Tests of the CURN model. The top two panels show that pulsar spectra of temporal correlations attributed to CURN are indeed consistent with representing the same spectrum (Goncharov et al. 2022). The blue lines show the measurement for DR2full, whereas the orange lines show the measurement for DR2new. The top left panel is the measurement of the standard deviation of the CURN amplitudes across pulsar spectra, σ_log10A, marginalised over the mean of the CURN amplitudes, μ_log10A, as well as pulsar-intrinsic noise parameters. The dashed vertical line denotes an upper limit at 95% credibility. The top right panel shows both the mean and the standard deviation. The inferred mean value is consistent with the measurement of log₁₀A_CURN denoted by a dashed vertical line. The bottom panel shows the logarithm of the dropout factors (Arzoumanian et al. 2020) which suggest pulsars’ contribution to the CURN model. Positive values represent support for the CURN model, whereas negative values point towards inconsistency of pulsar data with the CURN model. The differences in the dropout factors between the data sets could be due to differences in the recovered CURN or the pulsar red noise spectral properties.