1
$\begingroup$

**Anova for repeated measures, mixed effects regression or survival **

I have an RCT where I compare 2x 40 slopes (20 different skills learned over time by 80 subjects):

  1. Condition one (n=40) is learning skills assisted by a smartphone.
  2. Condition two, control group (n=40) : non-smartphone assisted learning. I have daily entries over a year (360 measuring points), and I'd like to be able to show difference between two groups.

I'll be comparing Days used pr. learned skill (it will probably take 10-30 days to learn, each skill, more skills can be learned at the same time) I have cluster randomized at 4 different sites. Not all will learn all 20 skills in the 1 yr. timeframe.

I'm uncertain as to what statistic to use here? Survival with each learned skill as an "event" (will help me by the possibility to censor) linear Mixed effects: Slope (days needed pr. skill)~time + smartphone condition + error Maybe anova for repeated measures

Any suggestions as to what would give me best use of my data. Best picture of differing p values? My research question is does smartphone-registration enhance learning of skills ?

Im no stats wizard, so all input about these or other procedures are very welcome Kindly Stig

$\endgroup$
4
$\begingroup$

Not ANOVA. It would be more applicable if your dependent variable were measured each day as something continuous. Some of your subjects never learn a given skill, so their number of days to learn is undefined; that will interfere with ANOVA.

Survival analysis on the other hand seems promising. As to site, how relevant do you expect that to be to the learning speed? If fairly relevant, consider a hierarchical (multilevel) survival model, where nesting occurs within sites.

$\endgroup$

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.