13 Keys to the White House (leaning)

[Bush did 311]

1992, 2000, 2004, 2012, 2016, 2020

Really this is the sample size of elections he predicted where the outcome is not self evident by October, and I'm being kind of generous allowing 1992 and 2012. Either he got 2000 or 2016 wrong depending what he claims the model does, so 18% failure rate for competitive elections.

Basically worthless.

[Randy Marsh]

13 keys really strikes me as overfitting past data and expecting it to hold when all you did was fit noise (with malleable subjective keys, too). Doesn't help that number of keys doesn't seem to be correlated at all to margin of victory.

But his track record is exemplary so I don't think we can write it off.

His record is 8 for 9, and that 8 includes freebies like 1996 and 2008. Impressive would be pushing it with that sample size, let alone exemplary

8/9 is better than polls did

Who did better?

[robocop]

I would like to know what the "major policy change" is that the article confidently states as TRUE is?

I‘ll give you a hint: what were the two high-profile bills passed during the 117th Congress?

Infrastructure Bill?

OK but we need concrete definitions on how major and significant it is. Looking at his past predictions the only "major policy change" bills he refers to seem to be Reagan tax cuts in his first term and Obama's Medicaid in his first term.

Again some of these keys definitions can be very ambiguous.

[SWE]

13 keys really strikes me as overfitting past data and expecting it to hold when all you did was fit noise (with malleable subjective keys, too). Doesn't help that number of keys doesn't seem to be correlated at all to margin of victory.

But his track record is exemplary so I don't think we can write it off.

His record is 8 for 9, and that 8 includes freebies like 1996 and 2008. Impressive would be pushing it with that sample size, let alone exemplary

8/9 is better than polls did

Who did better?

No it's not

The only popular vote post convention polling miss (remember, this is what Lichtman designed the keys to predict in 1988) was 2000. If you rewrite history to make the keys about the EC to save Lichtman from the 2016 embarrassment, then 2016 is the only polling miss. If you want to do be better than the 13 keys, just vaguely pay attention to the state of the race

[American2020]

https://www.youtube.com/watch?v=PB8JazHcQ_k

[Del Tachi]

The "keys" are so weasely and non-specific that you could make the test say anything you wanted.  Only with the benefit of hindsight does it become obvious that the test "worked."