Debates about what led to the defeat of Hillary Clinton, initially regarded as the favorite candidate in the 2016 presidential run, will be around for a while. Pollsters and scholars have been presenting different explanations for the unexpected Donald Trump’s victory ranging from bad campaigning strategy to the spiral of silence. I have my own pet hypotheses (some probably quite controversial) and hope to have a chance to test them in a future moment if I find data to do it.
Up to now, I have been playing with publicly available data. These data do not allow me to advance explanatory models but are enough to run descriptive analyses that end up being of some interest. A descriptive analysis may not answer the “why” question in detail but at least provides a picture of what happened and may facilitate to locate relevant pieces to be further investigated. I will be presenting vote counts as of 11 November 2016. As there are a few precincts to be added to the official vote count, figures may slightly change from now until December but fluctuations are going to be minimal.
The most immediately astonishing element (for me, at least) in Clinton’s defeat is how she underperformed in the popular vote compared to Obama in 2008 and 2012. Although the total number of popular votes received by Republican candidates from 2008 to 2016 — McCain, Romney, and Trump — remained mostly stable, Clinton managed to receive many millions of votes less than Obama.
This table reports the popular votes received by Democratic and Republican candidates in the three most recent U.S. presidential elections.
Popular vote for Democratic and Republican candidates in 2008, 2012, and 2016
|Obama 2008||69,498,516||McCain 2008||59,948,323|
|Obama 2012||65,915,795||Romney 2012||60,933,500|
|Clinton 2016||60,827,933||Trump 2016||60,261,913|
There are two important things to be noticed in this table. First, the Republican vote count has remained very stable from 2008 to 2016. Although Donald Trump received less votes than Mitt Romney, the difference is not substantial (less than 700,000 votes).
Second, and most important, Hillary Clinton received much less popular votes than Obama. Clinton got full 5 million less votes in 2016 than Obama received in 2012. If one compares Clinton’s numbers to Obama’s historical victory in 2008, the difference goes to 8.6 million.
Looking only at those numbers, my quick-and-dirty conclusion is as simple as this: Clinton lost because she was not able to mobilize Democratic voters to go out and vote for her.
This is, however, only part of the story. The U.S. elects its presidents using an indirect system via the electoral college and winning the popular vote matters less than its distribution across states. The important analysis here should focus on where Clinton won and where Trump won. [I will write about why the “Clinton won the popular vote” is not relevant in a future occasion.]
Compared to Obama’s state-level (50 states + DC) performance in 2012, Clinton received less votes in absolute numbers than Obama in forty six states; she was better voted than Obama ’12 only in five states: Florida, Georgia, Massachusetts, Nevada, and Texas. It is interesting to note that, although Clinton ’16 received approximately 248,000 more votes than Obama ’12 in Florida, she still lost the state because Trump overperformed Mitt Romney in this state by 442,000 votes. In a Clinton-Trump comparison, Trump bet Clinton by almost 120,000 votes.
Votes for Hillary Clinton in 2016 v. Votes for Barack Obama in 2012
|State||Clinton ’16||Obama ’12||Difference: Clinton ’16
– Obama ’12
|Did Obama win the state
|Did Hillary win the
state in 2016?
If one examines these numbers in terms of proportion of the state-level popular vote, Clinton received a larger vote share in Arizona, California, D.C., Georgia, Massachusetts, Texas, and Utah. Even in those states, however, the difference was above 2% only in Texas (2.02%) and Utah (3.05%), being less than 1% in Arizona, Georgia, and Massachusetts, and between 1% and 2% in California and D.C. Except for Texas and Utah, I do not think the difference is substantive enough to support the claim that Clinton overperformed Obama in those states.
This graph shows that Clinton’s percentage losses were way more substantial than her tiny proportional gains.
As shown in the first table, when Clinton’s performance is compared to Obama’s in 2008, the difference is even more expressive. Now the difference was 8.6 million votes. If look for states where Clinton bet Obama, we find that she received more individual votes than him mostly in the same states as in the previous 2012-2016 comparison. The only difference is that North Carolina is included in the list rather than Georgia.
For the sake of completeness, here is a list of states where Clinton received more popular votes in 2016 compared to Obama in 2008: D.C., Florida, Massachusetts, Nevada, North Carolina, and Texas.
Votes for Hillary Clinton in 2016 v. Votes for Barack Obama in 2008
|State||Clinton ’16||Obama ’08||Difference: Clinton ’16
– Obama ’08
|Did Obama win the state
|Did Hillary win the
state in 2016?
In terms of state-level vote share, the Obama 2008-Clinton 2016 is even wider. In eight states, Clinton underperformed Obama in 2008 by a margem of at least 7%. In Iowa, North Dakota, and West Virginia the difference is larger than 9%.
Back to the Clinton ’16-Obama ’12 comparison, Hillary Clinton failed to capture any additional state in comparison to where Obama won in 2012 and, in addition, lost six states that had gone Democratic in the previous election: Florida, Iowa, Michigan, Ohio, Pennsylvania, and Wisconsin. Losing these six states did cost 99 electoral college votes.
States lost by Hillary Clinton in 2016, compared to Obama in 2012
|State||Did Obama win the state in ’12?||Did Hillary win the state in ’16?||Did Hillary lose the state?|
The results presented up to now has shown that (1) although the Republican popular vote has remained stable since 2008, the Democratic one has dramatically decreased; (2) as a result, Clinton ’16 received 5 million votes less than Obama ’12; (3) this vote decline led to a Clinton’s loss of six states previously captured by Obama, what have a negative impact on her performance in the electoral college that ended up costing the presidency.
So far so good. A few questions remain to be answered in this analysis: Why did Clinton lose those states? Was Clinton’s underperformance hard enough to cost her the presidency? Or did Trump receive a spectacularly high number of popular votes in those places? As we see above, Clinton received million of votes less than Obama. Had her been able to warrant a similar number of popular votes in those specific states, could she win the election?
The best way to answer these questions is to compare how Clinton (and Trump) performed in those six states in 2016 in comparison to Obama (and Romney) in 2012.
Comparing Clinton (Trump) in 2016 to Obama (Romney) in 2012
|State||Hillary ’16||Obama ’12||Hillary – Obama||Trump ’16||Romney ’12||Trump – Romney||Trump – Hillary||Would Obama beat Trump?|
In five out of the six states Clinton lost (compared to Obama ’12), Clinton received less votes than Obama. The ‘vote deficit’ ranges from 145,000 in Pennsylvania to 510,000 (!) in Ohio.
Trump overperformed Romney in those five Rust Belt states. However, it is crucial to notice that Donald Trump bet Hillary Clinton in those states by lesser votes than the difference between Clinton in Obama in those states. For instance, Trump received 148,000 more votes than Clinton in Iowa at the same time Clinton got 171,000 less votes than Obama ’12.
In a state-by-state comparison, votes for the Republican candidate grew at a smaller pace than the Democratic decline in Iowa, Michigan, Ohio, and Wisconsin. The latter is probably the best example: Trump received only 1,500 more votes than Romney there but Clinton lost 240,000 there. Pennsylvania is the exception, being the only of those Rust Belt states where Republicans grew more than the Democratic retract.
Florida is the only state lost by Hillary Clinton where she was able to receive more support than Obama by approx. 250,000 votes margin. However, Trump was able to capture 442 thousand additional votes in Florida, overshadowing Clinton’s gains. If, in 2012, Obama bet Romney in Florida by a 74,000 votes margin, Trump bet Clinton in 2016 by the impressive advantage of 120,000 votes. In comparison to 2012, approx. 700,000 additional votes were casted in Florida in the 2016 competition, two-thirds of them going to the GOP. Donald Trump performance in Florida was strong enough to beat Obama ’12 in the state.
In a hypothetical scenario in which Clinton ’16 were able to warrant the same number of votes as Obama ’12, the Democratic candidate would still lose Florida for Trump but would be able to capture all the other five states where she lost. Those five Rust Belt states would give 70 additional electoral college to Hillary Clinton, more than enough to secure her rise to the White House.
My conclusion based on these state-level popular vote data is: The main cause of Clinton’s defeat / Trump’s victory is the considerable Democratic underperformance in 2016 compared to the 2012 election. Hillary Clinton was not able to turn any single 2012 red state into a 2016 blue state, what would be not a major problem everything else had been held constant. However, the silver bullet on her presidential aspirations was a dramatic decline in the Democratic vote in five Rust Belt states that turned those 2012 blue states in 2016 red states. Had Hillary Clinton been able to secure a similar vote count as Obama ’12 in those states, she would had enjoyed an easy ride back to D.C. as Madam President even if losing Florida. As Iowa, Michigan, Ohio, Pennsylvania, and Wisconsin turned their back to the Democratic candidates, they paved the road for Donald Trump to become the forty-fifty president of the United States.
What remains to be explained is why the Democratic popular vote in key states experienced a fast decline in four year. This is certainly going to be a hot debate as several alternative hypotheses will be presented. It may be the weak economic recovery in the Rust Belt during Obama’s administration. It may be a rejection of Hillary Clinton’s political platform. It may be Donald Trump’s rhetoric sounding appealing to voters in those states. It may be something else. To assess these and other hypotheses, micro-level data is needed to tell us what conditioned people’s voting decision as well as the motivations to reject the continuation of the Democratic administration.