Martin White makes estimates of contribution to power spectrum in section 3.3 and figure 2 of the first DASI analysis paper (97). He uses a luminosity function (flux distribution) at 1.5 GHz from the FIRST survey as provided in section 4.1 of Tegmark 1996. They both make the usual weak noises about assuming a single fixed spectral index to extrapolate to higher frequency.
Both White and Tegmark quote a guy Toffolatti who has made a career out of multi frequency source count modeling etc. He made a contribution to the 1999 microwave foregrounds book Section 3 of Toffolatti 1999 (in Microwave Foregrounds book so a-ph only) provides formula to convert dN/dS at one frequency to another on the assumption of a Gaussian distribution of spectral indexes. The model used here is described in Toffolatti 1998 MNRAS a-ph. Latest greatest paper from this series discusses millimeter counts from SCUBA/MAMBO etc Zotti 2002.
In principle if one has the differential source count spectrum at one frequency and the distribution of spectral indexes (flux ratios) to another frequency one can compute dN/dS at the second frequency. However published measurements of spectral indexes are usually incomplete samples. The best I've found between 5 and 31 GHz is Geldzahler 1981 which contains 333 5 GHz selected sources 232 of which are detected at 31 GHz. Geldzahler 1983 extends this work. The measurements made for us by OVRO guys were selected S_5>100mJy and are 95% complete.
Anyone who thinks radio sources have simple power law spectra (ie the spectral index distribution between two frequencies can be used between another two) should check out Herbig and Readhead 1992.
The VSA group has conducted a small 15 GHz survey Taylor 2001. They quote dN/dS at 15 GHz between 20 and 500 mJy.
There is a paper from the OVRO/BIMA SZ group on point sources..
Both Green Bank 5 GHz survey papers (ApJS 75 (1991) 1011 and updated ApJS 103 (1996) 427 from ADS) give tables of differential source counts and compare against the model of ApJ 287 (1984) 461, although on a quick scan through that paper the parameters and equations to generate the curve are not immediately apparent to me.
The PMN survey papers on a quick scan do not appear to give source count tables. For both PMN and GB6 I have source list tables and generate my own dN/dS distributions...
Nartallo 1998 contains 1.1 and 0.8 mm polarization data for a sample of 26 sources in the 1-10 Jy range. Mean polarization fractions run 0-15%.
Kronberg group papers: 59 @ 13.5GHz Vallee 1974, 79 QSO @ 9Gz Wardle 1974, 61 @ 9GHz Kronberg 1977, 68 @ 10.5GHz Simard 1978, 100 @ 15GHz Simard 1981, 142 @ 9GHz Simard 1981, 68 @ 10.5GHz Simard 1982, 154 @ 5GHz Zukowski 1999. This last paper the sources are in the 100mJy few Jy range. Distribution of polarization fractions looks exponential from 0-10%. These guys are after Faraday rotation in extragalactic space. Therefore their sample may be less biased towards exotic objects than others.
Simard-Normandin 1981 555 extragalactic rotation measures.
Mesa 2002 NVSS and VLA based estimates of polarization fraction distribution versus frequency and contribution to power spectrum.