1. Polls and Pundits
People don’t change, populations change. I know this sounds deterministic and seems to preclude the possibility of free will but, as far as voting habits are concerned, I think it is fundamentally true as I will argue in this article. It is a useful mantra I repeat to myself frequently during every election cycle to help me make my way through the thickets and swamps of misinformation, misunderstood information, information-free opinions and analysis, and generally depressing election coverage. The implication of this simple phrase is that none of the stuff I just mentioned matters, and it is indeed my hypothesis that most of what happens during an election cycle has another purpose that is essentially irrelevant to the actual events of election day.
As the countdown to election day approaches, patterns that I have observed in previous election cycles reappear: the frenzied horse race mentality of the media, the optimistic predictions of victory in the unlikeliest of states by surrogates of both candidates, the insistence on trying to parse daily actions of the candidates and the presumed effect on the critical demographic group required for the candidate’s success, the perennial discussion of the potential for one group of voters or another to swing the election in a new direction, and so on. Each day brings new exciting analysis that follows the same script each election cycle, even though the narrative arcs of the candidates superficially seem so different from those of previous election cycles. Will Trump’s latest outrageous statement finally cause voters to come to their senses? Is Hillary Clinton’s penchant for furtiveness going to cost her the election as she has a trust “issue”? What do the most recent polls show? Who is up and who is down? What does Nate Silver think about all this?
I understand that the news networks need a story line every day to keep the viewers hooked and their advertisers happy. The candidates also have a strong interest in staying in the public eye. Indeed, as we have learned from this election, some candidates will do anything to stay in the public eye with a clear preference for any publicity, good or bad. I also understand the tension and excitement that the uncertainty of the potential outcome brings, which leads to a compulsive trawling of the daily information stream for good news for your candidate and that piece of bad news for the opposition that will finally sink their candidacy. There is the corresponding corollary that every piece of bad news for your candidate of choice leads to thoughts of the coming apocalypse when the opposition wins. I suffer from this problem myself: How long until World War III starts if Trump takes office is the latest iteration of this long-playing record in my version of this nightmare. There is also good money to be made, reputations to be enhanced or diminished by being right or wrong in analyzing the potential outcome, and free air time for politicians and pundits of all stripes with an eye for career advancement. Election season is big business. The rise of the betting market mentality has served only to aggravate the horse race story line and to increase the tension now that the state of the election can be reduced to a single number, the percentage likelihood that one candidate will win the election.
I have observed with interest every election since 1984, and it has been my observation that in all these recent elections the story arc is remarkably similar: The incumbent or the scion of the incumbent is in a commanding position before the election begins as the opposition is weak and divided as exemplified by the intranecine squabbling during the primary. Then arises one great challenger, Candidate A, like a Phoenix from the ashes, strong and proud and ready to fight after battling bravely through the rough and tumble of the primary. Candidate B, especially if this candidate is the incumbent or of the incumbent party, is suddenly vulnerable, indecisive, distracted, untested in battle or weary and past his prime, dogged by the paper trail of a long political career. But Candidate A stumbles out of the starting blocks while Candidate B has suddenly become remarkably rejuvenated and is now manning the ramparts to bravely fend off the malicious attacks of Candidate A, who after all has quite a few Achilles Heels which were not apparent a few weeks ago. Then Candidate A is nominated at the party convention and has been given an elixir that makes him/her (well, always a him until this election) strong and invincible again while Candidate B has been dogged by yet another poorly timed international event or economic downturn or scandal that shows he or she clearly cannot handle the pressure of the task at hand. However Candidate B has a phenomenal convention, reconnects with the disaffected base and is back in the saddle, surely in too strong a position to be defeated now that the wind is at his or her back.
The summer breezes have died down by Labor Day and suddenly the race is, who could have guessed, neck and neck, too close to call, a nail-biter, as tight as it has been all year. Who could have known that Candidate A had such reserves of strength and that the base, which had been coy about its support have clearly been won over. The race is on now, inexorably tightening one day, then one or the other candidate seems to be pulling ahead, only to be pulled back and passed by the other candidate. Will Florida decide the election, or Ohio as usual, or will there be a wild card, perhaps Nevada, or North Carolina, or is this the year the red and blue map gets upended by this “new breed” of politician? Will there be an ‘October Surprise’?
Election Day arrives and the results are in doubt until late in the evening, but the candidate who was meant to have won all those months ago seems to have miraculously prevailed despite the head winds and the valiant efforts of the vanquished rival. Now begins the analysis of the result, the need to find the one single factor that propelled the candidate to victory when all was seemingly lost. Demographic subgroups of all types are brought forth as the linchpin of the election, a new group is crowned as the critical group required to achieve victory, and these voters are assiduously cultivated for four years until the cycle begins again.
2. Demography: The Only Constant is Change
One of my interests over the years has been to analyze election results mainly to try to understand long-term patterns. For example, when I moved to California in the 1980s the state had a reputation as a reliably Republican state; indeed the sitting President at the time, Ronald Reagan, was a former Governor of California. President Richard Nixon, who had resigned a mere half-dozen years before Reagan took office, was also from California, while the Governor of California for much of the 1980s and 1990s was also a Republican. Yet California today is one of the most reliably Democratic states in the country. Clinton will trounce Trump in California, and the race for the open Senate seat is between two Democrats as a result of the open primary system recently adopted in the state.
I studied the election results and determined that the demographic change that swept Californian Republicans out of office was predictable: first, the cities were strongly Democratic and were the engines of population growth. “Bullets of Blue” is how I think of these areas because on a map one sees large swathes of red covered with small dots of blue. Over time, as the populations in these “bullets” moved out to the suburbs, the blue started to spread and become larger in area. There was a critical moment when the balance between the densely populated blue areas and the red areas was contested, but the blue eventually triumphed and slowly continues to take over the landscape. There are still large patches of red on the map, but these are predominantly rural areas that are lightly populated.
I have subsequently noticed this phenomenon applies to many states and that the balance of power in most states is determined by the size of the “blue bullets” and the rate at which they are growing. States with no bullets, small bullets, or slow-growing, or even shrinking bullets trend Republican; states with many and/or fast growing bullets trend Democratic. California had two blue bullets at first, San Francisco and Los Angeles. As these cities grew, the suburbs became more and more blue, and eventually the sheer size of the electorate in these two cities was enough to make the state competitive for Democratic candidates. Then traditionally red areas such as San Diego and Orange County became progressively more blue, until the Republicans were left fighting each election to hold what they had. Currently, the Democratic party holds all the major offices, has a greater than 2-1 margin in both branches of state legislature, has both Senate seats, and a 39-14 margin in Congressional representation.
A glance at a national map showing the results by party makes it seem that the Republican party has overwhelming control of most of the country. This might be true from the perspective of physical territory, but in terms of population, the story is quite different. Unsurprisingly, more dense areas of the country appear to be smaller on the map despite the fact that they often hold far more people. For instance, the borough of Manhattan has a population of 1,644,518 people in less than 23 square miles, according to estimates by the United States Census for 2015. This is slightly more people than are found in the states of Montana and Wyoming combined, even though the land area of the two states together is 242,639 square miles, more than 10,000 times as large as Manhattan. Unsurprisingly, the tiny blue dots are lost in a sea of red and often result in the misleading conclusion that America is primarily “red” and the unfounded corollary that the vast red areas represent “real America.”
This is the source to my mind of most of the rancor that has emerged in this particular election, as older, less well-educated populations slowly are eclipsed in size and increasingly in power and influence by the younger, more educated, predominantly urban, and increasingly “non-white” Americans who make up a larger share of the population at each election. The arguments sound reasonable, that America used to be a better place that is being hijacked by lazy people in urban slums and “foreigners” who are changing the culture, and that clearly, as anyone can see from a map, malign secret forces are somehow ‘stealing’ the election from the vast majority in favor of a small minority. This argument is a modern resuscitation of an old version of racist and xenophobic arguments that have been used in countless elections to demonize those who came through the same door at a later date than the ‘real American’ making the argument. One of the few constants in American history is change, and change is painful for those used to being in the driver’s seat, and who may be unprepared for a modern economy dominated not by manual labor but by those in occupations such as finance or biotechnology.
So it seems reasonable to argue that there will be a surge of nativist support for Donald Trump from normally reliable Democratic working-class voters and that the absence of the unique circumstances that propelled Barack Obama to the White House will depress turnout among the minority groups whose support helped him to victory. Indeed this seems to be the predominant argument being promulgated by those who see Trump as a likely victor. Indeed, comparisons to the ‘Brexit’ vote in the United Kingdom are used to argue that there is a secret cache of disgruntled voters who are itching to turn up on election day and decide the election in Donald Trump’s favor. It is a plausible sounding narrative. I also believe it is a false narrative.
Which leads me to the main purpose of this article, to analyze the upcoming election from a completely different perspective, free from the distraction of prediction markets and breathless coverage of breaking news. I propose that the vast majority of people do not change in their voting habits, that they vote for what to them is the most accurate reflection of their ideology and that their ideology does not substantially change over time. I do not mean to say that people can’t change, but rather that they don’t change to such a degree that they substitute black for white or liberal for conservative. It is my contention that what actually changes slowly from election to election are the parties as they seek to find that perfect combination of principles and image to appeal to a broad enough coalition of voters to achieve victory. Insofar as people are born and die, as the old are displaced by the young, as certain demographic groups slowly shrink while others slowly increase, the makeup of these coalitions of voters inexorably changes, and it is the parties that shift their allegiances to accommodate the widest philosophical arc permitted to keep the often fractious pieces together. In short, I will argue that the results of every election hinge not on convincing voters to switch from voting Democratic to voting Republican but rather on shoring up the voter base by making sure that voters show up on election day. I will go further and state that election results, in the absence of traumatic events such as a war, are essentially predictable. The uncertainty is merely in places where the “blue bullets” and the “red sea” are finely balanced.
An implication of this hypothesis is that polls play a much less important role than they are often accorded. Insofar as a grand summary of polls offers a glimpse of the evolution of the demographic makeup of the electorate they have a purpose. But in general it is my opinion that the use of polling analysis to ascertain the state of the race is fundamentally flawed and that, to the extent that it augments the already misleading notion of a ‘horse race’ in an election, actually is deleterious to the intended goal of bringing clarity to election analysis. Thus, the constantly shifting dynamic of the 538 forecast graphs is a reflection not of a shift in the likelihood of a certain electoral outcome, but rather reflects the shifting dynamic of the perceived state of the race and is essentially an indicator of the ‘mood’ of the election industry. The results of an election, within reasonable parameters that, in only a few cases, drift into the realm of uncertainty of the outcome, are essentially unchanged. In most cases it is possible to predict the outcome of each state, even each county, to within a small percentage margin of the final outcome.
3. An Example: Boston
As this project is ostensibly about Boston I will offer Boston as an example. Based upon my analysis of the results of previous elections as well as my analysis of the demographic changes, I estimate that around 260,000 people will cast a ballot in the upcoming election, give or take 13,000 voters (about +/- 5%). Of these voters, about 78% will cast a vote for Hillary Clinton, giving her about 200,000 votes. If turnout is higher than expected, owing to good weather for example, Clinton might be able to achieve a higher total, perhaps 80%, or up to 210,000 votes. Donald Trump will receive slightly less than 20% of the vote, around 50,000 votes, perhaps as few as 45,000 votes. Despite the suggestion that the third-party candidates will pull a large number of votes, I anticipate the effect to be small, with Gary Johnson, Jill Stein, and any others earning at most 3-4% of the total vote, with 10,000 votes being the upper limit.
How can I be so confident of the results? Because the demographic makeup of Boston has not changed substantially from four years ago, because the number and distribution of registered voters has not changed much from the last election, and because there is no underlying change in the dynamic of the election compared to the previous election, despite attempts by the media to imply that Trump is a substantially different candidate from the typical Republican. Trump is the Republican candidate, about 20% of the voters in Boston vote Republican, although the number and makeup of this group is slowly changing (a future article), therefore Trump will get 20% of the vote. A corollary of this type of analysis is that there is no such thing as an undecided voter in the traditional sense: the only thing undecided about any voter is whether they vote or not, and even that is largely predictable. So-called ‘independents’ are independent only in that they do not wish to belong to a party, but it does not mean they do not have a party they support. I am a so-called ‘unenrolled’ voter, but I have yet to cast a ballot for a Republican candidate in 35 years of voting and do not see myself moving in that direction, unless the party moves in my direction, which currently seems very unlikely.
The rest of Massachusetts outside of Boston and Cambridge and other liberal cities sprinkled around the state is, on the whole, less ‘blue’. However, the balance of strongly ‘blue’ population centers combined with ‘pink’ suburbs and slightly more ‘red’ rural areas is such that, in a general election, turnout in Boston generally decides an election. In off-year elections, voter turnout is everywhere depressed, but is particularly low in Boston and other places with high student populations and/or non-white populations, and the balance between Republican and Democratic voters is much finer. The results of a presidential election in Massachusetts are very much not in doubt.
Things become more complicated in states where there is a finer balance between the ‘blue bullets’ and the ‘red sea’. Some states are slowly moving in the direction of California, where the blue is spreading slowly out from the blue bullets and the tide of red sea is ebbing: Texas and Georgia are two states that remind me of the state of California in the 1980s. Florida is a state on the brink of shifting in a permanent direction but as yet is still in flux. Colorado and Virginia are two states where the transition has occurred, and these states have reached the point where the blue is beginning to predominate, at least in Presidential elections. Other states are moving in the other direction: Ohio has seen slow declines in the formerly robust urban centers of Cleveland and Cincinnati, which has eroded the core of the blue areas, while the population is increasing slowly and aging; thus the state retains its status as a bellwether state more out of a lack of dynamism than because it accurately reflects the dynamic of the country as a whole. Smaller, more rural, slower growing, or regions with fewer minorities or fewer people with college degrees, slowly become more red at each election: Idaho, Arkansas, Oklahoma, Kentucky, even Iowa, once a more reliably blue state. On the whole, states that are growing rapidly are becoming more ‘blue’ while states that are aging and/or are more white are more reliably Republican at each election, and states where there is a little of each have become the “battleground” states. And, in this election, Pennsylvania has become the state to watch for exactly these reasons.
4. The State of the Election: Pennsylvania?
Political pspehologists, including the FiveThirtyEight.com website, which was created by Nate Silver, have proposed that Pennsylvania is in play for Trump. Is this true? Why it matters will be clear through an analysis of the electoral votes at stake in the upcoming election.
Result of the 2012 Presidential Election:
Obama 332 Electoral Votes, Romney 206 Electoral Votes.
270 electoral votes are required to win an election (269 tie throws it into the House where Republicans will win, so technically 269 votes is sufficient, but that is too complicated for this analysis). Therefore Trump has to take 64 electoral votes from the Democrats to win in 2016, assuming he holds all the states won by Romney, an assumption that is disputable but for the moment one I will use to advance the argument.
Some states are just not in the realm of possibility as the margins are so large that a fundamental sudden and unlikely demographic shift would be required to change the results from previous elections CA (55), D.C. (3), MA (11), MD(10), VT (3), IL (20), RI (4), NY (29), HI (4), DE (3). These total 142 electoral votes(EV). So in theory there are 200 EV out there for Trump. However, many states are not truly competitive either as they consistently vote for the democratic candidate by sizable margins and the type of voters there do not seem likely, based on demography and past results, to respond to the Trump message in a way that would alter the outcome in the state compared to previous recent elections: WA (12), OR (7), CT (7), NM (5), adding 31 more electoral votes to the Democratic column, for a total of 173 EV.
This leaves 159 EV in play. Let’s look at these states to assess the plausibility of a change from supporting the Democratic nominee to supporting the Republican nominee.
New Jersey (14 EV)
Voted Democratic every Presidential Election since 1992.
Increasing margins each cycle i.e. Kerry 52.92%, Obama 57.14%, Obama#2 58.38%.
The percentage of Non-Hispanic Whites has declined from 74.0% in 1990 to 59.3% by 2010 and is estimated at 56.2% in 2015. Almost no chance of switching, even if many white voters switch would require a 9 point swing from 2008.
Democratic Total 187. Remaining States (12) 145 EV.
Five States Won by Democrat since at least 1992: MN (10), WI (10), MI (16), PA (20), ME (4).
Total (60 EV). Last time state was won by Republican: MN (1972), WI (1984), MI (1988), PA (1988), ME (1988)
Remaining States: CO (9), NV (6), VA (13), FL (29), NH (4), OH (18), IA (6) All D in 2008, 2012—Total (85 EV)
Another way of looking at it is to ask what would it take for Trump to win? As I said before, if he is to win, he needs to hold the 206 electoral votes from 2012. Then he needs 64 more electoral votes. The most efficient route is to win the biggest states from the above list: FL, PA, OH makes 67 votes, which would be enough for 273 votes. If he loses one of these three he needs to make it up from the other nine.
Scenario #1: Assume he loses the smallest state, Ohio. This would leave him at 255 EV, needing 15 EV to win. He could still win by winning Michigan (16) for 271 EV. If he fails to win OH or MI, he would need to win two more states: Virginia is the largest at 13, for 268, requiring any other state in order to reach 270.
Scenario #2: Trump loses Pennsylvania. The math gets trickier. the 206 starting point plus FL and OH bring the total EV to 253. Michigan provides 16 EV which would produce a 269 tie, thus throwing the election into the House of Representatives, leading to a likely Republican victory. Failure to win Michigan would leave Trump needing to win two states totaling 16 or more EV. The shortest path to victory leads through Virginia (13) plus any other state: NH or ME for example provide 4 EV each for 17, leading to 270.
Scenario #3: Trump loses Florida. Now the number of routes to 270 drops dramatically. Assuming Trump wins the other 2 big states of PA and OH, he would have 244 EV, needing 26 for 270. Michigan (16) plus any state with 10 or more EV (VA, WI, MN) would make 270, or Michigan plus any two states where the EV total is 10 (all combos except NH+ME). In other words, 4-5 states won by Obama in 2008 and 2012.
What if Trump loses 2 of the 3 largest states? the least deleterious would be winning Florida and losing Ohio and Pennsylvania. In this scenario, Trump has 235 EV, thus requiring 35 EV from the remaining 9 states. The easiest route is through Michigan (16), then two of MN, WI, CO and VA . Thus this scenario requires a minimum of FL, MI and at least two more states, or FL plus at least 4 more states if MI is not successfully pulled into the Republican orbit.
What if Trump fails to win FL or PA? Here he requires 64 EV from states with 18 or fewer EV, so at least 5 will be required: Ohio and Michigan provide 34 for 240 EV but 30 are still required so any combination of three of VA, MN, WI, and CO are required for 270 (although MN, WI, CO results in a 269 tie as above). Alternatively 2 of the above 4 states plus two of the 4 smaller states. In other words, at least 5 states and possibly 6 or more of the states putatively “in play.”
If Trump fails to win FL, PA, and MI, then the path becomes very narrow indeed. Ohio is absolutely necessary for victory in any form. Even with Ohio, victory in almost every remaining state is essential for 270 EV. For instance, losing Virginia would require Trump to run the table with the remaining seven states (although there are two 269 tie scenarios). On the other hand, losing Minnesota requires Victory in all of WI, CO, and VA plus three of the remaining four states.
Unsurprisingly, Losing FL, PA, OH requires Trump to win Michigan and win almost all the remaining states (all but one at a minimum), and losing the four big states of FL, PA, OH, and MI is the death-blow from which no victories in other states will help.
What if Clinton takes a state won in 2012 by Mitt Romney. North Carolina for example, which Obama won in 2008? This truly complicates the Trump path to victory. On the one hand, if Trump were to take FL, OH, and PA, there would be multiple paths for victory for the Republican Party. However, if PA remained Democratic as well as two of the bigger states like MI or VA, then FL and OH are required for Trump to win.
As I mentioned above, Colorado and Virginia seem to be likely to become ever stronger firmaments of the democratic constellation. Minnesota, Wisconsin, and Michigan also seem unlikely from the perspective of previous results and electoral demographic shifts to suddenly move into the red camp. In other words, if one of these three states shifts to the Republican column, the election has likely already been lost. This leaves the democratic candidate with 245 electoral votes, with 25 required for victory and only 87 left to contest. Of these 87, Maine splits 2 of their 4 electoral votes by Congressional district. One district is very likely to be Democratic at the least so, Clinton will likely win at least one of the four electoral votes in Maine, bringing her total to 246, with 86 votes to fight for from 7 states: the 67 already discussed from PA (20), FL, (29) and OH (18), plus NH (4), IA (6) NV (6), and ME (3)
Thus it is clear that any key to victory for the Republican party requires an effort to win the 67 electoral votes of Florida, Ohio, and Pennsylvania. Florida and Ohio have indeed earned reputations as battleground states, and I have my doubts about the viability of Trump in Florida especially. However, Pennsylvania has been solidly in control of the Democratic party since 1992, when Bill Clinton defeated the incumbent George H.W. Bush with some help from Ross Perot. The presence of hidden Trump supporters is meant to put this state in play, made up of the aging, mostly white population of disgruntled ‘Reagan Democrats’ in formerly industrial areas like Scranton and the counties surrounding Pittsburgh in Western Pennsylvania. On the other hand there is the growing strength of the Democratic Party in the Philadelphia suburbs. So, is there any truth to the rumors that Pennsylvania is in play? Let’s look in detail at the demographics, the election history, and the current registration data to paint a picture of the likely electorate and the potential outcome of this race. If, as I will argue, Pennsylvania seems highly unlikely to shift to the Republican column, then the Trump campaign seems doomed to failure, regardless of the results in Florida or Ohio. Should the Democrats hold Pennsylvania, while not absolutely essential for a Clinton victory, it makes the election impossible for Trump to win unless he captures virtually every remaining electoral vote in the remaining 6 states.
5. Pennsylvania Election 2016 Analysis.
First a bit of data:
Electoral Votes 20.
Population 2015 estimate: 12,802,503 approximately 100,000 more than 2010.
2010 Population (US Census): 12,702,379
of which: 9,910,224 18+ years, therefore 2,792,155 under 18.
White Non-Hispanic 10,094,652 (79.5%)
Black Non-Hispanic 1,327,091 (10.4%)
Hispanic (any race) 719,660 (5.7%)
Asian alone 346,288 (2.7%)
Multiple races 178,595 (1.4%)
American Indian 16,909 (0.1%)
Pacific Islander 2,715 (0.0%)
Other race alone 16,469 (0.1%)
2014 American Community Survey Estimated Population: 12,787,209
Under 18: 2,700,178
Over 18: 10,087,031
of which native-born: 9,328,872
foreign born-Naturalized Citizen: 397,495
-Not a citizen: 360,664
Total Citizen Pop 18+ : 9,726,367
2014 US Census Voting Report
Voting Age Population: 9,924,000 (Citizen 9,511,000) Note that this figure is off of the above figure by 153,000 total and over 215,000 citizens!
White Non Hispanic 8,015,000 (7,934,000 Citizen)
Hispanic 689,000 (540,000 Citizen)
Black alone 1,001,000 (adjusted at 13.7% Hispanic= 864,000, Citizen: 920,000 x 86.3% Non-Hispanic=798,000)
Asian Alone 295,000 (adjusted at 1.5% Hispanic= 291,000, Citizen 165,000 x 98.5% Non-Hispanic=163,000)
Therefore Other = 65,000
Reported registered by the Census, 2014: 8,121,000
Actual Registered according to the state of Pennsylvania, 2014: 8,251,050
Reported Voting by the Census: 3,803,000 voters
Actual Voted: 3,495,866 (for two candidates) : at least 8% “overvote”–Why does this matter? Because this is a survey of voters and clearly some fraction of voters ‘misremembered’ whether and likely who they voted for which, to me, implies that any poll is at least potentially off by perhaps as much as 8% merely because of inaccurate statements. Since most races in competitive states are determined by less than 8%, any poll of these states is liable to be wildly off base for this reason alone.
What follows is a detailed breakdown of the Pennsylvania electorate, starting with the overall electorate and then focusing on the biggest vote centers.
Ironically Philadelphia, the county with 40% of all growth in Pennsylvania in the years from 2010-2015, will likely make up a smaller fraction of the total vote than it did 12 years ago. Concomitantly, areas of the state with substantial population drops since 2010 have actually seen an increase in total voter registration in the last four years. This is the dilemma of the Clinton campaign in 2016: they have the demographic winds at their back, but have not succeeded in taking advantage of them as most of the changing demographics are occurring in the under 18 age group, while the general population of older (and whiter) voters has actually increased in the last four years in line with the increasing population of older people in Pennsylvania. It looks on paper as though things are moving favorably in the Democratic column and yet the opposite seems to be happening in the short-term. Since 2012, total numbers of registered Democrats has dropped from 4,266,317 to 4,116,095, about 150,000 voters, while the Republican total registration has increased from 3,131,144 to 3,200,191, an increase of about 70,000 potential voters. The number of people registered as ‘other parties’ has increased about 43,000 in the same period.
Overall, voter registration in Pennsylvania will probably match or slightly exceed that of 2012 (8,508,015). Compared to 2012, the totals favor the Republicans, who will likely finish with a total of about 3,240,000. The Democrats will likely finish with about 4,150,000, and other parties will end with about 1,180,000. In total registration is likely to finish at about 8,570,000, 0.7% higher than in 2012 and 2.4% higher than the 2004 total (8,366,663), but under the record 2008 registration of 8,755,588 potential voters.
A look at longer term changes in registration by party at the state level since 2004 shows a clear movement towards the Democratic Party, unlike the pattern of the last four years. In 2016 registered Democrats will exceed the 2004 total by 165,000 (to 4,150,000 from 3,985,486), while Republican registration has declined by exactly the same amount (3,405,278 down to 3,240,000). Interestingly, the registration of ‘other parties’ has increased by more than 200,000 over the same time period, which makes up the entire 203,000 increase in total voter registration. It seems that, rather than switching parties, voters are increasingly likely to choose to belong to neither party (although, as I mentioned earlier, I am one of these ‘other parties’ people and my preferences are clear).
Philadelphia in 2012 cast 12.00% of the total ballots and the 2010 Population was 12.01% of the total population of Pennsylvania. In recent years the city has given the Democrats an increasingly larger share of their total vote. Gore 2000-18.07% of all the votes won in the state of Pennsylvania; Kerry 2004-18.45%; Obama 2008-18.18%; Obama 2012-19.69%. This trend looks likely to continue as Wasserman has shown that most of the non-Philadelphia metropolitan area counties are becoming more red as measured by voter registration shifts.
As for demographics, Philadelphia in 2015 census estimates population 1,567,442 and PA estimate of 12,802,513; thus Philadelphia makes up 12.24% of the Pennsylvania population. Of the roughly 100,000 person increase in the population of Pennsylvania 41,436, or over 40% of the total, was in Philadelphia. This would seem to indicate that if voting levels remain the same that Philadelphia will exert more influence on election results. Indeed, in the past three Presidential elections, the fraction of the total vote cast by Philadelphia has increased from 11.68% (2004) to 11.94% (2008) and finally to 12.00% in 2012.
The down-side of the Philadelphia story is that overall voter registration is down from 2008 and 2012, although final numbers will probably show Philadelphia with 1.07 million registered voters, a small increase over 2004. As a percentage of the total registration, Philadelphia will likely be in the 12.60% range, lower than all three previous elections (12.92% in 2012, 12.87% in 2008, and 12.70% in 2004). This will likely result in a small drop in the proportion of the total Pennsylvania vote that comes from Philadelphia: in the last three elections the actual turnout as a percentage of total vote is usually about 1.0 % points below the Philadelphia fraction of state voter registration (2012- Votes 12.00%, registration 12.92%-difference -0.92% points; 2008- Votes 11.94%, registration 12.87%-difference -0.93% points; 2004- Votes 11.68%, registration 12.70%-difference -1.02% points). This implies that the Philadelphia fraction of the total vote will decline from 12.00% to 11.6-11.7%.
It also implies that Philadelphians are less assiduous in voting, where they are slightly underrepresented, than they are in registering to vote, where they are slightly overrepresented relative to their share of the population of Pennsylvania. This may actually be a false impression for three reasons: first of all the share of the population under 18 is higher in Philadelphia than that of the state of Pennsylvania (fill in the data here), while the share of non-citizens is also higher than in the rest of Pennsylvania as is unsurprising for a major international urban center. Thirdly, there is a large student population in Philadelphia, which explains in all likelihood the excess of voter registration (easy to do on campus), but the lower turnout totals (requires actually showing up to vote). My point is that it seems that there is unlikely to be either a conspiracy to attempt voter suppression or conversely an election fraud scheme to steal the result, despite the minor discrepancies in the data, and despite the accusations being hurled by Trump and elements of the ‘Alt-right’. Perhaps we should look into the mysterious increase in voter registration in Westmoreland County, the county which had the single largest population drop of any county in the past five years, he asks sarcastically.
2004 -Kerry 542,205 (80.44%), Bush 130,099 (19.30%). Democratic margin: 412,106 votes
2008 -Obama 595,980 (83.08%), McCain 117,221 (16.34%) Democratic margin: 478,759 votes
2012 -Obama 588,806 (85.29%), Romney 96,467 (13.97%) Democratic margin: 492,339 votes
Immediately the problem for the Republican Party in Pennsylvania comes into focus. If Philadelphia makes up 1 in 8 votes and the Republican candidate loses by an increasingly large margin as is clearly shown by the results above (and by the registration data, which as of September 26, 2016, shows that 121,131 people are registered as Republicans, down 11,000 from the 132,574 of the previous election, down from 147,068 in 2008 and from 175,434 in 2004. This is a 30% drop from 12 years ago; the 2012 figure was close to 25% below that of 2004 and was reflected in the 26% drop in votes for the Republican candidate).
Thus, the Democratic candidate comes out of Philadelphia with a 400,000 to 500,000 vote lead. The statewide margin of victory in the last three elections was 310,000 (2012), 621,000 (2008), and 144,000 (2004). Thus the close to 500,000 vote head start in Philadelphia has been sufficient to overcome any inroads made in the rest of the state by the Republican candidate. Is there reason to doubt that the margin of victory in Philadelphia will be any lower in this election? No, as the registration breakdown is virtually unchanged and the demographic changes favor the Democratic candidate. Thus turnout is the only variable in this equation, which is clearly why the campaign has become so nasty as the Republicans attempt to make things so ugly that turnout is suppressed, which they believe will help them win. My analysis of voting trends indicates that a likely Trump maximum vote in Philadelphia would be about 98,000 votes while a ‘low’ turnout result for Clinton would net about 559,000 votes, resulting in a margin of 461,000 votes. On the other hand, if turnout is high or is as one might anticipate from normal demographic change, Clinton stands to receive as many as 590,000 votes, while if Trump fares as trends indicate for Republicans, he is likely to receive under 90,000 votes. Here the gap between the two candidates reaches 500,000 votes. So, the effective range of vote margin is between about 460,000 and 500,000 votes, a number that is almost entirely dependent on turnout.
Personally I believe that turnout will be on the higher end rather than the lower end and thus I believe Clinton will have closer to a 500,000 vote lead in Pennsylvania before any other counties are factored in. If, as expected, Philadelphia represents about 11.6-11.7% of the electorate, then the electorate will total about 5.8-5.9 million voters, of which 700,000 are from Philadelphia. Of the remaining 5.1-5.2 million voters, Trump would need to win at least 2.8-2.9 million (about 54% of the remaining vote) to Clinton’s 2.3-2.4 million (about 46% of the remaining vote; also I am ignoring third-party numbers for the moment) to win the state. Is this possible?
7. Philadelphia and the Rest of Pennsylvania.
Pennsylvania is a large state with many other vote rich centers of population. Indeed the Republican candidate in the last three elections has been victorious twice in these three contests outside of Philadelphia. The one time the Democrats won, 2008, is arguably one of the “special event” elections I referred to earlier, in that an economic meltdown coincided with the appearance of a unique candidate able to motivate people previously less motivated to vote. Turnout for the 2008 election in Pennsylvania, exceeded by 200,000 votes the total of both the previous election and of the subsequent election. Thus the 2008 result was anomalous and, in general, the Republican candidate wins the rest of Pennsylvania.
In order for the Republicans to win the election in Pennsylvania, Trump needs to defeat Clinton by about 500,000 votes in the rest of Pennsylvania. In 2004, and 2012, the two most recent ‘normal’ elections, turnout was a little under 5.8 million voters. Philadelphia accounted for a little under 700,000 of these votes, leaving 5.1 million votes in the rest of the state. The total number of registered voters and actual voters in the state was not dramatically different in 2012 from the 2004 figures. As I have mentioned, Philadelphia and the four counties immediately surrounding Philadelphia, are the major sources of population growth in Pennsylvania. In fact, the population of the 5 county Philadelphia metropolitan area in the past 5 years alone is 85% of the total growth in the state over that time, with the majority of counties in Pennsylvania undergoing population loss over the same period. One might expect voters from the five-county Philadelphia area to make up a slightly larger fraction of the electorate in 2016 than they did in 2012 and as they did from 2004 to 2012.
Figures from the Pennsylvania Secretary of State for current voter registration (as of September 26), which does not officially close until October 11 for the November election, currently exceed 2004 figures and are likely to match or exceed the registration numbers from 2012. The figures for the five counties of Philadelphia metropolitan area already exceed those of 2004 and are 4,000 short of the 2012 registration total 2,821,068, a number which will certainly be matched or surpassed. Figures for the rest of Pennsylvania are down about 45,000 registered voters from 2012, which may also be made up in the remaining time at the current rate of registration.Thus, there is clearly no upsurge in potential new votes which will change the balance of voting in Pennsylvania. Donald Trump’s candidacy has not resulted in a significant surge in new voter registration at all; the likelihood is that the 2016 registration figures will be almost the same as, or slightly higher than those of 2012. The rise in Republican registration, while significant, has failed to reach the level of either 2008 or 2004, both years the Republicans lost the election. Registration figures essentially show a relatively static electorate, increasing slightly in places where the population is increasing and decreasing where the population is decreasing. The supposed rise of Republicans engendered by Trump’s ascent during the primary is minimal: the total gain in Republican voter registration from 2012 is 69,000, a 2% increase. It has been argued by others (Wasserman) that Pennsylvania in general is becoming red based on changes in voter registration dynamics since 2012 (particularly the drop in registered Democrats); however, the number of registered Democrats in Pennsylvania has increased since the 2004 election by over 100,000, while the number of registered Republicans in Pennsylvania has decreased by over 200,000, and John Kerry won Pennsylvania by more than 100,000 votes against a ‘war’ President in the aftermath of September 11, 2001.
Thus, in order for Trump to win, he has few strategies. Of the 5.1-5.2 million remaining votes, Trump must win by about 500,000 votes to win the election, either a) by increasing voter turnout to a number higher than 5.2 million, b) by depressing Clinton turnout or convincing Clinton voters to try another candidate such as Johnson or Stein, or c) by getting likely Clinton voters to switch (the most efficient method since a vote gained is also a vote lost) . As you may have surmised by now, the first method is probably a nonstarter for the reasons already outlined, specifically the fact that people die in Pennsylvania at an even higher rate than they do elsewhere, so the massive surge in voters in non-Philadelphia Pennsylvania is extremely unlikely as there are not enough bodies to make up the numbers as I will show later. I also do not believe that the two ‘minor’ candidates, who have received little and mostly negative press, will succeed in attracting more than a few thousand new supporters, and if anything, they will be peeled from the Republicans at least as much as the Democrats. By the way, my opinion about most of the supposed diehard ‘Never Trump’ voters is that they will largely do exactly what they said they would not do, and cast a ballot for Donald J. Trump. As for depressing turnout, that may affect the race to some degree, but the implication that the necessary 300,000 voters of an electorate already favoring the Democratic candidate will sit on their hands out of disgust at the system is not something I would put money on.
That leaves the third way, a version of which has been mooted by many; that is, the ‘Reagan Democrat’ who will vote Republican. The best place to find these voters is in the area surrounding Pittsburgh, particularly Westmoreland County, which has the largest voting bloc outside Allegheny County (Pittsburgh) in western Pennsylvania. It is true that the number of registered Republicans has increased by 10,000, from 90,032 to 100,073 registered voters, or 10%, since 2012 and it is also true that the number of Democrats has declined by a similar number from 122,432 to 112,508 at the same time. However, George Bush received 100,000 votes to Kerry’s 78,000 in 2004 when the number of Democrats was over 142,000 while the number of Republicans was 89,000. Obama lost both times in Westmoreland, in 2008 by 73,000 to 103,000 and in 2012 by 64,000 to 104,000. So Bush, McCain, and Romney each won with a vote total larger than the entire number of registered Republicans. Clearly, despite protestations to the contrary, many Democrats have already been casting their ballot for Republicans in Westmoreland for some time. More interestingly, although the number of votes for the Democratic candidate has been dropping there for years, the number of Republican votes is increasing only very slowly, from 100,000 to 104,000 votes, so that the margin of victory has gone from 22,000 votes to 30,000 votes to 40,000 votes in 12 years. It is likely to increase again, perhaps to 50,000 votes.
The problem is that this cuts only an extra 10,000 votes from the 500,000 vote head start Clinton will have from Philadelphia, and Westmoreland is the 11th largest county by population in Pennsylvania and is likely to be the 6th or 7th largest source of votes for the Republican candidate. (Ironically, the largest source of votes for the Republican candidate is in the five county Philadelphia metro area which, as I shall show below, is likely to widen the gap between Clinton and Trump, rather than shrink it.) What seems to be happening in the suburban Pittsburgh counties is that Democratic numbers are falling but that, as the population in general in Western Pennsylvania declines, the Republicans are gaining ground only very slowly in absolute numbers, so that the net impact is to depress the influence of counties like Westmoreland on a statewide race. Overall registration in the county has declined from 256,365 in 2004 to about 240,000 for the 2016 (estimated final figure). In 2004 3.10% of the total vote in Pennsylvania but by 2012 that figure had declined to 2.95%. As Westmoreland County was according to the most recent Census figures, the county with the largest absolute population decrease (over 7000 fewer people) in the past 5 years, it seems unlikely that the influence of the county can make up for the Republican decline in the Philadelphia area.
Thus far I have discussed Philadelphia and Pennsylvania as if they were two alien entities with nothing in common. What has not been said thus far is that there are a few other counties where Clinton is likely to do very well, thus increasing the margin over Trump. An annoying article in Philadelphia Magazine after the 2012 election seemed stunned to discover that Obama won only 13 counties in the state out of 67, neglecting to mention that these 13 counties account for more than 54% of the entire population of Pennsylvania (blue bullets, my friend, blue bullets). It was also not pointed out that the margin Obama racked up in these thirteen counties was about 800,000 votes. Trump would need to win the remaining 54 counties, with approximately 2.6 million votes, by a larger margin than 800,000, which means he would need to win the rest of the state (including Westmoreland County) by at least 1.7 million votes to 900,000, almost two to one. Romney won the same counties by 1.5 million to 1.0 million, while McCain received 1,443,000 votes to Obama’s 1,347,000 in the same counties and Bush earned 1,514,000 votes to Kerry’s 1,033,000 votes. Notice how similar the 2004 and 2012 results are. Not only does Trump have to increase his vote total by 200,000, he has to keep Clinton’s down 100,000. Voter registration has declined in these counties in the last four years, so Trump is reliant on a massive upsurge of voters plus a nearly 10% switch of voters from the Democratic party to the Republican Party, which is certainly not seen in registration numbers and is unlikely.
A further point I have not mentioned is that the four largest counties in which Obama lost in 2012, Berks, Chester, Lancaster, and York, all less than 100 miles from Philadelphia, have all seen an increase in the percentage of registered Democrats relative to registered Republicans, unlike the general trend in the state. In 2012 Obama received 370,000 votes in these three counties to Romney’s 452,000, a gap of 82,000 votes. In the unlikely event that there is a reverse in voter registration, a sudden change in the demographics which day by day favor the Democratic party, or a massive rise in secret votes for Trump from disaffected Democrats, this gap is likely to shrink in 2016 rather than grow. These four counties contain almost 2 million inhabitants, more than 15% of the state population. Thus the 17 counties thus far discussed account for 70% of the population and votes. The total number of available votes to Trump if these three growing counties are removed is now close to 1.7 million at most. In 2012 Romney won by a margin of 1,041,000 to 638,000 or 400,000 votes. Thus, even if Trump holds the margins in the four above counties, he is still more than 700,000 votes behind. It is virtually impossible make up more than 300,000 extra votes in this small remaining voter pool, requiring half of Obama’s remaining votes to switch to Trump and assuming he wins all of Romney’s votes. Did I mention there are fewer registered voters in these remaining 50 counties than there were in 2012?
To recap: Clinton will have a large lead in Philadelphia and will have to “hang on” in the rest of Pennsylvania. As I have said, however, there are other pockets of Democratic strength in other parts of Pennsylvania which will be crucial in maintaining the margin necessary for victory. I will examine some of these in detail below.
8. Suburban Philadelphia.
For the purposes of this article this region is essentially the four counties nearest to Philadelphia: Montgomery, Delaware, Bucks County, and Chester County. These four counties, along with Philadelphia, make up an area of southeastern Pennsylvania which has more than 4 million residents, 32% of the population of the state (versus 31.6% 5 years ago). The five counties have a total of 2,817,058 registered voters as of September 26, 2016, slightly below the 2012 total of 2,821,071, but more than the 2,790,646 voters of 2004. Almost 1.6 million of these voters are registered Democrats, while 837,581 are Republican. In 2012 the Democrats had 1,601,682 and the Republicans had 852,169 voters. Thus, the total number of Republicans has declined from 2012 while the number of Democratic voters has stayed almost the same and will likely exceed the 2012 number by October. Compared to 2004, the number of Republicans has declined from 1,036,277 to the current 837,581 a drop of almost 200,000 voters. The Democratic numbers have almost exactly reversed rising from 1,414,151 to the current 1,600,330. The number of ‘other party’ voters has also increased from 340,218 to about 379,147 (fix), to complete the picture. If the Republican total is combined with the ‘other parties’ total, the number of Democratic registered voters still exceeds that total by almost 400,000 potential voters.
How did the five counties vote as a whole in recent elections? Unsurprisingly, the Democratic Party has performed very well in the area in the last three elections. A brief aside here: I only consider the last three elections because the elections prior to 2004 were quite different in their overall structure; in particular voter registration and turnout has increased dramatically in the last three elections relative to prior elections. In Pennsylvania, for example, no election had more than 8,000,000 registered voters prior to the 2004 election, which saw an increase of more than 500,000 voters over 2000 and had at least 2 million registered voters more than virtually all the other elections ever held in Pennsylvania. As for actual voters, the 5 million votes cast mark was exceeded in 1960 (5,006,541), but subsequently was not exceeded until 2004, when almost 5.8 million votes were cast. This election has already achieved 8.4 million registered voters and may well pass 8.5 million before October.
To return to the results of the five counties in the last three elections:
Year D Vote R Vote Democratic % of PA total Vote 4 County margin
2012 1,278,876 663,120 +615,756 34.26% +123,417
2008 1,345,107 662,715 +682,392 33.85% +203,633
2004 1,200,000 700,770 +499,230 33.16% +87,124
Obama did have a fall-off in his vote total in 2012, but his numbers were still much higher than Kerry’s 2004 numbers. More interestingly, though the total fraction of the state’s vote total increased by 1.10% from 33.16% to 34.26%, the actual number of Republican votes decreased by 5%. Another thing to notice is that even excluding Philadelphia, the differential for the Democratic candidate increased in the four remaining counties. In fact, in 2004 Kerry had a net gain of 87,124 votes in the four counties, but by 2012 Obama had maintained a 123,417 vote advantage in the four counties, difference of over 36,000 votes. As I explained earlier, these numbers merely reflect the changing registration data, which in turn reflects the change in the region from a predominantly white, working class area to a more ethnically mixed and much more educated population. There is little reason to believe these numbers will decline much if at all in 2016. Thus it seems likely that Clinton will have at least somewhere in the range of a 600,000 vote advantage in these five counties in 2016.
I have neglected to discuss demographics in as much detail as I would like to only in the interest of keeping this article focused. Suffice to say, the demographics of these five counties, as well as those of another half-dozen or so of the largest counties in Pennsylvania favor the Democratic party in the long-term. Pennsylvania today resembles the demographic balance of New Jersey 12-16 years ago when it too was considered winnable by the Republicans (and was not).
To summarize: The five counties of the Philadelphia Metropolitan area will likely cast one-third of the vote in the upcoming election and there is little prospect of Trump closing the margin of 600,000 votes that has been the rule the last two elections. If anything, the trend seems to favor the margin widening. Thus the remaining two-thirds of the state must provide Trump with a similar margin in order to have a chance at victory. Now we will turn our attention to the next largest voting bloc in the state, Pittsburgh.
Pittsburgh is more complicated. Allegheny County is the source of the second largest vote haul in Pennsylvania, and has long been the single largest source of votes for Republican candidates. Pittsburgh makes up a quarter of the population of the county and is a Democratic stronghold. The suburbs, like other parts of Western Pennsylvania, have long been conservative and, as the population ages and declines, increasingly vote Republican. So, it is certain that Clinton will win Allegheny County but the margin of victory here will decline in all likelihood from the previous election.
Like Westmoreland County discussed above, the general trend in voter registration in Allegheny County is a decline in Democratic party registration accompanied by a very slow increase in Republican registration. Relative to 2004, Republican registration has actually declined from 262,692 to 251,845 at the most recent count. Since 2012 Republican registration has increased by about 2,000 voters, from 250,279. Democratic registration has been in a steeper decline, currently at 527,995 voters, down almost 30,000 from both 2004 and 2012 (557,900 and 556,819 respectively).
Also like Westmoreland County, Allegheny County has seen Republican vote totals in the last three elections that exceeded the total number of registered Republicans, a very unusual occurrence as a rule but one that strongly implies that many democrats in the county actually vote for the Republican candidate for President. Thus it is a little difficult to predict what will happen as a result of the changing voter registration statistics. Will the percentage voting for the Republican increase or decline relative to the total number of registered Republican voters? The absolute number of votes cast for Romney was down by 10,000 compared to both John McCain and George Bush. Assuming the decline is reversed by Trump, can we expect more than a 10,000 vote increase? Are there more Trump voters out here than one might predict? It seems extremely unlikely to me that voters who did not cast a ballot for the war hero John McCain or the ‘action’ president George Bush will suddenly decide that Trump is ‘the one’ and convert en masse to the Republican ticket. What is more likely is that the total Democratic vote is likely to decline while the Republican vote holds the line or increases slightly, correlating with what the voter registration statistics suggest. In the last three elections the Democratic vote totals in Allegheny have been 368,912 (2004), 373,153 (2008), and 352,687 (2012). If votes correspond to the change in voter registration, one would expect that Hillary Clinton will receive between 330,000 and 340,000 votes to about 265,000 to 275,000 for Donald Trump, a margin of Democratic victory ranging from 55,000 to 75,000 votes, down from slightly over 90,000 votes for Kerry in 2004 and Obama in 2012.
It is this type of result that the Trump campaign is touting as the formula for victory in November, as the margin of Democratic victory has been reduced by about 15,000-35,000 votes. Like Westmoreland County, however, it was achieved not by increasing Republican votes (at least not by much, especially compared to George Bush in 2004) but by reducing the total Democratic vote in these areas. A low turnout is the best chance for a Trump victory as it will disproportionately affect the Democrats. This, in my opinion, goes a long way to explaining why this election campaign has been so ugly; turning off voters is seen as a winning strategy for the Republican Party, so hold your base and convince likely Democratic voters that Hillary Clinton is not worth showing up to vote for with a steady stream of ugly revelations about the corrupt practices of the Clintons. As we have seen as the election progresses, Trump supporters seem absolutely uninterested in any dirty laundry in Trump’s past, so it makes sense for Trump not to not care about how he seems to the mainstream American political establishment. There is no magic formula here, just a conscious effort to get out the base and hope that the other teams players decide not to show up.
Only I do not believe this strategy will work, any more than I believe that polls swinging wildly from day-to-day are any remote indication of the state of play in Pennsylvania or in any state for that matter. The dynamic of the electorate is the critical factor in an election; polls are not refined enough to capture the change, which is why they often have a bias in one direction or another as they try to assess the likely voter. But ultimately, the voters who tend to vote, show up to vote, the voters who do not vote do not show up to vote (despite all the stories about massive surges in Hispanic voter registration to counter Trump support, and plumbers who never vote who plan to vote for Trump). The only exception is that, in election years, there is a special voter who shows up every four years to vote for President but essentially does not otherwise vote. So, in the presidential election years the total turnout was 5,753,670 (2012), 6,010,519 (2008) and 5,769,590 (2004), while for Governor’s race in 2014 and 2010 (in which there was also a Senate race) turnout was under 4,000,000 voters.
Low turnout races favor candidates whose voters are more reliable, which typically turns out to be Republican voters. The higher the turnout, the higher the likelihood that the Democratic candidate will win. For example, the Democrat Joe Sestak lost the 2010 Senate race to Pat Toomey by 80,000 votes. Had the city of Philadelphia voted in the proportion that it typically does in Presidential election years, where it makes up anywhere from 11.6-12.0% of the total electorate, in line with its relative share of the state population and registration, Sestak would likely have shaved more than 40,000 votes off that margin. Getting another 40,000 votes from the other 66 counties, representing 88% of the population would similarly have been likely if the voting patterns of presidential elections held for all elections. It would not surprise me at all to see Toomey lose his reelection battle this year, as he has the same electoral structural problems that face Donald Trump.
10. The “Scranton Strategy”
The anecdotal evidence that Trump is winning often is accompanied by a story about some lifelong Democrat in one of the smaller old industrial towns like Scranton or Allentown, or Wilkes-Barre, who is switching over to vote for Trump in this election. Is this story even true ? if it is true, can it affect the outcome of the election? Scranton is in Lackawanna County, Wilkes-Barre is in Luzerne County, and Allentown is in Lehigh County while the city of Easton is in Northampton County. These were all counties won by Democrats in recent elections, so the putative switch of allegiance by white working class men should in theory strongly impact the results in these counties. The total population of these four counties is slightly under 1.2 million, slightly under 10% of the total Pennsylvania population. In 2012 Obama took 272,014 votes to Romney’s 221,730 votes, a margin of 51,000 votes. In the four years since Democratic registration has fallen slightly from 422,276 to about 405,000 while Republican registration has increased by 12,000 voters from 250,232 to 262,140. This is a similar, although not as drastic pattern as that of Westmoreland or Allegheny County, where the Democratic decline was steeper than the Republican increase. However, should the results of changes in voter registration be reflected in the voting results, one might expect the Republican total to increase to about 233,000 votes, which is fewer than the 241,981 votes George Bush received in 2004, when he lost the state to John Kerry by over 144,000 votes, despite holding the margin in the five county Philadelphia area to under 500,000. Kerry took 266,532 votes from the four counties under discussion here. Should Clinton lose 10% of the vote total from 2012, she would still earn about 245,000 votes, so even if Trump were to match Bush’s vote total from 2004, the Democratic margin would be about 4,000 votes, a gain of about 47,000 votes. Again, this is essentially a strategy of hoping Democratic numbers fall off and trying to get the maximum turnout which, in all honesty, is unlikely to exceed George Bush’s 2004 vote total of about 2.8 million votes statewide.
11. The Rest of Pennsylvania
Thus far I have discussed counties in Pennsylvania making up almost 60% of the entire population and I have concluded that, in the five county Philadelphia area, the margin of victory is likely to be in the range of 550,000 to 650,000 votes. Pittsburgh might give back anywhere from 15,000 to 30,000 votes, but the Clinton margin will still grow after the votes in Allegheny County are counted. Westmoreland might give Trump 10,000 more votes to reduce the margin. And the four counties just discussed at most might give 50,000 votes to Trump in a perfect world. This totals at best 90,000 votes. If we assume that Clinton does relatively poorly with her base in the Philadelphia area, she may have only the minimum 550,000 vote margin. Adding Allegheny County votes, the margin at worst will increase to over 610,000 and even the four counties just discussed will likely add to the Clinton margin. Westmoreland might reduce the margin by as much as 60,000 votes. Thus, after 60% of the electorate has been examined, Trump at his best might have reduced the margin to about 550,000 votes. Thus with about 2.3 million votes remaining, Trump must win by a minimum of more than 550,000.
The difficulty is that the Democratic Party normally wins two other counties in Pennsylvania, Erie County and Monroe County, and is likely to win in Dauphin County, where Harrisburg the state capital is located, as well as running evenly or losing by a very small margin in fast-growing Centre County, home of Penn State University, and Berks County, where Reading is located. Together, these four counties have over 1.1 million residents. Now we have accounted for about 70% of the electorate where does Trump stand? In the same spot he was standing after the five county Philadelphia metro votes were tallied, at a minimum 550,000 votes short of victory and most likely closer to 600,000 votes behind Clinton. The remaining 1.8 million voters must give Trump a victory of more than 550,000 for him to win the state, so he must win about 1.2 million of these votes, or close to two-thirds of the vote. The single largest outstanding county is Lancaster, a reliably Republican stronghold. Even here, however, trouble lurks for Trump, as voter registration since 2004 has strongly favored the Democrats, who have gone from 82,172 registered voters to about 105,000 as of this writing, while the Republicans have dropped from 184,852 to about 170,000 voters. So, Bush’s winning margin of over 71,000 votes is unlikely to be matched. Even the 2012 margin of Romney over Obama, 42,000 votes, is unlikely to be much expanded and is likely to be slightly reduced by Clinton.
Fewer than 1.6 million votes remain and the margin continues to hover at around 500,000 votes. Now Trump needs to win 2/3 of the remaining vote. How did Romney do against Obama? Romney took slightly over 1 million votes while Obama earned a little more than 600,000 votes. Thus Romney clawed back 400,000 of the 710,000 vote gap he had to make up after the counties above had been tallied. In the worst case scenario for Hillary Clinton, she will have only about a 500,000 vote lead after all the above counties have been accounted for. If the margin is similar to Obama’s against Romney, Clinton will win with about 100,000 votes to spare. But in a very bad climate, it is theoretically possible for Trump to claw back the votes to win. A look at the largest remaining county, York County, illustrates why that might be difficult. It is true that the number of Republicans has increased slightly over the past four years while the Democratic registration has very slightly dropped, but Obama lost by 40,000 votes and Kerry lost by 50,000 votes 8 years earlier, so the worst likely outcome for Clinton is a performance like Kerry, losing by about 50,000, only any Republican gains in the vote total are likely to be matched by Clinton gains as well. In 2004 there were 89,646 registered Democrats, while the most recent figure is 100,026, whereas the Republican margin has increased by only 5,000 voters in the past 12 years. Even if Trump does very well in York, Clinton is likely to get about the same number of votes as Obama, diminishing any presumed success of the Republicans in this, their largest single source of remaining votes, by a large margin.
There are now fewer than 1.4 million votes left to contest and the Democratic advantage at the worst for Clinton is still over 400,000 votes. Romney’s margin in 2012 was 362,000 votes (925,000 to 563,000). So a swing of close to 50,000 votes needs to occur to close the gap. Another problem now emerges. The remaining vote tally assumes that the voting pattern of the last election will hold for this election, but it should be apparent from my discussion of demographics above that most of the remaining counties have seen population declines in the last five years according to the Census. In fact, the entire remaining population of Pennsylvania under discussion had a net loss of over 30,000 residents in a five-year span. Though the number is not particularly large, it points to the relative lack of growth compared to the Philadelphia five county metro area, which increased in population by 85,000 people in the same period. Unsurprisingly, voter registration for the 25 largest counties in Pennsylvania has increased as a fraction of the total from 82% to 82.5% in the last four years. These 25 counties already made up almost 83% of the vote in the 2012 election, so it seems likely they will make up an even larger proportion of the election in 2016. There is, in all likelihood, a good chance that the figure of 1.4 million remaining votes is on the generous side. It is possible that the numbers could be made up in this pool of voters, but doubtful, and it is important to remember that I have tried to use the likely worst case scenario for Clinton, yet she is still likely to eke out a win.
And of course there is one final problem for Trump which I alluded to earlier: third-party support. In 2016 much has been made of the population of undecided voters who may ultimately plump for Libertarian Gary Johnson or Green Party candidate Jill Stein. If the predictions come true, Trump is finished because the number of votes being suggested (as much as 10% of the vote) are as, or more, likely to come from Trump than from Clinton, and so the math is not there to overcome the already daunting gap. In my opinion, as I stated previously, I do not believe the hype, and expect a modest increase in third-party vote casting in 2016 from about 1.24% to as much as 2 or perhaps 3% at most. An extra 1-1.5% of the vote being distributed elsewhere may not seem like much of a problem, but in a presidential election one candidate needs only defeat the other candidates, not win a majority. I do not believe it is possible for Trump to win a majority, but the prospect of Clinton winning even without a majority of the vote is a definite possibility. It is, of course, possible for Trump, in theory, to win by denying Clinton a majority and winning a plurality of the vote. This is the scenario I have been describing thus far. Even here, the odds of victory for Trump are quite small.
I tried another approach to see what Clinton’s situation is likely to resemble. After collecting data for the 25 largest counties in Pennsylvania, which account for 83-85% of the vote, I calculated that Obama in 2012 had collected 2.67 million of his 2.99 million votes, earning the remaining 332,000 from the remaining 900,000. Romney earned 568,000 of the remaining votes, recovering about 246,000 votes of the more than 550,000 vote gap from the 25 largest counties. Clinton, according to the type of calculations I have described above for the counties I have discussed, will at the absolute worst likely have over 2.5 million votes after 25 counties. Even if she performs 10% worse than Obama did in 2012 in the remaining pool, Clinton will still receive over 2.8 million votes in Pennsylvania. George Bush received a little under 2.8 million votes in 2004, Romney and McCain earned 2.68 million and 2.66 million votes respectively in 2012 and 2008. Trump would need to have 120,000 more votes than Romney to match the total of George Bush, and assume the impact of third-party candidates is negligible, in order to perhaps match Clinton’s worst possible vote total. Based on the evidence presented, I think that these are all unrealistic situations. In the best of circumstances, Trump will perform slightly better than Romney, perhaps even matching Bush, although the presence of that small increase in third-party voters will make even that task very difficult. Clinton will perhaps in the worst case, earn fewer votes than President Obama did in 2012 but, the fact that this is an open election will likely result in a higher turnout which will favor Clinton as her margin will increase in the largest counties, leaving more room to perform poorly in the counties such as Westmoreland. It is even likely that Clinton might end up with more than 3 million votes if the turnout is slightly higher than it was four years ago, which seems possible. It seems improbable that she will earn fewer votes than John Kerry, who won the state in 2004 by almost 150,000 votes.
Hillary Clinton 2.9-3.0 million votes; Donald Trump 2.7-2.8 million votes. Third Party Candidates: 150,000 votes. Turnout: approximately 5.9 million.
The simplest way of explaining the result in a broader sense is to say that the Republican high point was in 2004, that the demographics have inexorably changed the dynamic of the state of Pennsylvania, and that this is likely to be the last close election if the current version of the two parties continues unchanged and if the population of Pennsylvania continues to evolve in the current direction from the standpoint of education, age, racial make-up, and geographical concentration.
The campaign of Donald Trump relies on victory in at least two of the three largest states in the group I discussed earlier in this article: Florida, Ohio, and Pennsylvania. A failure to win one of the three makes the probability of a Trump victory in 2016 low at best. Victory by Clinton in Pennsylvania, coupled with the expectations I outlined above regarding the distribution of electoral votes, leaves Hillary Clinton with 266 electoral votes. Victory in any one of Maine, New Hampshire, Nevada, or Iowa, all states won by Barack Obama twice, is sufficient to garner the 270 electoral votes necessary to win the election. It seems very unlikely that Trump will win every one of these states, and so I conclude that Hillary Clinton will win the election.
Of course, this is a prediction and predictions are, by definition, uncertain. It may come to pass that I am wrong and that Trump manages to win. In that event, it is almost certain that, rather than an upsurge of hidden support, the likely reason will be that Clinton’s vote total was suppressed by lack of enthusiasm rather than that thousands of voters suddenly came out for Donald Trump. Whatever the final outcome, I have presented data that has made a prediction. If the prediction turns out to be incorrect, I can examine the data that results from the upcoming election to refine and improve my methodology. The fact is, the data is already sitting there to be studied. Change in a state’s voting patterns does not happen in giant bursts or in wild swings as is often implied, but rather is the result of slow evolution in the makeup of the population. The type of analysis I have tried to do here for Pennsylvania is an attempt to examine the data for signs of this change, a method I think much more productive than some of the material being promulgated that purports to be analysis. I thought of using Massachusetts as a test case but in the likely event that Clinton wins by either 56% or by 62% the result will be of less interest to the reader than if the result is 49% to 47% and the balance is in doubt.
An underlying implication of my hypothesis that the makeup of the electorate changes only very slowly is that the Electoral College vote is less dynamic than it is presented in the media and by the campaigns themselves. I am not saying that change is not possible, nor I am saying that it is easy to predict the exact number of electoral votes. What I am saying is that there is a narrow range of possible electoral scenarios based on the fact that most states do not change enough in one election cycle to affect the overall final results from election to election. I am not saying this is a good thing; it is merely an observation. The implication of this observation is that the endless parsing of primary results to make statements about the general election, the discussions of various theories about the impending transformation of the traditional party appeal to one group or another, the coverage of head-fakes by candidates, who go to a state they have no chance of winning to attract a day or two of coverage, the daily examination of national polls and the countless hours spent poring over the polls to assess the effect of an event, a speech, a stumble or cough by a candidate, is all a waste of time. The election results of most states are already settled before any candidate tosses his hat into the ring. The few states where the election is contested are the result of local factors, and are mostly to do with demography.
A deeper analysis of the likely voter, a more informed discussion of the actual changes in the makeup of the electorate in various competitive states, and a change in the ESPN-style coverage of elections that has taken over the airwaves, although superficially less ‘exciting’ than the current horse race style of coverage, could prove useful in other ways. It might allow us to focus attention on the value of the current system itself: if most states are not competitive in each election cycle, is there something wrong with the system? If election dynamics are less malleable than is generally understood, is the problem human nature, a cynical and inflexible political structure, or do people not care enough about elections to make them worth even having? Is a two-party system the optimal system for government? Would more get done in Congress if there were many parties that formed changing coalitions to pass pieces of legislation than the current system where a vote with the other team is an invitation to be kicked out in the next primary by a more hard-line replacement? Does the focus on winning the electoral college, on cobbling together the minimal coalition of voters sufficient to win the current election, narrow the appeal of one party or the other to the majority of the electorate, thus increasingly alienating the overall electorate? A friend of mine living in Asia recently wrote me to say that the results of the election were irrelevant as both candidates were controlled by global corporate interests and that US foreign policy was unlikely to change either way. Is he right? Is the skeptical electorate right that it does not matter? Personally, I do not believe that is true and is part of the reason I have tried to take a closer look at this election.
Hillary Clinton will spend almost no time in a city like Boston, precisely because she will win overwhelmingly. Donald Trump was mocked for ‘wasting’ time in Mississippi this summer when he could have been wooing undecided voters in Ohio. Is it a good thing that candidates are expected to take massive chunks of the country for granted and ignore them? Voters in uncontested places such as Boston have the option of voting for a third-party candidate without fear of throwing the election to the ‘bad guys’ but only because it is fully understood that the vote is essentially a protest vote and will not actually result in a change in the political structure. The less it matters, the more likely it is to occur, which seems exactly backwards. The last real change in the parties contesting the election occurred prior to the Civil War. Is that something to be proud of or is it something that needs to be fixed? Why is it that candidates as disparate in temperament, background, even basic moral beliefs as George Bush, John McCain, Mitt Romney, and Donald Trump can all be nominated by the Republican Party and yet each election cycle ends with almost the same basic outcome (the few changes in the last four or five elections are precisely the effect of the focus on a few states in a transitional demographic pattern)?
Below I have appended some more general thoughts about the current state of political analysis. I have not analyzed the data in this article as deeply as I would like to have done in the interest of presenting my conclusions before the election. I would have preferred to present a deeper analysis of the breakdown of the electorate by race, class, level of education, age and other factors. I also would like to have been able to do the same type of analysis for other states like Florida, but time has prevented me from being able to present data that is currently in a half-finished state. However, I believe that the results of my research as they currently stand and the type of analysis that I have done here are more useful to contemplate than most of the predictive analysis of pollsters and others who seek the answer using more ephemeral data.
13. Some General Thoughts About Election Analysis
a. Polls and their fluctuation are assessments and reflections of the mood of the election analysts rather than actual indicators of fluctuations in voting outcomes. The outcomes of elections are far less variable than polls indicate. Prediction markets and pundit web sites assessing the state of the race are looking at just that, the effect of the opinions and pronouncements on the public of the professionals who are analyzing the state of the race at any given moment. They do NOT reflect, in my opinion, variability in the likely outcome of the race. This explains the volatility of polling averages, which often shift wildly from day-to-day, implying that the voting public changes their mind with alarming frequency. Were that true, then states like South Carolina, Rhode Island, and about 40 other states would be much less reliable in their voting outcomes. Campaigns and the events surrounding them are analyzed in-depth by the media and the message is then reflected back to the media by the collection of opinions of the general public on the state of the race.
b. Anybody with the ability to look at a map and some basic numbers (i.e. previous election results) can almost certainly predict the correct outcome in 80% of the states of the union, and can likely determine the correct result in 5 of the remaining 10 states as well merely by flipping a coin. In other words, pundits and betting market predictors are essentially speculating on the results of 4 or 5 races at most. In the case of 4 races one in 16 prognosticators will be correct by randomly choosing the winner in each case. Predicting 48, 49, even 50 correct results is not particularly difficult and someone relying on luck and a small amount of knowledge would have a decent chance of predicting every result correctly. The fact that with polls, as the data points pile up and the election deadline approaches, the numbers tend to fall into a consistent pattern that is often close to being correct, if not 100% correct, should be unsurprising, since they are merely taking the collective guesses of a large number of people and aggregating them. It is a bit like Wikipedia: if enough people work on an entry, the accuracy generally improves and the quality of the article should approach that of one written by a professional from the Encyclopedia Brittanica for example.
c. Betting markets, prediction markets, any organization which attaches a number to the chances of a given candidate winning are merely aiding and abetting the media penchant for a horse race by giving them a veneer of ‘hard data’ to support their theories about the changing dynamic of the election. Polls are small snapshots of small groups at a given moment. They are likely less accurate than the stated margin of error because of what I will call ‘lying’. As I stated above, more people claim in surveys to have voted than actually voted. If asked whether someone is likely to vote in a survey, surely the temptation to say yes is strong, even if the truth is less clear. The problem, in my view, with analyses of polls by an organization like 538 is that they are taking an aggregation of small, mostly untrustworthy data samples and applying a formula to them that tries to make sense of their lack of trustworthiness. This is certainly superior to the old method of prediction, putting ‘experts’ like Dick Morris on TV and having them say whatever they feel like without having any data whatsoever to back up their choice. However, It is unclear to me how useful an exercise it is at all. If I was told there was gold buried in my yard, I am sure that if I used a spoon to dig up small sections of the yard at a time I would eventually discover the correct location by process of elimination, just as I am sure that if I surveyed Massachusetts voters 1000 times I could eventually filter out the noise and home in on the signal and be able to ascertain what I already knew anyway, that they will vote for Hillary Clinton. The thing is, I already have the gold laying in front of me in the form of previous elections and demographic analysis of the likely electorate. So why would I need to ask a few people their opinions many separate times, then collect the totals of these many small samples and aggregate them into the pool of data already that already exists. The existing pool also has the advantage of not lying, and includes only actual votes and voters. Since the overwhelming majority of the voters do not change from one election to the next, the main purpose of political analysts should be to determine the potential changes in the electorate over time and to then compare that to any new developments in the dynamic of the two parties in order to be able to make intelligent statements about the likely result of a particular election.
d. The voting habits of individuals do not change as much as we are led to believe. Individuals typically vote the way they always vote, and what changes from election to election is the demographic dynamic of each electorate. Put more bluntly: older voters die and are replaced by younger voters. Parties respond to the changes as necessary. Corollary- there is no such thing as a swing voter. People in general do not change their basic opinions, attitudes, habits. Ask yourself how many people you personally know who were once conservative and are now liberal. By conservative I do not mean Republican (which as I argue, is not constant), I mean people who have a fundamentally conservative outlook of the world and have changed to a more liberal outlook, or vice versa. What changes over time is not the particular world view of a single voter, but the collective world view of the particular aggregation of people at a given moment of time in a given geographical unit which is electing a candidate for that particular geographical unit (a state, a congressional district etc.). Instead parties change in response to the search for the perfect dynamic that will result in electoral victory. As stated above, populations change as a result of varying birth rates and death rates, young people grow up in a changed world from their parents, immigration and internal migration change the local dynamic and parties shift their collection of messages in response to the changing dynamic.
e. Corollary of above: the perfect dynamic that will result in electoral victory does not mean that the platform of a party appeals to a majority of people, just to a plurality of people likely to vote in a given election. Thus many “liberal” states have conservative governors as liberal voters are much less likely to show up in non-presidential elections. It is a myth that elections reflect the majority opinion of a constituency as the majority of the population does not vote in a typical election. Even in a presidential election the actual turnout is only marginally higher than the number of people not voting, so cannot truly be said to reflect the will of the majority. Perhaps 60% of those not voting might have voted for the losing candidate, in which case the election outcome might have been different. Perhaps there is an argument that people not voting are making a statement of preference, but the truth remains that most elections are not in fact necessarily an actual preference of a majority of the population, only the preference of the people who actually turned out. A hypothetical example: a room has ten people and 6 of them hate spinach while 4 of the ten love spinach. A subset of the ten, say 5 of the ten chosen at random, is asked to state their preference for spinach. If the result is that the majority of the ‘electorate’ likes spinach that should be unsurprising as it possible that 3 or even 4 of the people chosen at random are spinach lovers. The results of that particular poll/election do not accurately reflect the opinion of the majority in the room. Perhaps people who vote more frequently are similar, more likely to own a home for example, than those who vote less frequently. Perhaps they are not a majority of the population, but are a majority of the electorate.
f. Turnout of the base of support in each camp is the primary decider of elections. Elections are not won by appealing to the swing voter or to the voter from the other side who might be inclined to switch teams, they are won by getting a few more of your own voters to make a trip to the polls. Historically, Democrats have higher turnout for presidential elections and Republicans have higher turnout for non-presidential elections. This is perhaps a reflection of the adage that a Democrat is a Republican without a mortgage. Older voters, voters with assets and with roots in a community, tend to be the most likely to vote from year to year, and these tend to skew Republican as a rule. Charismatic elections such as the presidential election attract the largest number of voters.
g. At least 80% of voters (and, in my opinion, virtually all voters) know exactly who they are voting for before the election even begins and any of the candidates have been put forth. In other words, if the corpse of Ronald Reagan were to be brought out his coffin and nominated for President by the Republican Party, he would be guaranteed to receive about 40% of the vote in the general election, while around 40% will oppose the corpse of Reagan or any candidate nominated by the opposing team. Look at the results of the vast majority of two-party races and you will observe that the losing candidate virtually always receives at a minimum 40% of the vote (occasionally in a particularly one-sided electorate, such as Utah or D.C. the loser gets far fewer votes, but that is a demographic quirk) A dead candidate actually won the Senate race in Missouri a few years back. Corollary: the myth that difficult and drawn-out primaries result in fractured parties that will result in many of the losing candidates’ supporters switching teams or not supporting the victorious primary candidate is just that: a myth. Cruz now supports Trump, Sanders now supports Clinton, Clinton supported Obama in 2008, all despite much media hype that the candidacies were doomed from intra-party rifts. And their supporters do exactly the same thing. The overwhelming majority of individual voters know exactly how they will be voting in a given election well before the election season begins; the only question is whether they actually vote. They may not like the eventual choice of their particular side of the divide, but they will support that candidate over the one from the other team unless there are extreme extenuating circumstances. Conservatives have lately taken to using the word binary to justify why they support a candidate they previously may have stated was the devil incarnate as in “this election is a binary choice and the devil incarnate is preferable to Hillary Clinton, who is worse than that!”
h. Charismatic politicians can have an impact but as a rule they succeed mainly in turning out more of their base than less charismatic candidates, not in changing the dynamic of elections by peeling off voters from the other side. Evidence: Western Pennsylvania and Trump support, which will likely be high. This will be reported as an example of the ability of Trump to pull in voters from the other party, but the fact is that Democrats have been losing voters day by day for more than a decade in Western Pennsylvania as demographic shifts result in an older, more conservative electorate. Trump is not responsible for this change, he is the beneficiary. The number of voters in Western Pennsylvania voting for the Republican candidate in the last three general elections has EXCEEDED the number of registered Republicans every time. This is an indication that some voters are local Democrats and National Republicans, much the same way conservative voters in South Boston typically vote for the Republican in general elections and yet all the local elected officials are Democrats. George Bush, John McCain, Mitt Romney, and Donald Trump are ostensibly different candidates with different personalities and approaches to governing yet all 4 received (or will receive) roughly the same number of votes in the election in Pennsylvania for example, between 2,650,000 and 2,800,000 votes. Similarly the Democratic candidates all received (or will receive) between 2,900,000 and 3,300,000 votes. The wider margin of votes in the case of the Democratic result is a reflection of the volatility of turnout of the democratic base.
i. Most, if not all, analysis of day-to-day events as transformational or dynamic-changing is misguided; it is a self-referential attempt to assess the dynamic of the dynamic of the race. In other words, there is a vast bubble of punditry that seeks to apply a daily filter to the developments in the campaign which ultimately, has little impact on the actual result. Debate performances, behavioral or temperament ‘issues’, and all the ‘developments’ during a campaign merely serve to reinforce the image of your candidate as the correct candidate and the opposition as the ‘bad’ candidate. All spin is ultimately designed to make the individual who was ‘wavering’ settle themselves into voting for their candidate with a relatively clean conscience. Use of the word ‘binary’ is the surest sign that a pundit is trying to justify why an individual should vote for the candidate despite misgivings: in a two-horse race you have to go with the ‘best’ one.
j. I will continue to watch the pundits on TV, read the daily developments of the campaign in the New York Times and the Washington Post, agonize over coughing fits or new racist comments, wonder in amazement at the gullibility of people who support Donald Trump, spend hours looking at the ever-changing forecasts on websites such as 538, and generally get less sleep than I might otherwise until election day, despite all of what I have said and despite the fact that I consider most of it a waste of time. People don’t change. I am a person. ergo…
14. More Articles
Articles I have seen since I wrote this that I find interesting and relevant. In other words, they support my hypothesis by explaining the election dynamic in a more productive manner (marked with $) or articles that do exactly what I have been complaining about in this essay but superficially seem to be serious (marked with &).
$ Why Trump Won’t Win Pennsylvania: The 2016 election offered the Republican Party its best chance to win Pennsylvania since 1988. Instead, GOP voters chose the one candidate who might ensure it remains blue for another twenty years. by Varad Mehta in The Federalist.
I wish I had seen this one earlier as it would have probably dissuaded me from spending so much time on data analysis for Pennsylvania and I could have gone to Florida instead.
& Election Update: Even A Small Post-Debate Bounce Could Make A Big Difference For Clinton, By Nate Silver at 538.com.
I am unhappy with the speculative nature of this article. It is less about data analysis and more about what would or should or could happen, with speculations about what upcoming data might show, which is not far off from conventional punditry, and disappointing to me.
$ Ohio, Long a Bellwether, Is Fading on the Electoral Map, by Jonathan Martin, New York Times
I could not have said it better myself.
$ Pennsylvania is Clinton’s Firewall: So she’s pouring troops and surrogates into the state in hopes of giving Trump no route to victory, by Albert R. Hunt, columnist for Bloomberg View, from the Pittsburgh Post-Gazette
I particularly like the bit about the Trump ‘negative’ strategy.
& Clinton Leads in Key Battlegrounds; Seen As Big Debate Winner, Public Policy Polling. This won even blatantly uses the words ‘Horse Race’
$ ‘Missing’ White Voters Could Elect Trump. But First They Need To Register., David Wasserman in 538. Wasserman consistently produces the most interesting articles on 538 and in the media in general. They appear to be based on the analysis of actual data.
$ Donald Trump’s Claims Haven’t Panned Out In Pennsylvania, by Dylan Purcell and Maria Panaritis, Philly.com