These were internet panel polls conducted by YouGov Nov. 15–19. YouGov gets a C+ rating from 538. All participants were registered voters who had previously voted in the Democratic Party’s primaries or caucuses. The majority were participants in CBS/YouGov’s October polls who were recontacted to participate in this month’s polls.
Summary by YouGov
Shorter summary by CBS News
CBS/YouGov “Battleground” Poll results
|
IA |
|
NH |
|
SC |
Clinton |
50% |
|
45% |
|
72% |
Sanders |
44% |
|
52% |
|
25% |
O’Malley |
5% |
|
3% |
|
2% |
|
|
|
|
|
|
Sample size |
602 |
|
561 |
|
420 |
Margin of error |
+/- 7.6 |
|
+/- 6.9 |
|
+/- 8.7 |
Detailed results for each state: Iowa, New Hampshire, South Carolina
According to YouGov, Sanders has picked up increased support from African-American South Carolinians, while Clinton has increased her support from white voters in the same state, with a net gain for Clinton. There is little change in the results from the other states.
Comment on the results from me: In this election cycle, Sanders has usually gotten better name recognition and more support from participants in internet panel polls than traditional telephone polls. Clinton has gotten more support in polls like this one, restricted to previous participants in Democratic primaries and causes. In the case of the CBS/YouGov “Battleground” polls, these two types of sampling bias may sort of cancel out. (Their results for Iowa are very similar to Selzer’s most recent Iowa primary poll, after all.) But combining two different kinds of sampling bias, even though they favor different candidates, cannot be counted on yield better predictions of the primary and caucus results.
Take every poll with at least a grain of salt. Remember that sampling methodology matters. Remember that the primaries are still a ways off. Kudos to everyone who’s volunteering for their candidate and trying to make a difference. (Yes, even if you’re not volunteering for my candidate.)