What do we recommend for qualitative research, where replicability is not a quality criterion?