In 2010, the Republicans took over control of the state machinery in a number of blue and purple states, including Ohio, Pennsylvania and Michigan. One of the first things some of them did was enact tough voter-ID laws, in the hope it would keep enough Democrats from voting that the Republican nominee would carry the state. For a variety a reasons, mostly court challenges, it didn't work. Now they have a new plan.
The idea is stop awarding the state's electoral votes on a winner-take-all basis, but to split it per congressional district, as Nebraska and Maine do. There is no question that such a move is legal as the Constitution gives each state the power to determine how its electoral votes are cast. In Virginia, for example, although President Obama carried the state by a healthy margin, he won only four of the state's 11 congressional districts. So under this plan, Obama would have gotten four electoral votes and Romney seven.
In Maine and Nebraska, the statewide winner gets the other two electoral votes, but the Virginia plan would award the extra two electoral votes to the candidate who won the most districts. Thus under the proposed plan, Obama would have gotten four electoral votes to Romney's nine, even though Obama carried the state. Similar results hold for Ohio, Michigan, Pennsylvania, and other states. If all these states switched to a district-based scheme, it would become extremely difficult for Democrats to win the White House in the future, even if they got far more total votes and won far more states than the Republicans.
The Democrats' problem--which the Republicans want to exploit to the hilt--is that Democrats tend to be packed into a small number of highly gerrymandered urban congressional districts with Republicans having small majorities in many rural districts. If all congressional districts were drawn fairly, this problem would not exist, but most districts have been very carefully drawn for maximum partisan advantage, and in 2010 the Republicans did the drawing.
Another issue is that this change in the way electoral votes are allocated is only being contemplated in blue states. If the idea were also applied to big red states, like Texas, Georgia, and North Carolina, the Democrats would pick up quite a few electoral votes. But the Republican-controlled state legislatures there have no interest in such a plan at all as it is simply a partisan gimmick to help Republicans.
Democrats don't really have any way to fight back now, but there are two developments on the horizon that could possibly help them in the long run. First is the National Popular Vote Interstate Compact. The idea here is that a state allocates all of its electoral votes to the candidate who won the national popular vote, irrespective of who won the state. If states with 270 electoral votes sign on, this de facto eliminates the electoral college since then the candidate with the most votes nationwide wins the electoral college automatically. Currently eight states, with a combined total of 132 electoral votes have signed up. If the Republicans manage to change the rules in several blue states, this may motivate Democrats in other states to join the compact, although currently there are not enough states the Democrats control to reach 270, although that could change in 2016.
The second factor is that the Republicans managed to take over state governments in 2010 because it was a midterm election year and Democrats don't bother to vote in midterm elections. In contrast, 2020 is a presidential election year, when Democratic turnout will be high. It is certainly possible that the Democrats will make a major push to win state legislatures that year and change the laws concerning the electoral college, for example, by joining the Interstate Compact.
In any event, Republicans are counting on the fact that they can probably change the rules to their advantage where it helps them this year and not suffer in public opinion for it. In principle, there is nothing wrong with allocating votes by congressional district--if it applies to the entire country, not just blue states. But that is certainly not going to happen any time soon.
This sure-to-be-controversial hypothesis is being floated by political scientists at Vanderbilt and Princeton Universities. In short, they say that robopolls taken after a poll using live interviewers, does pretty well but one taken without a previous human poll for "guidance" does poorly. Here is the full paper.
In effect, these professors are suggesting that the robopollsters are manipulating their numbers to make them agree with the "real" pollsters, the ones who have live human beings doing the calling. Our view is completely different. The robopollsters don't really have any vested interest in publishing numbers that look like the live polls. They have a vested interest in getting results that match the election results. If their actual data show that the live polls are way off, they would be crazy to "copy" those numbers and get the results wrong, too.
Possibly the two professors have this somewhat outdated view that live polls give good results. As anyone who was paying attention last year knows, Gallup, the gold standard of live pollsters, did an awful job, predicting a Romney landslide right up until the moment he conceded, whereas some of the robopollsters, such as PPP and SurveyUSA hit the nail on the head.
We analyzed the pollsters on Nov. 23, 2012. One thing that comes out of that study is that there is quite a bit of variance among each category. Roboposters PPP and SurveyUSA had very slight (Republican) biases, whereas robopollsters Rasmussen and subsidiary Pulse had biases of more than double that. ARG, the University of New Hampshire, and Quinnipiac University also had substantial Republican biases as well as large absolute errors, despite using live interviewers.
The paper used the Republican primaries as the data source whereas our analysis used the general election results. Nevertheless, lumping all the robopollsters into one basket and all the live-interviewer pollsters in another doesn't seem like a good idea, given the large spread in each category.