4
$\begingroup$

I have done a RCT (randomized control trial) of an intervention (which is of 6 week duration) versus control. The outcome measure (an exercise test) has been tested pre-intervention (Pre), immediately after (Post-1), 6 months after (Post-2), and one year after intervention (Post-3).

I have done a repeated measures ANOVA in SPSS. There is no group difference at Post-3. However, looking at the means/plots at each time point there appears to be a difference between the groups at Post-1.

How would I test to see if there is a significant group difference at Post-1?

$\endgroup$
  • 1
    $\begingroup$ Welcome to this site, @neil. I considerably edited your question. Please check and re-edit if necessary. Also, what does mixed model tag do here? Did you mean ANOVA? $\endgroup$ – ttnphns Mar 1 '13 at 5:30
  • 1
    $\begingroup$ I edited references to time-points, hopefully it's a little easier to read now. If you are after a fun and readable chapter on repeated-measures ANOVA, Discovering Statistics Using SPSS by Andy Field is a good start. With multiple time points, you might also consider how you would deal with missing data, as repeated measures ANOVA does list-wise deletion. $\endgroup$ – Sootica Mar 1 '13 at 21:32
  • 1
    $\begingroup$ It is very bad practice to look at the descriptive results, pick the time point with the largest effect and then happily report the significant result. Each time point needs to be treated and reported in a similar way unless otherwise stated in the study protocol. $\endgroup$ – Michael M Oct 28 '13 at 6:36
1
$\begingroup$

Although the answer could be very extensive explaining all aspects of a repeated measures design I will keep it short and answer the question directly. I think, however, that it is wise to do a bit more study on your own to be able to comprehend the answer completely.

  1. You have to look whether the interaction between the group variable (control vs. intervention) and the factor variable (the measurements over time). Although you can proceed to the next step either way it is advised that you only do so in case this step is significant.

  2. In the SPSS repeated measures menu you click options >> select interaction term >> drag to the right column >> click continue >> click ok. You now get descriptive statistics for each group for each time interval. Using the confidence intervals you can determine whether or not the two groups differ on a certain time point.

The syntax you have to add in the second step is:

/EMMEANS=TABLES(Group*factor1) << replace variable with your variable
$\endgroup$
0
$\begingroup$

Are you sure you want to know whether the groups are statistically different or if there is a significant time by group interaction? I also do exercise studies and I am interested if the parameter changes over times significantly different in the respective groups. This would be the line labelled time*group.

Regarding the question directly: When I am interested in the changes between Pre and Post-1 I just do a seperate repeated measures ANOVA for this two time points.

$\endgroup$

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.