I wanted to test the effectiveness of a particular type of "talking" therapy on depression. I envisaged selecting ONE group of people and measuring their heart rate and scores on depression scale for a period of 8 weeks. So I would have 8 measurements for the heart rate (session heart beat average) and 8 measurements for depression scale. How do I go about choosing the right statistical test, and how do I get my head around this in SPSS?!

  • 3
    $\begingroup$ That would be a very advanced analysis. If you aren't terribly familiar w/ stats, you may want to work w/ a statistical consultant. $\endgroup$ – gung - Reinstate Monica Oct 11 '12 at 21:10

One way to approach this is to use multilevel modeling for longitudinal data.

If you would like to stay in SPSS, it is possible too. Below is a very readable book available through Amazon on this topic. It has a nice introduction to this method in general, and step-by-step examples with screenshots.

Multilevel and Longitudinal Modeling with PASW/SPSS (Quantitative Methodology Series) by Ronald H. Heck, Scott L. Thomas, & Lynn N. Tabata

For adding time-varying covariates, please see Chapter 6, Model 2, Adding Time-Varying Covariates.

| cite | improve this answer | |
  • $\begingroup$ If you are going to use multilevel modeling, or latent growth modeling (from structural equation modeling framework) which has similar capacity in handling the type of data you have, it would be very helpful to first read and study these methods. Most books on these use other softwares such as Mplus, R, HLM, MLwiN, and it's easier to find examples outside SPSS environment (which means making life easier in a long-term). Also these methods are often quite demanding on sample size, which is not always easy to get with intervention studies. $\endgroup$ – Sootica Apr 2 '13 at 7:25

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.