Silver Bulletin's weighted generic congressional ballot average stood at D+6.3 in mid-July, and Nate Silver wrote on Monday 13 July that it was "approaching D+7"1. The generic ballot asks which party's candidate a respondent would back in their own district, and it is the crudest available read on the national environment. The average had eased to D+6.1 on 30 June from a D+6.9 peak in late May, a dip attributed partly to June payrolls landing at 57,000 against a 115,000 consensus.
Nothing structural or economic arrived between 9 and 17 July to explain the climb back. Absent a trigger, the June easing looks like sampling noise that has now washed out, rather than a movement that reversed.
Silver notes that most of the panel's inputs are registered-voter polls, and that a likely-voter screen could put the true figure at D+8 or D+92. That cuts against the historical pattern, in which registered-voter samples overstate Democratic performance in midterms because turnout skews older and more Republican. The claim worth contesting sits there, in the inversion, not in the topline.
The panel also disagrees with itself more than the average admits. NBC News, Cygnal and TIPP show roughly D+5; YouGov and Morning Consult roughly D+3; Focaldata, J.L. Partners, Quinnipiac and Emerson range from D+7 to D+113. Eight points separate the ends of that spread, and a single number drawn across it carries a great deal of weight.
