Open ramiromagno opened 1 year ago
Hi Ramiro, it's good point and one we have discussed. The main thing is that the column index is consistent for effect size. Second, we originally had effect_size
as the label, but this would force users to read the metadata file to correctly interpret the data. Given that the metadata is in a separate file (an issue discussed elsewhere!), this was deemed potentially laborious so we've put that metadata into the column header itself.
Currently the standard allows for three alternative column names to represent the effect size:
beta
odds_ratio
hazard_ratio
This is a mandatory field. These alternatives, particularly when one of them is mandatory, makes the parsing more difficult.
I would prefer to have one column named
effect_size
, and then have the metadata explain which of the three cases is it.