Respondents in the poll also strongly favored a measure legalizing medical use of marijuana, which supporters are trying to put on the 2014 ballot as an amendment to the state Constitution.
Crist, a Democrat, led Scott by seven points in a head-to-head matchup, 47-40 percent.
In the same poll in June, Crist led 47-37 percent, a 10-point lead, and in March, Crist led by 50-34 percent.
Crist has has long been expected to run for governor, and appears to have received no boost from his formal announcement two weeks ago.
Crist did somewhat better among voters of his own party than Scott, and led narrowly among independents.
Democrats in the poll favored Crist 86-4 percent; Republicans favored Scott 80-11 percent; and independents split 44-41 percent in favor of Crist.
The poll also showed a large gender gap, with women favoring Crist 50-34 percent and men narrowly backing Scott 46-43 percent.
Scott’s job approval ratings declined slightly, however. He got a negative rating of 42 percent approval to 47 percent disapproval, compared to a 43–44 percent score June 18, his best ever in this poll. Respondents said they approved by 53–36 percent of the job Crist did as governor from 2006-10.
Scott beat Nan Rich, the other declared Democratic candidate, by 43-35 percent in the poll. It didn’t include a matchup with Sen. Bill Nelson, who has said he might enter the race if it appears the Democratic candidate is losing to Scott.
Respondents in the poll said they favor “allowing adults in Florida to legally use marijuana for medical purposes if their doctor prescribes it” by 82-16 percent. They also said by a margin of 48-46 percent they favor allowing adults to possess small amounts for personal use.
The medical marijuana measure would need a 60 percent vote in its favor to pass if it goes on the 2014 ballot, and Quinnipiac pollster Peter Brown said the measure is “overwhelmingly favored” to pass if it goes on the ballot.
Opponents plan to portray the measure, however, as opening the door to widespread recreational use of marijuana, while backers say it would provide only for tightly controlled medical use.
The Nov. 12–17 included 1,646 registered voters for an error margin of 2.4 percentage points. The error margin would be larger for results from portions of the sample such as gender or partisan breakdowns.