We conducted a RCT of four arms, each corresponding to different versions of the internet-delivered 4-week (8 sessions, 2 per week) CBM intervention. Time Frame is baseline (pre-intervention), post-intervention at week 4, 1-month follow-up. Eligible 94 participants were randomly assigned to one of four arms. Of these 94 participants, 82 adequately adhered to the protocol. 6 participants dropped out before the intervention; 5 participants drop-out during intervention; and 1 participant in the wait-list control dropped out at 1-month follow-up.
I would like to include all participants in the analysis regardless of adherence to the protocol (intention-to-treat analysis). I read many articles about the intention-to-treat analysis for repeated measures data. I would like to use a mixed models approach instead of analysis of variance for this purpose, as suggested by Gueorguieva and Krystal (2004). However, I didn't figure out how to conduct such analysis in SPSS, interpret and report.
I would greatly appreciate it if you helped me.
Thank you in advance.
You can use the SPSS mixed command with the repeated subcommand as below:
mixed bp by time diet /fixed diet time /repeated time | subject(id) covtype(ar1) /print = solution.
It is easy to exclude unwanted interaction terms by simply not including them on the fixed subcommand. You can change the covariance structure by altering the structure given in the covtype option on the repeated subcommand. A great thing about mixed effects models is that they don’t require subjects being measured at the same intervals. So you’d be definitely in good shape. if I can be of further assistance, please let me know.
Thank you so much for taking the time for sharing your knowledge with me. I would like to ask you kindly to inform me about participants', who dropped out, missing data before conducting the analysis. I am planning to use EM algorithm to impute missing values. (of course if data is missing at random). Is this the proper way?
In short: ITT analysis refers to handling losses of follow up in prospective trials, not missing values.
ITT is a recommended conservative approach - you should consider lost to follow up as an untoward event. Missing values, on the other hand, missing data is an statistical issue to apply some tests and should be treated appropriately.
Missing outcomes due to attrition (loss to attrition) is handled by censoring. I suggest you read up on survival analysis (also known as time-to-event analysis). I am attaching a link to a tutorial paper I wrote on the topic a long time ago.
In SPSS there are four standard methods to deal with missing data - you can find on:
www.csun.edu/sites/default/files/statistics20-core-system-guide-32bit.pdf
- Badges
- Science topic
- Similar topics
- Mathematical Sciences
- Statistics
- Statistical Software
- SPSS