Currently, the subject_id variable in the data must be a valid index variable (i.e., start with 0 and be consecutive, without missing values). If they are different, this results in shape errors.
One fast way to prevent errors is to re-assign IDs, but this might be not ideal, because parameter estimates, etc. are then associated with a new index that was different in the data. Full support of non-consecutive indices, however, would require us to keep a mapping between indices and IDs and convert back and forth between the two formats, throughout the code.
For now, data going into the model should already be formatted with a consecutive participant ID, starting at 0.
Currently, the
subject_idvariable in the data must be a valid index variable (i.e., start with 0 and be consecutive, without missing values). If they are different, this results in shape errors.One fast way to prevent errors is to re-assign IDs, but this might be not ideal, because parameter estimates, etc. are then associated with a new index that was different in the data. Full support of non-consecutive indices, however, would require us to keep a mapping between indices and IDs and convert back and forth between the two formats, throughout the code.
For now, data going into the model should already be formatted with a consecutive participant ID, starting at 0.