This afternoon, FiveThirtyEight is presenting forecasts of the Republican contests in Nevada, Arizona, Michigan and Ohio.
The polling in these states has been a little thinner than ideal — just one or two recent surveys in each state. So for the most part, the forecast is simply taking the most recent poll and translating the data into odds of victory for each candidate. The forecasts will become more reliable once additional surveys are released.
Still, it’s clear who has the edge in three of the four states: Mitt Romney.
In Nevada, which votes on Saturday, a poll released by the University of Las Vegas for The Las Vegas Review-Journal gives Mr. Romney a 20-point lead over Newt Gingrich. A 20-point lead with just a few days before a vote is normally quite safe, and the forecast model gives Mr. Romney a 99 percent chance of winning.
It should be noted that polls of the Republican caucuses in Nevada were extremely inaccurate in 2008. Caucuses in general are hard to poll. And Nevada has been a nightmare for pollsters in recent years. The state’s rapidly growing population; its unorthodox working hours, linked to the pre-eminence of the gambling industry; and the presence of some unique demographic groups — like Mormons — all present challenges.
In 2008, the polls vastly underestimated Mr. Romney’s standing — so saying that he is the overwhelming favorite there this year seems a safe assumption.
There is more time for events on the campaign trail to intervene before Arizona and Michigan vote on Feb. 28. The most recent polls there, however, give Mr. Romney big leads as well. The FiveThirtyEight model gives him a 91 percent chance of winning Arizona and an 89 percent chance of taking Michigan — although these estimates should be thought of as provisional until one or two more surveys in these states are released.
Ohio, on the other hand, which votes on Super Tuesday, March 6, looks as though it could be close. A Public Policy Polling survey there, completed last week, showed a competitive three-way race between Mr. Gingrich, Mr. Romney and Rick Santorum, with Mr. Gingrich holding a nominal one-point lead.
The survey was conducted before Mr. Romney’s win in Florida. Still, earlier polls of Ohio sometimes showed middling results for Mr. Romney — a Quinnipiac poll released in early December gave Mr. Gingrich a significant lead there, for instance.
Ohio could be an interesting test case: like Florida, it is a populous state with a number of media markets, potentially giving Mr. Romney’s campaign some advantages because of its superior resources. However, it is also a working-class state, a group with which Mr. Romney has sometimes struggled.
There are also caucuses in Maine, Minnesota and Colorado this month, as well as a “beauty contest” primary in Missouri that has no effect on delegate allocation. But there is little recent polling in these states, and surveys in low-turnout events like caucuses and beauty contest primaries can be inaccurate. The forecast model is specifically designed for states where the polling is reasonably robust. Still, we may release forecasts for these states if additional surveys come out.