What do you consider to be a landslide?

[muon2]

It's useful to look at more than the last couple of elections, so I've graphed some data for the elections from 1900 to 2008. This first chart is the EV total of the winning candidate compared to their advantage over the other party's candidate in terms of the fraction of the vote cast.



Based on the graph, 400 EV is much more reasonable than 350 for a cutoff for a landslide. A cutoff at 400 EV would make half of all the elections since 1900 landslides. I think that may even be too generous, since saying half of the elections are landslides weakens the notion of a landslide.

Placing the landslide threshold at 450 EV would reduce that to 25% of the elections since 1900 and make the notion of a landslide more meaningful to me. A threshold of 450 would say that Reagan won with landslides in both elections, but GHW Bush did not in 1988.

The comparison between EV and fractional margin is fairly linear. The next chart shows the GOP EV total compared to the fraction of GOP votes minus the Dem fraction from 1932 to 2008.



The trend is quite linear for fractional margins within +/- 0.10 (10%). That same range also is a good fit to EC wins of less than 450. Note that for above 450 EV or under 90 EV the trend flattens out suggesting that a threshold in the behavior of the EC has been reached. This behavior tends to reinforce the choice of 450 EV or a 10% popular vote margin as an appropriate level to declare a landslide.

[muon2]

It's useful to look at more than the last couple of elections, so I've graphed some data for the elections from 1900 to 2008. This first chart is the EV total of the winning candidate compared to their advantage over the other party's candidate in terms of the fraction of the vote cast.



Based on the graph, 400 EV is much more reasonable than 350 for a cutoff for a landslide. A cutoff at 400 EV would make half of all the elections since 1900 landslides. I think that may even be too generous, since saying half of the elections are landslides weakens the notion of a landslide.

Placing the landslide threshold at 450 EV would reduce that to 25% of the elections since 1900 and make the notion of a landslide more meaningful to me. A threshold of 450 would say that Reagan won with landslides in both elections, but GHW Bush did not in 1988.

The comparison between EV and fractional margin is fairly linear. The next chart shows the GOP EV total compared to the fraction of GOP votes minus the Dem fraction.



The trend is quite linear for fractional margins within +/- 0.10 (10%). That same range also is a good fit to EC wins of less than 450. Note that for above 450 EV or under 90 EV the trend flattens out suggesting that a threshold in the behavior of the EC has been reached. This behavior tends to reinforce the choice of 450 EV or a 10% popular vote margin as an appropriate level to declare a landslide.

Agreed on the  EV question. However, I would say that a 10% margin is not a ladslide, but just a solid victory...

But the data in the second graph (which goes back to 1932) show that a 10% margin is nearly equivalent to a 450 EV total. The strict exceptions would be: 1980 -- 489 EV 9.7% margin; 1952 -- 442 EV 10.8% margin; 1940 -- 449 EV 10.0% margin. The percentages are extremly close to 10% in all three cases.

Before 1932 the effect of the South on Republican victories throws off any clear analysis along this line.

[muon2]

Yeah, I just don't like it on a theorical point : I consider a landslide to be a really outstanding margin of victory ( 15 or more ). I don't consider Ike'52 or Reagan'80 to be real landslides, just comfortable victories.

Then you could achieve that by requiring both 450 EV and greater than 10%. That would exclude the elections of '52 and '80. It is hard to get a feel for whether 10 or 15 makes more sense. Since 1912 no winner has had between 10.8% ('52) and 15.4% ('56) and 1912 (14.4%) had less than 450 EV. With that gap we have no way of saying whether a 13% win would feel like a landslide by historical proportions. That's why I looked to a correlation in the data.

[muon2]

At least 400 EVs and at least an 9-point or 10-point margin between the two candidates, such as Bush vs Dukakis in '88 or Reagan vs Carter in '80.  ('84 was a given).

1988 a landslide ? LOL

Well given my standards of at least 400 electoral votes then yes it was a landslide.

Many of the thresholds proposed here seem to make an awful lot of landslide elections. Maybe posters think that either an election is close or a landslide with very little room in between. I think that both very close elections and landslides are at the edges of the spectrum and that comfortably large wins should be the norm when elections are graded.

I think the best idea is to determine what elections are landslides, then set a threshold so that only those landslide elections pass the cut. For instance, I wouldn't call the 1988 election a landslide nor would most observers who watched it. It was a solid win, but that doesn't make it a landslide.