Open LukasWallrich opened 2 years ago
This is now in ESMARconf presentation - yet with wholly different set of considerations. How can we best enable an inclusion of this? Simply allow for duplicate effect size IDs and then compare models based on them - and require user to prep this in advance?
Analysts already have multiple degrees of freedom when converting effect sizes - e.g., to use Fisher's z transformation for correlations or to calculate the df in various instances. It would be great to show the impact of such choices on the estimates - but to do that, we'd first need to define a set of common decisions faced.