Astronomy databases include both homogeneous and inhomogeneous ones (i.e. collected from a variety of sources). Recall there is no such thing as perfect data, so one should always consider how the data were obtained and potential limitations. This may become more of an issue as we evolve towards larger and larger projects for which automated processing becomes essential.
Example: pipeline data quality: bogus SDSS quasars in the EDR
Example: selection effects: high-redshift rotation curves