[NYRepublican72]

How much would this change if the numbers were normalized for D+4?

It would roughly be Trump +4.

Here's the actual D/R/I amongst registered and likely voters, buried of course in the second-to-last paragraph of the ABC article.

"This ABC News/Washington Post poll was conducted by landline and cellular telephone Sept. 19-22, 2016, in English and Spanish, among a random national sample of 1,001 adults, including 651 likely voters. Results have a margin of sampling error of 3.5 points, including the design effect, for the full sample, and 4.5 points for likely voters. Partisan divisions are 33-23-36 percent, Democrats-Republicans-independents, in the full sample, and 37-27-28 among likely voters."

[CatOwner]

Having it go from Hillary +2 to Trump +4 under my assumption (D+4) instead of D+10 would happen if every DEM voted for Hillary and every REP voted for Trump. But that's not quite the case.

Still, I could see this going to Trump +2.5 or so, discounting the rest of the biased opinions that forms the basis of this poll.

[DrDude]

There is no “other” category. It has to add up to 100%. So which 8% did they leave out? There mere fact they don’t account for the “other” is significant. D+8 plus 8% unexplained. It is creative but doesn’t pass the smell test.