Is it appropriate to use the SMD when some studies report raw data and some report log transformed data? ie cortisol data
Please have a look on the below articles (links). The articles have reported some classic approach for dealing with skewed data.
https://www.ncbi.nlm.nih.gov/pmc/articles/PMC2978323/
https://www.ncbi.nlm.nih.gov/pmc/articles/PMC4120293/
Unfortunately, you can't mix them. I think your best bet is to try and get the 2 means and SD for the logged data, and back transform them, to get the correct SMD.
Am I correct in understanding the articles posted are discussing analysis of data where the outcome is always skewed, but sometimes has been transformed and sometimes not?
My circumstances is that in most cases the raw data weren't skewed, and consequently data has only been transformed in a few cases. Is is acceptable to run SMD analysis in such a circumstance?
The standardized mean difference (SMD) is always suggested to use in meta-analysis when the studies evaluate the same outcome but measure it in a variety of ways (for example, measuring low back pain [pain intensity] via visual analogue scale [VAS] scores [scale range, 0–100] and numerical rating scale [NRS] scores [range, 0–10]).
Your raw (cortisol) continuous data may be available in mean with either SD/SE/CIs or their log-transformed measurements.
As Adrian Esterman said already, you cannot mix raw and log transformed data under SMD.
Use the formulae given in the article "Meta-analysis of skewed data: Combining results reported on log-transformed or raw scales" to convert to either the mean and SD for raw measurements or the mean and SD for log-transformed measurements.
Some more information from the Cochrane Handbook for Systematic Reviews of Interventions.
http://handbook.cochrane.org/chapter_7/7_7_3_4_transformations_and_skewed_data.htm
http://handbook.cochrane.org/chapter_9/9_4_5_3_meta_analysis_of_skewed_data.htm
try this paper
Article TO DETERMINE SKEWNESS, MEAN AND DEVIATION WITH A NEW APPROAC...
- Badges
- Science method
- Similar topics
- Mathematics
- Statistics
- Statistical Techniques
- Meta-Analysis