That’s according to the latest Morning Call/Muhlenberg survey, which has Hillary Clinton at 44% and Donald Trump at 41%.
Just a week ago, Morning Call and Muhlenberg College had Clinton with a 47% to 38% advantage over Trump.
If Libertarian Party nominee Gary Johnson and Green Party nominee Jill Stein are included, Clinton’s lead over Trump narrows to two points, 40% to 38%. Johnson and Stein get 8% and 3% respectively.
Digging into the crosstabs, Trump’s gains seem to come from additional support from Republicans (85/5) and the middle of the commonwealth (57/31). On the other hand, Clinton leads with Democrats (78/9) and Independents (40/34) as well as Southeast residents (57/26). Trump edged ahead among Southwest residents (44/42).
The gender and educational gaps were both present again. Clinton has an eleven-point advantage with women (48/39) and college graduates (49/38). Meanwhile Trump is ahead by five with men (45/40) and six with non-college graduates (45/39).
The biggest discrepancy, however, is racial. By a 49% to 38% margin, whites favor Trump while non-whites prefer Clinton 68% to 13%.
Finally, 18 to 29 year-olds are with Clinton 51% to 29% whereas 50 to 64 year-olds favor her 45% to 43%. 30 to 49 year-olds are tied 42% to 42% while those over 65 years-old prefer Trump 47% to 41%.
How did one week make such a difference? As implausible as it may seem, there might not have been any change at all.
It really doesn’t really make sense for Clinton to post a strong result in the midst of a public bout with pneumonia, but then have worse numbers a week later when no major events happened in between.
It’s likely as simple that both surveys were outliers and/or subject to sampling error. It may well be the case that a nine-point lead is Clinton’s best-case scenario and a three-point lead is the worst. Monday night’s debate could very well swing the numbers between the two poles of a solid Clinton lead and a within the margin of error toss-up.
The Morning Call/Muhlenberg Poll was conducted by Muhlenberg University from September 19th to 23rd. They surveyed 486 likely Pennsylvania voters. The margin of error is +/-5%.