As I said in another topic, the image I see when I hear "landslide" is this giant, unstoppable force coming down a mountain and wiping out the totally helpless citizens below. So my metric for determining whether or not something was a landslide is not just the number of states won or anything like that, but how strong the victory was... or, equivalently, what percentage of the popular vote would need to be changed to reverse the victory. The bigger the percentage, the littler chance the opponent had of stopping the election, which is what I feel a "landslide" implies.
The percentage for each of the elections listed is roughly as follows:
1936: 11% shift to Landon.
1964: 13% shift to Goldwater.
1972: 12% shift to McGovern.
1984: 10% shift to Mondale.
So by this metric, the biggest landslide was 1964, with 1972 close behind, and then 1936 and 1984 respectively bringing up the remainder.
I disagree. A lot of Johnson's lead was padded by a massive national divide, which wasn't quite as present in the other races. It is hard to imagine how Goldwater wins, but it is not hard to imagine him getting 150 EV's or so if the election were a little different. But there are many states that Goldwater had no chance in.