Down the home stretch

Currently Hillary Clinton leads Donald Trump by about seven percentage points in the polls (48.5% to 41.5%).  That's a good lead, but not an overwhelming one--it's a little smaller than the lead that Barack Obama had over Mitt Romney at this time in 2012.  I looked at the polls going back to 1952, picking the one or two that were closest to August 25.*  The closest race was in 1960, when Kennedy and Nixon were tied at 46%.  The most lopsided was in 1964, when 67% said they were for Johnson and only 26% said they were for Goldwater.  The difference between Democratic and Republican shares in the actual votes (V) could be predicted from the difference in the polls in August (A) by:
V=.63*A
For example, in 1952, Eisenhower led Stevenson by 55%-38% in August, so A was -17.  The predicted margin in the actual vote was .63*(-17)=-10.7, which was almost exactly equal to the actual margin (55.2% to 44.3%).  Clinton's +7 leads to a predicted margin of 4.4%.  The standard error is about 4.2.

The biggest residual was in 1980, when Carter and Reagan were tied in August, but Reagan went on to win easily (about 51%-41%).  I think that even the final polls showed a close race, but I also recall than Carter's campaign seemed to be floundering in the last few months.  So probably some of it was survey error but some of it was a real change.  The other two large residuals were in 2008, when the polls were pretty much tied in August, and in 1956, when Eisenhower was ahead by 13 in August and increased the margin to 15 in September.  The explanation for 2008 is obvious--the financial crisis that started in September 2008.  I don't know anything about the details of the 1956 campaign, but my guess is that since it was a rematch of 1952, people made their minds up earlier than usual.

The experience of 1952-2012 suggests that a Trump victory is unlikely but not impossible (maybe 15%).  A Clinton landslide (say a margin of 10 or more) is also unlikely.

The raw data:

     Aug  Final
1952 -17 -11
1956 -13 -15
1960   0   0
1964  41  24
1968  -6  -1
1972 -37 -23
1976   9      2
1980   0 -10
1984 -26 -18
1988  -8  -8
1996  15   9
2000   5   1
2004  -1  -2
2008   0   7
2012   9   4

*I omitted 1992. because Ross Perot had dropped out, but resumed his campaign in September.

[Data from iPOLL, Roper Center for Public Opinion Research]