Swing State Pres

Sunday, November 9, 2014

Grading BTRTN Election Projections: How Did I Do?


% Correct Among All Races
Overall it was a pretty good night for the forecasting.  But I missed three Senate races (assuming Alaska goes to the GOP, they are still counting absentee ballots).   The chart below is my scorecard in gory detail, and out of 499 races that are final (including several runoffs that will clearly go to the GOP), I called 484 correctly, or 97%.  (Eight races remain completely undecided; I’ve included a few that are officially still open as called because they seem pretty much over.)  But I am kicking myself over Alaska.  More on that later.

Certainly 97% sounds great, but of course most races are foregone conclusions.  Many Senate and Governor races are in deep red or deep blue states, and the propensity for all-but-uncontested races is even starker in the House due to gerrymandering.  By my count there were only 67 races that were competitive, or “in play.”  Of these, 30 leaned one way or the other, while 37 were toss-ups.  Among these 67 races, 62 are finalized, and I got 48 right, or 77%.  Among the true toss-ups, I was 23 for 33, or 70%.  I’ll take that; as best as I can tell, I was right up there with the professional services.

SENATE
Right
Wrong
Undecided
Total
Solids
28
0
0
28
Toss-ups
5
2
1
8
Total
33
2
1
36





GOVS
Right
Wrong
Undecided
Total
Solids
23
1
1
25
Toss-ups
6
4
1
11
Total
29
5
2
36





HOUSE
Right
Wrong
Undecided
Total
Solids
385
0
2
387
Leans
25
4
1
30
Toss-ups
12
4
2
18
Total
422
8
5
435





TOTAL
Right
Wrong
Undecided
Total
Solids
436
1
3
440
Leans
25
4
1
30
Toss-ups
23
10
4
37
Total
484
15
8
507





TOTAL %
Right
Wrong
Undecided
Total
Solids
99%
0%
1%
100%
Leans
83%
13%
3%
100%
Toss-ups
62%
27%
11%
100%
Total
95%
3%
2%
100%
Among Called
97.0%



In  Play
48
14
5
67
Among Called
77.4%




SENATE

I was 33 out of 36.  I have two wrong right now, North Carolina and Kansas.  And Alaska is likely to go into the wrong column as well, which I have assumed. 

I called a number of toss-ups correctly, including New Hampshire (the only one to go for the Democrats), Iowa and Colorado.  A week ago I put reasonably close races in Arkansas and Kentucky into the “solid GOP” column and clearly those were good calls.

North Carolina was razor thin, but I called it for Democrat Kay Hagan because she led in 19 out of the 27 polls conducted since October 1st , and four of the other eight were ties.  David Perdue led in only four.  The makeup of the race was similar to New Hampshire.  But on Election Day it was not to be, as Republican Thom Tillis pulled it out 49% to 47%.

Kansas was as close as could be; of the last nine polls, Independent Greg Orman led in six, incumbent Republican Pat Roberts in three, and the average was Orman up 43% to 42%.  While Orman was ahead, there were many undecideds, and clearly they broke for Roberts, the incumbent.  A lesson learned.

Alaska was somewhat different, and that is why I am kicking myself.  The polls were very erratic.  Five of the last seven polls had either Mark Begich, the incumbent Democrat, or Dan Sullivan, the GOP challenger, up by 4 or more points.  Two other polls were essentially tied.  I went with Begich largely because I had read he had a pretty good ground game and figured that that would really matter in a state as dispersed as Alaska.  Another lesson learned. Sullivan leads 49-45% but, again, this is not final.

The two runoffs were interesting.  I expected both to ultimately go Republican, so I take credit for being right.  Hoever, I also thought both would go to a runoff first, but only Louisiana will.  In Georgia, I thought the Libertarian candidate would siphon off enough votes to keep both candidates under 50%.  But Perdue won by eight points, 53% to 45% to only 2% for the third party.  His margin in the recent polls was on average only one point.

Nate Silver got only two wrong (Kansas and North Carolina) and Larry Sabato only got North Carolina wrong, so kudos to both of them!

GOVERNOR

I did worse among the Governors, 29 out of the 34 decided to date (two remain uncalled).  But there were fully 11 toss-ups and I’m 6-4 on them so far.

The real shock here was MarylandMaryland was the only “Solid” call thus far to go the other way.  Democrat Anthony Brown led in every single poll from May forward, on average by nine points – except one, an in-house poll (and I always exclude partisan polls from my analysis) for Republican Larry Hogan that had Hogan up by five with a week to go.  It turned out to be the only accurate poll out of them all: Hogan indeed won by five.

HOUSE

There are still five races out there, all in the 50/50 range.  Two of them were viewed as “Solid” as well.  But the House is becoming depressingly easy to call…even among the toss-ups, I was 12 for 16.  Only eight wrong overall thus far is a pretty good outcome.


Let the games begin for 2016!

No comments:

Post a Comment

Leave a comment