# Was the Democratic vote less efficient in 2018 than in 2016?

In 2016, Donald Trump won the election despite losing the popular vote by about 3 million votes. This is because the Republican vote was distributed efficiently in the Electoral College.

The map above shows Democratic advantages at state-level in Arizona, Iowa, Michigan, Pennsylvania, and Wisconsin in the 2018 House election. All of these states are targeted and/or have been won by Obama twice. This gave the House Democrats 295 "electoral votes".

The reason this is important is because I am curious about how the vote was distributed across states compared to the 2016 presidential election. I feel that the Democratic vote distribution was less efficient than 2016's election even though Democrats got a majority this time. I say this because it was an 8.6 point win as opposed to 2.1. When Obama pulled off a win almost as large in percentage in 2008, he won 365 instead of 295. (The win was 7.2 percentage points.) Is this an accurate assessment?

Note: I mean by "efficiency" how well winning by more popular votes translates to winning by more electoral votes. This is different than a classical example of House districts. This is aggregate at state level.

• It doesn't really make sense to look at House elections by state, because most states are not politically uniform. Compare e.g. NYC with upstate, or California urban areas with the rural northern parts. You also have a large bloc of independent voters who vote for individual candidates on their merits, rather than by partisanship. – jamesqf Jun 12 '20 at 15:38
• I believe this could look be the 2020 presidential map but with Iowa red and different margins. Electoral College is generally determined by popular vote in the state not congressional district, though it is sometimes (eg Obama won one in Nebraska in 2008, and Trump won one in Maine). – Number File Jun 12 '20 at 15:39

This is an interesting question - it depends by what metric you measure efficiency. I've looked at it by classifying a 'redundant' vote as a vote cast in excess of the amount required to win the state. For example, in 2016, the Democrats won 8,753,788 votes in California compared to the Republican's 4,483,810. This would result in 4,269,978 redundant votes. If the Republicans won the state, this means that there were 0 redundant votes.

Using this metric is slightly limited by the states which don't use the 'winner-takes-all' distribution of electors, these being Maine and Nebraska. Using this concept doesn't really make sense for these states without going down to district level, so I've excluded them from the totals.

Using this definition, and the state-level results for the 2016 Presidential and 2018 House elections, the number of redundant votes cast in 2016 was ‭10,934,112‬, while in 2018 there were ‭14,474,157‬. Therefore, despite winning an increased number of states and (a majority of) electoral votes, had 2018 been a presidential election, the Democrat vote was less efficient in 2018 than in 2016 by this metric.

Below is the CSV data should anyone want to test a different metric.

Alabama,1318255,729547,0,975737,678687,0
Arizona,1252401,1161167,0,1139552,1179193,39641
Arkansas,684872,380494,0,556339,312978,0
California,4483810,8753788,4269978,3973396,8010445,4037049
Connecticut,673215,897572,224357,520521,849341,328820
Delaware,185127,235603,50476,125384,227353,101969
Florida,4617886,4504975,0,3675417,3307228,0
Georgia,2089104,1877963,0,1987204,1814468,0
Hawaii,128847,266891,138044,87348,287921,200573
Idaho,409055,189765,0,367993,207303,0
Illinois,2146015,3090729,944714,1754449,2757540,1003091
Indiana,1557286,1033126,0,1247978,1000104,0
Iowa,800983,653669,0,612338,664676,52338
Kansas,671018,427005,0,563190,464380,0
Kentucky,1202971,628854,0,935304,612977,0
Louisiana,1178638,780154,0,835715,553184,0
Maine,335593,357735,22142,250119,343635,93516
Maryland,943169,1677928,734759,737906,1493047,755141
Massachusetts,1090893,1995196,904303,497953,1943597,1445644
Michigan,2279543,2268839,0,1853459,2175003,321544
Minnesota,1322951,1367716,44765,1125533,1420769,295236
Mississippi,700714,485131,0,471162,398770,0
Missouri,1594511,1071068,0,1330975,1027969,0
Montana,279240,177709,0,256661,233284,0
New Hampshire,345790,348526,2736,249685,310269,60584
New Jersey,1601933,2148278,546345,1198664,1856819,658155
New Mexico,319667,385234,65567,264701,404026,139325
New York,2819534,4556124,1736590,1859054,3990483,2131429
North Carolina,2362631,2189316,0,1846041,1771061,0
North Dakota,216794,93758,0,193568,114377,0
Ohio,2841005,2394164,0,2291333,2082684,0
Oklahoma,949136,420375,0,730531,428452,0
Oregon,782403,1002106,219703,702531,1061412,358881
Pennsylvania,2970733,2926441,0,2206260,2712665,506405
Rhode Island,180543,252525,71982,129838,242575,112737
South Carolina,1155389,855373,0,927494,758340,0
South Dakota,227721,117458,0,202695,121033,0
Tennessee,1522925,870695,0,1279655,846450,0
Texas,4685047,3877868,0,4135359,3852752,0
Utah,515231,310676,0,617307,374009,0
Vermont,95369,178573,83204,70705,188547,117842
Virginia,1769443,1981473,212030,1408701,1867061,458360
Washington,1221747,1742718,520971,1048712,1888593,839881
West Virginia,489371,188794,0,337146,234568,0
Wisconsin,1405284,1382536,0,1172964,1367492,194528
Wyoming,174419,55973,0,127963,59903,0
• That is what I was looking for. I knew what it meant, but I couldn't put a finger on it. – Number File Jun 12 '20 at 12:06
• Using a slightly different definition, where redundant votes are >50% instead of >Republican, I get totals of 5,467,060 redundant votes for 2016 and 7,237,086 for 2018. This is still the same percentage change, though (+32.38%), so I'm not sure there's a meaningful difference. – Bobson Jun 12 '20 at 17:15
• You should also consider counting all votes cast where the Republican won, since they are also wasted. – obscurans Jun 12 '20 at 20:52
• @Bobson: If you exclude spoilt, blank, and invalid ballots, and third parties generally (all of which form a relatively small proportion of the total vote), the >Republican number is equal to the margin of victory, and the >50% number is equal to half the margin of victory. So you should expect them to change by the same percentage, because one is (roughly) half the other. – Kevin Aug 9 '20 at 1:05