Only one of Utah’s three professional pollsters was reasonably accurate with almost every prediction it made.
In the two weeks prior to the election, several pollsters tried their hand at forecasting the election results in Utah’s various races. I thought I’d find all the predictions and put them into a single table alongside the actual results.
The competitors
- Key Research with BYU’s Center for the Study of Elections and Democracy. Predictions were posted here. (Although I am involved with CSED, I was not involved with this poll. In fact, the first I heard of it was when the predictions were posted to this blog.)
- Utah State University student poll. Run by students, not by a professional polling firm. Predictions were posted here.
- Mason-Dixon poll, commissioned by the Salt Lake Tribune. Results for different races were posted at different times (1, 2, 3, 4, 5). Mason-Dixon revised its prediction in the Salt Lake County mayor’s race from R+10 to R+1; I put both numbers into the table below.
- Dan Jones poll, commissioned by KSL and the Deseret News. Results for different races were posted at different times (1, 2, 3, 4, 5).
- I took the official results from the Tribune since the state’s official website appears to be down right now.
How to read our comparisons
One major difference between prediction polls and official results is that prediction polls will have lots of “undecided” responses. To make a more valid comparison, I compare each poll’s predicted margin to the actual margin.
For example, the actual result in the presidential race (for Utah) was 73% Romney to 25% for Obama, a margin of 48 percentage points in favor of the Republican. I write that below as “R+48,” where “R” means “Republican” and “+48” refers to the margin. In this race, the BYU/Key poll predicted R+51 while the Tribune/M-D poll predicted R+45. Both were off by 3 points.
Comparison of prediction polls
In the table below, I use yellow to highlight any poll that was correct within 4 points. We’ll call this “reasonably accurate.” I use red to highlight any poll that was incorrect by at least 10 points. We’ll call this “wildly inaccurate.” Take a look.
| Office | Actual result | BYU, Key | USU | Trib, M-D | DNews, Jones |
| Pres. | R+48 | R+51 | R+53 | R+45 | R+43 |
| Gov. | R+40 | R+46 | R+59 | – | R+45 |
| Sen. | R+35 | R+39 | R+44 | – | R+47 |
| UT1 | R+46 | R+42 | – | – | R+57 |
| UT2 | R+28 | R+26 | – | – | R+16 |
| UT3 | R+52 | R+53 | – | – | R+50 |
| UT4 | D+1 | Tie | – | R+12 | R+4 |
| SLCo mayor | D+10 | – | – | R+1, R+10 | D+3 |
The results speak for themselves. Only one of Utah’s three professional pollsters was reasonably accurate with almost every prediction it made.
The exit poll
The Utah Colleges Exit Poll released its predictions at 8:00pm last night, immediately after the polls closed but before any official results had been posted. The poll is run by student volunteers who fan out across the state. The exit poll released updated predictions at 10:00pm after additional polling forms were returned by the student volunteers. The exit poll results were posted here (first you’ll see the 10pm numbers; scroll down to see the 8pm numbers). Let’s see how they did:
| Office | Actual result | Exit poll, 8pm | Exit poll, 10pm |
| Pres. | R+48 | R+48 | R+47 |
| Gov. | R+40 | R+41 | R+40 |
| Sen. | R+35 | R+36 | R+35 |
| UT1 | R+46 | R+46 | R+45 |
| UT2 | R+28 | R+21 | R+20 |
| UT3 | R+52 | R+61 | R+59 |
| UT4 | D+1 | D+3 | D+4 |
| SLCo mayor | D+10 | D+19 | D+22 |
On the whole, the exit poll fared very well. The predictions were off noticeably only in the Salt Lake County mayoral race. Of course, in posting the poll results (before the official results came out), Quin Monson pointed out that the poll hadn’t sampled enough people in Salt Lake County to make a firm prediction there.
Possibly related posts:
Possibly related posts (automatically generated)
Key Research Poll General Election Predictions (October 22, 2012)
