To give you a breif summary of my work, I am mostly looking at the percentage time a participant spends in a certain play state when compared to the entire observationn duration. I was using parametric measures to analyse the data provided that all assumptions were met. However, I have recently come across CoDA (Compositional Data Analysis) and a recent study which shows significant differences between standard tests and CoDA tests. My question is whether it is neseccary to redo my data analysis or whether it would be ok to continue using standard testing meassures. Most instructions on CoDA also get a little bit too technical for me to understand. I have been struggling to really wrap my head around the differences betweeen the methods and how to accurately do CoDA. I am attaching the paper citation mentioned for your reference.
Gupta, N., Mathiassen, S. E., Mateu-Figueras, G., Heiden, M., Hallman, D. M., Jørgensen, M. B., & Holtermann, A. (2018). A comparison of standard and compositional data analysis in studies addressing group differences in sedentary behavior and physical activity. International Journal of Behavioral Nutrition and Physical Activity, 15(1), 53.
Hi Kiyoshi,
I think it depends on what you are testing, if just differences between absolute time spent in MVPA, for example, then your standard tests would likely suffice.
However, you must acknowledge you are unable to account for all the other concurrent behaviours, or changes in those behaviours.
For Coda, the data must be converted to isometric log ratios, to account for compositional nature of the data. As you said, there can be quite a bit to get your head around. I would suggest searching "opencoda" and taking a look there. They have set up an online programme where you can run some. Coda.
I'm happy to chat further too, my email is [email protected]
Best wishes
Cain
Hi Kiyoshi Bhuiyan , compositional methods and conventional methods can give occasionally similar results, and sometimes really different ones. The problem is you never know, and sometimes conventional methods can give spurious results. I would recommend to use compositionally compliant methods. You can ask Javier Palarea-Albaladejo in Edinborough for advise.
- Badges
- Science topic
- Similar topics
- Mathematical Sciences
- Statistics
- Statistical Software
- SPSS