The lack of a plateau in the average structure function (hereafter SF) for the complete flux-limited sample of Seielstad et al. (1983) [Fig.1] suggests that the mean time scale of radio sources variability at 10.8 GHz should be longer than the 3.8-yr time base of those observations. A comparison of the variability time scales derived from the SF analysis with those predicted by the statistical model of Magdziarz & Machalski (1993, AA, in print) [Fig.2] confirms consistency of that model. The mean time scale of variability of <τ> = 6±2 yr was found in both cases for v = 10.8 GHz.