Will Romney Have Edge When Pollsters Move to Likely Voters?

[WI_DEM]

Most national pollsters are still using samples of registered voters rather than "likely" voters and some suggest that Mitt Romney will have an advantage when this change is made but Mark Blumenthal suggests it's too early.

"In almost every election dating back to 1980, the margins separating the top candidates in horse race polls shifted significantly after the party conventions. Only in 1996 did those margins remain roughly the same throughout the year. In other years, the shifts in voter preferences that occurred after the party conventions, shifts that have benefited both Democratic and Republican candidates, would have overwhelmed the relatively modest differences that earlier likely voter screens would have produced."

"In the end, if all pollsters applied likely voter screens right now, Romney's numbers would be slightly better, but there is a long way to go before any horse race poll should be considered an accurate forecast of the outcome."


http://politicalwire.com/archives/2012/07/18/will_romney_have_edge_when_pollsters_move_to_likely_voters.html

[Arkana]

1. They're all using LV screens from 2010, when liberal voters stayed home and angry white people

voted in droves.

Of course Romney's numbers are going to be better if you inflate the number of angry racists in the poll.

[SlimJimmy]

3. Actually, they are asking sample questions to extrapolate the likely voters.

"Did you vote in an election in the last four years?"
"Do you always vote?"
"Do you know where your polling place is?"
"Do you follow politics closely?"
"Have you given a lot of thought to the coming election?"

There are other questions that are crudely predictive of actual voting. Drawing again from that final week 2006 Pew Research poll, some examples include voters who report:

Having previously cast a vote in their precinct (65% of adults)
Always or nearly always voting (63%)
Knowing where people in their neighborhood vote (63%)
Following "what's going on in government and public affairs" most or some of the time (63%)
Giving some or a lot of thought to the upcoming election (50%)

Again, none of these are perfect predictors of turnout, and the percentage answering in the affirmative in each case still far exceeds the actual turnout of eligible adults in 2006. However, the 1999 Pew study confirmed that those who report greater intent to vote, a history of past voting, knowledge of voting procedures or high interest in or knowledge about politics a week or so before an election are more likely to vote than those who do not.

In the 1950s, Paul Perry, the legendary president of the Gallup Organization, combined seven questions much like these into an index. The greater respondents score on the index, the greater the the probability that they would vote He then made an educated guess about the likely level of voter turnout -- say 40% of adults -- and used the index to select the 40% of the sample that scored highest on the index. That procedure proved to produce more accurate election forecasts than alternatives they tested, and remains the basic concept still used in the Gallup likely voter model to this day.

http://www.huffingtonpost.com/2010/10/05/likely-voters-how-pollsters-choose-them_n_751560.html

[SlimJimmy]

2. I'm a big fan of likely voter polls. They always give a much clearer picture of

what is actually happening on the ground. In my opinion, using registered voters or adults in a poll is a colossal waste of time and money.

[Marzupialis]

4. It depends on whether Obama extends his RV lead

If things stay the same by say, October or November when all pollsters or almost all pollsters have switched to LV, then Romney will take the lead or at least be tied with Obama because LV historically changes the margin 4-5% toward the Republican.

However, I don't know if minorities register to a greater extent than whites late in the game (October or November). If that's the case, then Obama will extend his lead among RV in October-November, and this would serve as a buffer against the LV change. In summary, Obama needs to open a lead of at least 5% among RV in order to feel somewhat safe.