Was 2016 a Hillary Decline or Trump Surge? Examining Voting Trends
Overview: Today's post is going to be relatively short and simple, as I originally had nothing available and decided to do this at the last minute. The inspiration for this post was an Incredibles-derived meme on Twitter from several weeks ago regarding the 2016 presidential election. The meme highlights an ongoing debate about the 2016 presidential election, which is whether Trump's victory in Michigan (and by extension, Wisconsin and Pennsylvania) was a one-off occurrence or part of a larger trend, notating that "We [The Republicans] went from 45% in Michigan to 47% [between 2012 and 2016]," and that it "created a new path to victory." Seeing this meme inspired me to want to look into this case quantitatively, particularly to judge whether Trump's victory in those states was the result of a Trump surge or Clinton decline.
As a side note, the phenomena explored in this post has probably been looked over before, but I wanted to try it out for myself using my own methodology and put it in the current context. I did not look into any existing studies that examined the same exact thing before writing this, but again, I am also sure that others have probably done this already because of the simplicity of the methodology and because of the intense fascination with the 2016 election by political nerds like myself.
Comparing 2012 and 2016: This project consists of analyzing 2012 and 2016 battleground state presidential election data, both of which were taken from The American Presidency Project at the University of California, Santa Barbara. I narrowed down the data to 10 of the closest states in 2016, some of which were also close in 2012. These states include the six states that Barack Obama (D) won in 2012, but Trump (R) flipped in 2016 (Florida, Iowa, Michigan, Ohio, Pennsylvania, Wisconsin), another state Obama won in 2008, lost in 2012, and was won by Trump in 2016 (North Carolina), and one that neither Obama or Clinton won but moved into competitive territory in 2016 (Arizona). For this analysis, I did not include Maine's or Nebraska's individual districts, as I wanted to simply focus on statewide voting patterns.
Table 1: 2016 Close States Raw Vote Changes from 2012
Notes: Raw voting data came from the sources listed under "data sources,", but the party vote changes, net changes, and 2012/2016 winning margins by candidate were self-calculated by the author. Italicized states are those that voted for Trump, and states in plain text voted for Clinton.
Table 2: 2016 Close States % of Total Vote Changes from 2012
Notes: Raw percentages came from the sources listed under "data sources," but the party vote changes, net changes, and 2012/2016 winning margins by candidate were self-calculated by the author. Italicized states are those that voted for Trump, and states in plain text voted for Clinton.
Raw Votes: As can be seen in Table 1, the only major 2016 close state where Clinton improved in relative to Obama was Arizona, although such improvement was not enough to win that state. Otherwise, in most of the other close states, particularly in the Midwest, Rust Belt, and Northeast, the drop in Clinton's total votes relative to Obama's 2012 numbers was greater than Trump's gains (or losses) in those states relative to Romney's. The only major exception is Pennsylvania, where Clinton dropped by 63,000 votes from Obama, while Trump gained almost 300,000. Even in New Hampshire, Minnesota, and Maine (Clinton's 3 closest states), Clinton's vote decline relative to Obama was larger than Trump's gains in those states. Despite Trump not winning them, these results seem to show that the close results in these states were the result of Clinton declining relative to Obama, and not because of a significant Trump surge relative to Romney. In interpreting the Florida, Nevada, and North Carolina results, while both parties gained votes relative to 2012 (because all three states continue to rapidly gain population), Trump's gains far outstripped Clinton's, locking her out of Florida and North Carolina and keeping Nevada close.
Percentage of Total Vote: Referring back to the original meme linked above, the percentage of the total vote by party results in Table 2 tell a similar story to Table 1, in that Trump's victory was not accompanied by a significant surge ahead, but was because of a Clinton decline relative to each party's 2012 results. Notably, while Trump increased the Republican share of Michigan's votes by 2.8% between 2012 and 2016 (the "45 to 47% in the original meme), the Democrat's share of the state's vote declined by 7%. In all of the other close states except Arizona, the same pattern remained true, and in some cases, occurred even if the Republican's share of the vote also dropped. This was prominently seen in Minnesota, where the Republican share of the state's vote dropped by .1%, while the Democrat's share dropped by 6.3%. Even in Arizona, where Clinton did better than Obama, much of her growth there was not the result of a significant surge on her end, but a drop in the Republican percentage of the total vote (this may have to do with a significantly higher third-party vote, something that is another topic for another post). Even in Ohio and Iowa, where Trump did show significant gains relative to Romney, Clinton's decline still outpaced such gains, allowing Trump to win both.
Implications: The 2012-2016 presidential vote shift in the critical states has significant implications for this year's election, particularly that these results seem to back up the argument that Trump's 2016 victory may have been an unusual fluke resulting from a significant Clinton decline, and not a "Trump surge" in several critical states. Referring back to the original meme, it is true that Trump received 47% in Michigan versus Romney's 45% in 2012 ("2012" and "2016"). However, Clinton faced a 6.9% decline in vote share from Obama's 53%, mainly caused by a decline in total Democratic votes. This likely (or almost) doomed her chances of winning the critical Midwest, Rust Belt, and Northeast states. A similar story emerges in the Southeastern battlegrounds of Florida and North Carolina, as the total votes Clinton gained from Obama's 2012 performance were far outpaced by Trump's gains, thus locking her out of both states. Assuming all other factors held equal, if Clinton had simply matched Obama 2012's vote numbers, she would have won Iowa, Michigan, Pennsylvania (by a hair-thin 19,000 votes), and Wisconsin, and would have lost Ohio by a similar amount as winning Pennsylvania. Of course, this is without factoring in North Carolina, Florida, and Arizona, much less other similar Sun Belt states such as Georgia and Texas that could become competitive in this and the next few cycles (Moskowitz). Regardless, the results of this short quantitative analysis show that Trump's 2016 victory was not so much a surge of votes on his end, and was instead more likely the result of a massive Clinton vote decline from Obama in several key states.
As of the time of writing this post, polls are showing incumbent President Trump in a very precarious position, with Biden leading in most of the 2016 close states profiled above, and Trump falling behind in other normally Republican states such as Georgia and Texas (Jacobson). For a while, many pundits have speculated that unless Trump expands the Republican electoral college map beyond the "Rust Belt Three," Trump's path to re-election will have to rely on a similar strategy, and one that is more riskier this time around (Kondik). On the other hand, Joe Biden would simply have to replicate Obama's 2012 results in the "Rust Belt Three" to reach 270 votes, while winning Florida, Arizona, and others would provide a safety cushion (e.g. in case of faithless electors defecting, bringing Biden below 270) (Mosowitz). Recent reports have emerged that Republican insiders and strategists are becoming increasingly concerned that Trump's path to re-election is very narrow, particularly with his declining approval ratings and Biden's polling strength being far higher than Clinton's (Fedor).
Works Cited:
Fedor, Lauren. "Republicans warn of ‘grim’ outlook for Trump in Florida." Financial Times, 5 Jul. 2020, www.ft.com/content/1d81d237-8b76-48eb-9b7e-c3da6bdbabdf. Accessed 8 Jul. 2020.
Jacobson, Louis. "Biden Gains Ground Over Trump in Electoral College Outlook." US News and World Report, 18 Jun. 2020, www.usnews.com/news/elections/articles/2020-06-18/joe-biden-gains-ground-over-trump-in-electoral-college-outlook. Accessed 8 Jul. 2020.
Kondik, Kyle. "The Electoral College: Expanding the Map." Sabato's Crystal Ball, 19 Sept. 2019, centerforpolitics.org/crystalball/articles/the-electoral-college-expanding-the-map/. Accessed 8 Jul. 2020.
Moskowitz, Seth. "The 2016 Electoral College Map: A Template for 2020." Sabato's Crystal Ball, 22 Aug. 2019, centerforpolitics.org/crystalball/articles/the-2016-electoral-college-map-a-template-for-2020/. Accessed 8 Jul. 2020.
Data Sources:
"2012." The American Presidency Project, University of California, Santa Barbara, updated 16 Dec. 2016, www.presidency.ucsb.edu/statistics/elections/2012. Accessed 7 Jun. 2020.
"2016." The American Presidency Project, University of California, Santa Barbara, updated 16 Dec. 2016, www.presidency.ucsb.edu/statistics/elections/2016. Accessed 7 Jun. 2020.
