How large is your sample size?

With kind regards

Karabiner
 
#3
Thank you for your question @Karabiner
Emotion dysregulation was measured with the questionnaire "Emotion Regulation Checklist" and behavior problems were measure with the questionnaire "Disruptive Behavior Rating Scale" and diagnostic interview Kiddie-SADS. The whole baseline sample is included into the analysis and my sample size is 620 children
 

Karabiner

TS Contributor
#4
Assuming that both variables are measured on an interval scale, a Pearson correlation
or a simple linear regression (of f'up problems on baseline dysregulation) would seem
obvious. If you have additional variables which you want to include into the analysis
(which you didn't mention, but would seem quite natural in an observational study),
for example as control variable, then multiple regression could do this.

With kind regards

Karabiner