Whether the methodology is fatally flawed or a consequence of the shy Trump voter phenomenon, the post-election polls do not inspire much confidence that the problem has been adequately addressed. If calibration is required, that hasn’t been done either. From a poll released yesterday from SurveyUSA concerning the Georgia Senate runoff elections in January comes this information about the composition of the respondent pool:
Of the Georgia registered voters polled, 47% reported voting for Biden and 43% for Trump. Barring changes from additional recounts or disqualified ballots, Biden will have narrowly edged Trump by 0.2% in the state, not a full 4%.
That the poll finds both Democrat Senate challengers in the lead, Ossoff up by 2 points and Warnock up by 7 points, is entirely unsurprising in this context. So unsurprising that given how recently the presidential election occurred it would seem advisable to randomly cull Biden voter responses until the composition of the respondent pool is at parity with regards to the presidential vote. The blog is downstream of the polling operations. We want clean water, not this detritus!