That graph doesn’t match the headlines about Clinton maintaining or expanding the national lead. If shows the race tightening.
It’s just torture looking at this stuff. I try not to do it, but always fail. However close this looks, I’ll be surprised if Clinton doesn’t win easily. I just want everything else to assure me of that. I think we’re all in for a lot of worry until this godawful election is over.
Try pollster.com or fivethirtyeight.com or election.princeton.edu for poll graphs that are smoothed better (pollster) or weight and adjust polls better (fivethirtyeight) or try to look at the bigger picture (election.princeton.edu)
TPM’s Polltracker just jumps all over the place and tends to induce whiplash.
It also doesn’t match the Clinton number in the article, which puts Clinton at 47.6, while the graph has her at 44.7. That’s a decline of nearly 5 points since yesterday. I just don’t believe that.
TPM might as well bag it on their poll-tracking. It makes zero sense and it doesn’t even look like they are doing the math right. They had it at 8.5 points on Friday but now it’s down to 2.7 even though Clinton went UP in most of the polls release over the weekend and today – even Rassmussen?
The REAL difference is probably closer to the 8.5 TPM had on Friday, based on the polls coming out recently including NBC, ABC, GWU etc. Internet polls under-report Clinton’s support by about 3%, and ABC has been decidedly anti-Clinton this entire cycle, probably because of how they weight the potential electorate.
Fivethirtyeight.com is probably the most rigorous and analytical of the polling aggregate sites, but in trying to be “fair” they overvalue internet polling with larger sample sizes, even when they clearly may just reinforce a sampling bias (not intentionally) against minorities less likely to use the internet a lot.