The distribution of QSO radio luminosities has long been debated in the literature .
Some argue that it is a bimodal distribution , implying that there are two separate QSO populations ( normally referred to as ‘ radio-loud ’ and ‘ radio-quiet ’ ) , while others claim it forms a more continuous distribution characteristic of a single population .
We use deep observations at 20 GHz to investigate whether the distribution is bimodal at high radio frequencies .
Carrying out this study at high radio frequencies has an advantage over previous studies as the radio emission comes predominantly from the core of the AGN , hence probes the most recent activity .
Studies carried out at lower frequencies are dominated by the large scale lobes where the emission is built up over longer timescales ( 10 ^ { 7 } -10 ^ { 8 } yrs ) , thereby confusing the sample .
Our sample comprises 874 X-ray selected QSOs that were observed as part of the 6dF Galaxy Survey .
Of these , 40 % were detected down to a 3 \sigma detection limit of 0.2–0.5 mJy .

No evidence of bimodality is seen in either the 20 GHz luminosity distribution or in the distribution of the R _ { 20 } parameter : the ratio of the radio to optical luminosities traditionally used to classify objects as being either radio-loud or radio-quiet .
Previous results have claimed that at low radio luminosities , star formation processes can dominate the radio emission observed in QSOs .
We attempt to investigate these claims by stacking the undetected sources at 20 GHz and discuss the limitations in carrying out this analysis .
However , if the radio emission was solely due to star formation processes , we calculate that this corresponds to star formation rates ranging from \sim 10 M _ { \odot } yr ^ { -1 } to \sim 2300 M _ { \odot } yr ^ { -1 } .