In other words, in order to rigorously test the US, we need a sample of objects that are intrinsically similar. We therefore need to (i) quantify their definition, (ii) identify their orientation-independent parameters, (iii) find objects with a range of orientation and (iv) choose those which are matched in these parameters.
Step 1: We define Seyferts to be bona fide AGN (Sy 1s with broad lines of FWHM 1000 km and Sy 2s with ) that are (a) optically weak, (Schmidt & Green 1983), (b) in hosts of galaxy-type S0 or later, and (c) are radio-quiet, (Kellermann et al.1989). We thus exclude quasars, elliptical host galaxies and radio-loud objects.
Step 2: : In order to match the Seyferts in both their intrinsic AGN power and host-galaxy properties, we chose the following orientation-independent parameters: (a) The [OIII] luminosity, the bulge luminosity and stellar luminosity of the host, and its Hubble type. Since all of these parameters are optically derived, we also required that the host galaxy not be edge-on, in order to avoid selection effects due to extinction by the galactic disc.
Step 3: Range of orientation: Since in the radio-quiet regime a quantitative handle on the orientation of the AGN is not available, we simply chose equal numbers of purportedly pole-on and edge-on AGNs, Sy 1s and Sy 2s.
Step 4: Matching the pole-on and edge-on Seyferts: Before matching the Sy 1s and 2s, we needed to ensure feasibility of VLBI imaging. We arrived at a heuristic constraint that the Seyferts had to have a compact feature as imaged by the VLA-A or -B array with a flux density at 5GHz of mJy. The short list contained 54 Seyferts, of which we could chose 10 Sy 1s and 10 Sy 2s that conformed to our definition, that had non-edge-on host galaxies, and that were matched in the above orientation-independent parameters as well as redshift.
Vindication: It turned out a posteriori that the selected Sy 1s and Sy 2s are also matched in their total radio, mid-IR and far-IR luminosities.
Caveats: (1) We only choose Seyferts with previously known compact kpc-scale radio features. (2) The avoidance of edge-on host galaxies results in an avoidance of Sy 1(2)s with the AGN axis perpendicular(parallel) to that of the host galaxy.
Observations: We obtained 24hrs of observing time with a 14-station global VLBI array, during which we could observe 15 sample objects at 5GHz that had no prior VLBI data; data on the remaining five were added from the literature.