Trajectories in longitudinal data

gvleioras

Hi JASP team!

I have a dataset in which we measured academic expectations (5 scales) and college adaptation (3 scales) at two time points (end of first academic semester and end of second academic semester).

I am looking for a way to see if I can identify groups of students who have similar trajectories in their scores in these scales. Given the two measurements, I could expect three possible trajectories (increase, decrease, stable) per study variable. However, it would also be interesting to look for groups that have similar changes across scales (per study construct). For instance, to find that there is a group of students that remain stable in College adaptation scale 1, but increase in College adaptation scale 2.

Given my restricted knowledge of person-based statistics, I would expect a cluster analysis to be the right analysis. But is this so? Can I use cluster analysis for longitudinal data? And if so, what variables do I need to enter in the analysis to identify the trajectories I am looking for?

If cluster analysis is the right analysis, then which of the JASP option is appropriate? And if not, is there an analysis available on JASP that is appropriate?

Thank you in advance for your feedback!

Georgos.

EJ

Hi Georgos,

My main advice is twofold:

1. Be very careful. It is easy to explore the data with all kinds of tools and fool yourself into "detecting" a pattern.
2. I would use simple visualizations to see whether I could detect separate groups by eye.

But yes, most people would probably explore this using machine learning techniques (as included in our machine learning module)

E.J.