|
2012 New Hampshire Republican
Presidential
Primary
Pollster
Ratings The following table
lists the
accuracy of pollsters in the January 10, 2012 New Hampshire
Republican
Presidential primary using the measure
of polling accuracy
proposed
by Martin, Traugott, and Kennedy.
The order
of finish in New Hampshire was Mitt Romney 39.4%, Ron Paul 22.8%, Jon Huntsman 16.8%, Rick
Santorum 9.4% (23,312 votes), and Newt Gingrich 9.4% (23,174 votes).
An accuracy measure of 0.00 indicates
that the odds ratio of the survey results matches the odds ratio of the actual results (the
closer to 0.00, the better).
| 2012 NH Republican Presidential Primary
|
|
Romney |
Paul |
Odds
Ratio |
Accuracy |
|
Washington Times/
JZ Analytics |
38% |
24% |
0.92 |
-0.09* |
| NBC/Marist |
42% |
22% |
1.10 |
0.10* |
| PPP |
35% |
18% |
1.13 |
0.12 |
| Suffolk |
37% |
18% |
1.19 |
0.17 |
| ARG |
37% |
17% |
1.26 |
0.23 |
| Rasmussen |
37% |
17% |
1.26 |
0.23 |
| WMUR/UNH |
41% |
17% |
1.40 |
0.33* |
|
| Actual |
39.4% |
22.8% |
|
|
The accuracy measures for all seven polls fall
within the 95% confidence intervals. Also, there are no differences between pollsters at the
95% confidence interval.
* The accuracy measures for the Washington Times/JZ
Analytics survey, the NBC/Marist survey, and the WMUR/UNH survey in the match-up between Romney
and Huntsman fall outside the 95% confidence intervals.
Unlike the Washington Times/JZ Analytics and
the NBC/Marist surveys, the WMUR/UNH survey has the distinction of having the lowest accuracy
of all seven polls in the Romney/Paul match-up and having its accuracy measure for the
Romney/Huntsman match-up fall outside the 95% confidence interval.
If Suffolk University stopped its daily tracking
with the January 7-8 survey (Romney 33%, Paul 20%, Huntsman 13%), it would have been the most
accurate survey in the Romney/Paul match-up (odds ratio of 0.95 and an accuracy measure of
-0.05). Its accuracy measure in the Romney/Huntsman match-up, however, would have moved from
first place to third place.
These are accuracy measures for the Romney/Huntsman
match-up.
| 2012 NH Republican Presidential Primary
|
|
Romney |
Huntsman |
Odds
Ratio |
Accuracy |
|
| Suffolk |
37% |
16% |
0.99 |
-0.01 |
| Rasmussen |
37% |
15% |
1.05 |
0.05 |
| PPP |
35% |
16% |
0.93 |
-0.07 |
| ARG |
37% |
18% |
0.88 |
-0.13 |
| WMUR/UNH |
41% |
11% |
1.59 |
0.46* |
| NBC/Marist |
42% |
9% |
1.99 |
0.69* |
Washington Times/
JZ Analytics |
38% |
8% |
2.03 |
0.71* |
|
| Actual |
39.4% |
16.8% |
|
|
There are differences between the accuracy measures
for the Suffolk, PPP, and ARG surveys and the WMUR/UNH, NBC/Marist, and Washington Times/JZ
Analytics surveys at the 95% confidence level. There are no differences between the accuracy
measures for the Rasmussen, the WMUR/UNH, NBC/Marist, and Washington Times/JZ Analytics surveys
at the 95% confidence level. And there are no differences between the Suffolk, Rasmussen, PPP,
and ARG surveys at the 95% confidence level.
* The accuracy measures for the Washington Times/JZ
Analytics survey, the NBC/Marist survey, and the WMUR/UNH survey
in the match-up between Romney
and Huntsman fall outside the 95% confidence intervals. |
