By Jim Ellis
Sept. 20, 2016 — The Reuters/Ipsos “States of the Nation” project finds Donald Trump taking a surprising 243-242 electoral vote lead over Hillary Clinton, but are the individual state projections reliable? Many news sources covered the weekend story, yet it appears that the R/I numbers leave much to be desired. In fact, they show several states going for a candidate either for the first time in this election cycle, or in a manner that other data fails to substantiate.
Reuters/Ipsos is using a huge pool of almost 16,000 respondents, which allows them to segment results for most states. They are also casting several turnout models, and then calculating various Electoral College scenarios based upon the percentage of individual voter groups expected to participate in the November election.
The model forecasting the one-vote Trump advantage features an overall 60 percent turnout ratio among eligible voters, 43 percent for all minority groups, 59 percent African-American participation, and 69 percent from Anglo males. Even with his Electoral College edge, the Reuters/Ipsos results still find Trump trailing Clinton in the national popular vote by a 40-45 percent margin, however.
Questionable conclusions occur in several states. First, addressing the entities where Reuters/Ipsos did not have a large enough respondent sample to determine a trend, several have a clear and decided voting history. Adding Democratic Rhode Island and the District of Columbia’s electoral votes to Clinton’s total would increase her share to 249. For Trump, all other polling data supported by a long vote history would yield both Alaska and Wyoming to him. This would add another six votes to the Republican’s national total, thus leading the adjusted national split to a 249-249 tie.
Additional Reuters/Ipsos states are completely out of whack with other published survey research. One state that suggests their data lacks credibility is Vermont. This heavily liberal and reliably Democratic state is actually placed in the Trump column. Colorado and New Mexico, both of which could split their votes between the parties, have been consistently for Clinton throughout this election. Yet, the “States of the Nation” survey awards the pair to Trump. Adding the combined 17 electoral votes from the three states to Clinton’s total, where they belong, and removing them from Trump would adjust the national count to 266-232, in the Democrat’s favor.
The other state that appears to be at odds with current trends is Ohio (18 electoral votes). Polls from last week clearly point to Trump holding a lead, but the Reuters/Ipsos data puts the Buckeye State squarely in the Clinton column. If we further adjust to place Ohio on the Trump side of the political ledger, the national count would then spin the Republican back into the lead, 250-248. In the adjusted scenario, three states worth a combined 40 electoral votes (Maine, Michigan, Pennsylvania) are too close to call.
Obviously, the Reuters/Ipsos national data appears to be flawed, but even this statistical model suggests that the race is unquestionably close. Still, attempting to factor in the three remaining toss-up states under their projection model, adding a Trump win in Pennsylvania to his 250 electoral votes would take him to 270, the bare minimum needed for national victory. Failing to carry the Keystone State, a Michigan/Maine Trump combination would also equal exactly 270.
A more realistic fully projected Electoral College count, based upon all available data, could reasonably end with Clinton winning the Electoral College with 272 electoral votes as compared to Trump’s 266. This outcome would find Clinton winning all of the 2012 Obama states with the exceptions of Florida (29 electoral votes), Ohio (18), Iowa (6), Nevada (6), and the 2nd Congressional District of Maine (1). This configuration leaves him four votes short and would award a close national victory to Clinton.