On the morning after, the aggregate story of the American elections of 2016 was one of surprise. Professional pundits – not to mention the pollsters, those masters of disaggregation – had been rather strikingly wrong. So, one of the major products of the elections of 2016 was a major research project for these professional disaggregators. But can those who treat elections as aggregate phenomena – as patterned wins and losses across time – fit these results into an ongoing pattern? Within this pattern, what looks familiar and what looks anomalous? Or, said the other way around, when does a recognizable “modern world” begin, and what are its contours? And in the end, even if 2016 proves to have more similarities than differences to some (surely not all) of its predecessors, what would have to happen in order for its anomalies to become the shape of a successor world, making 2016 look like the beginning of something seriously new? Those are the questions that motivate this paper.
In their aftermath, the American elections of 2016 provided fresh grist for those whom we might call the “disaggregators”, those analysts who break the result down into its individual pieces. What did various social groups do? Which issues had major traction? Which states conformed to – or deviated from – ongoing expectations? And “why” was this so in every case? Many of the authors in this issue of The Forum are masters of this approach. Yet there is another way to analyze these results, one treating them instead as a collective whole. In effect, they become an N of one: one presidential outcome, one Senate outcome, one House outcome, and, most especially in the USA, one separationist outcome. 
Enter the aggregators, whose questions are inevitably different. Where does the election of 2016 fit into the long flow of American electoral politics? How do its major pieces – elections to the Presidency, the Senate, and the House – slot these individual institutions into that longer flow? More to the practical point, how does their composite result refer back to earlier incarnations, as well as forward to the context for policy-making that this composite outcome more or less automatically produces? Even the richest comparisons across time cannot predict the political or governmental future. But they can interpret the latest result. And they can constrain alternative futures.
For the aggregators, then, three things are required. Their analysis must begin with the evolution of the separate pieces of the 2016 election. It must then put them back together: what we need, after all, is some perspective on the current incarnation of the separation of powers, the real institutional context for policy-making in American national government. Yet the ultimate purpose of any such long-running analysis is to offer perspective on two future-oriented questions:
Does the election appear as part and parcel of an ongoing electoral era, and if so, how should this be described?
And in light of the answer to question #1, what are the alternative futures that become more and less plausible in the aftermath of this election?
All too much of the coverage of the election(s) of 2016 was focused on the Presidency. In a separationist system, this is an immediate distortion. Still, the presidential outcome is often thought to be both a harbinger of the political future and a key practical propulsion toward it. Moreover, this particular outcome – the victory of Donald Trump and the defeat of Hillary Clinton – has been treated as the largest single surprise of 2016. So it may on both historical and interpretive grounds be best to use the presidency to begin building a picture of 2016 in the longest run.
Figure 1A offers the full record of presidential outcomes from 1828 through 2016. Arraying results before 1828 as if they were lineal predecessors of the modern world requires too much supposition and extrapolation. But from 1828 onward, with the crystallization of pro- and anti-Jackson factions, the enterprise becomes more direct and plausible.  The picture that emerges is curiously dual, featuring stretches of partisan domination coupled with insistent reminders that these have rarely been absolute. There is one 24-year stretch, one 20-year stretch, and one 16-year stretch within this 188 years. No others rise to the equivalent of two presidential re-elections in succession. Only two others, 1828–1840 and 1980–1992, offer even a presidency-and-a-half.
Figure 1B, on the other hand, treating single terms as a kind of anomaly, pushes back toward stretches of partisan dominance. Single-term presidencies can produce important policy developments. Yet they cannot well serve as evidence for a new and continuing partisan era. With this particular brand of “smoothing,” then, the American electoral world begins to look different:
There was an extended period of Democratic predominance from the formation of the two-party system to the Civil War, 1828–1860. It was in some sense the undoing of this domination that brought on that horrific conflict.
This was followed by the longest stretch of partisan predominance in all of American history, under the Republicans from 1860 to 1912. Were the single presidency of Woodrow Wilson to be further smoothed away, this period would extend from 1860 to 1932.
The Democrats then sustained their largest unbroken return to predominance by way of the extended period from 1932 to 1952. In turn, if the single presidency of Dwight Eisenhower were additionally smoothed away, this period would run from 1932 to 1968.
Finally, the Republicans produced their largest unbroken return to predominance from 1968 to 1992. Though after that, no defensible smoothing produces a further stretch of one-party dominance.
So, there are two grand stories within this historical sweep: the recurrent ability of individual candidates to intrude into partisan predominance, but the insistent reassertion of this predominance across most of presidential time since 1828. With a focus on the most recent election of 2016, however, the outstanding take-away is that these smoothed and extended periods appear to end in 1992. Seen presidentially, that year is the putative beginning of a “modern world,” one whose central characteristic is partisan balance and party competition. The election of Hilary Clinton would have disrupted that story-line and raised questions about its persistence. The election of Donald Trump instead fits neatly into (and extends) this modern era.
The House of Representatives, the other body that is nationally elective across this entire period, manages to bring further practical possibilities and theoretical challenges. In principle, on its own terms and courtesy of a 2-year term, the House allows for more deviations from the dominant partisan outcome, whatever that outcome is. Yet smoothed in even the most elementary fashion, the House instead shows an impressive record of partisan continuity, greater than that of the Presidency or (as we shall see) the Senate. This raises the inescapable question, both historical and theoretical, of whether it is the House or the Presidency that better captured mass partisan alignment in all the years before it was possible to survey party identification.
Figure 2A, absent any smoothing, provides only the slightest evidence of those potentially greater changes in party control for the House as compared to the Presidency, 23 versus 21, despite its two- (rather than four-) year term. Moreover, still without smoothing, the House offers a longer stretch of unbroken party control than the Presidency has ever generated, by way of the 40 years from 1954 to 1994. Yet when the search is for partisan domination or party competition, some elementary smoothing seems even more essential with the House: the argument that one 2-year interim is evidence of a major and lasting partisan change seems implausible on its face.
When this is done, Figure 2B throws up even more striking results. Dropping only one-term deviations where the Congress in question was only 2 years long, the House was actually less prone to partisan shifts than the Presidency: 11 such shifts for the Presidency, only seven for the House. Moreover, that record-setting stretch of partisan dominance, 1954–1994, is now partisan predominance on a gigantic scale: 1930–1994. At the same time, isolation of any plausible “modern era” becomes automatic. The only available candidate is the years after that longest unbroken stretch of partisan predominance, namely 1994 to the present. As with the Presidency, so with the House, the elections of 2016 simply slot into and extend this modern period.
Those are the two great, longest-running, individual stories of American electoral politics at the national level. Yet it is time to stop treating them individually and instead elicit the composite story, already breaking through almost on its own. In this, partisan electoral accounts of the Presidency and the House nearly cry out for a “separationist” presentation, that is, for treating them jointly, as integral components of the constitutional separation of powers. To that end, Figure 3A takes the unsmoothed results from Figures 1 and 2, puts them together in the same graphic, and codes them as unified Democratic (red), unified Republican (blue), or split (green).
One immediate and insistent finding jumps out: their joint “choppiness” has contributed a great deal of split partisan control to American national government across time. Still with only the Presidency and the House in the picture, the results of American national elections have divided rather neatly into thirds – one-third unified Democratic, one-third unified Republican, and one-third split – with the modal outcome being split rather than unified partisan control. That said, this picture is otherwise so misleading that the analysis must shift directly to a different electoral development, present but masked in Figure 3A.
For this, it is necessary to begin by noting that there are actually two distinct kinds of “split partisan control” when the focus is the joint electoral evolution of the Presidency and the House, distinct types that must be disentangled in the interests of meaningful interpretation:
In the first type, split control registers an impending shift in partisan owner-ship of the Presidency. Having a shorter term, the House could always move earlier. Yet when the voting public was in the process of shifting preferences for the Presidency too, appearances of split partisan control were effectively just reflections of the unified control to follow.
In the second type, by contrast, instances of split control were not part of a simple and direct transition. For whatever reason – and the reasons themselves varied – the public chose split control in its own right. Unified control in the hands of the opposite party was not about to happen, and if analysts of the time could not have known this, analysts with the benefit of hindsight certainly can.
Accordingly, Figure 3B removes those split controls that proved in practice to be simple transitions. In response, the story becomes immediately different. Now, there are only four such elections in the entire period from 1828 to 1946, and three of these – 1874, 1876, and 1878 – are bogus: they are the result of the “corrupt bargain” that kept the Presidency in Republican hands in 1876 but terminated Reconstruction.  In a real sense, 1854 was the sole incarnation during this extended historical stretch. By contrast, the mid-term election of 2006 was the only election from 1946 onward where the old pattern of off-year change in partisan control of the House rolled on to produce on-year change in partisan control of the Presidency – compared to 11 such instances in earlier times. All others were instead split partisan control in its own right.
The result was cascading. From 1946 onward, split control became a familiar outcome. From 1954 onward, it became the most common result. From 1980 onward, it was overwhelmingly dominant: it had become the default outcome. That is the modern world in which we all live, though the jury must remain out on whether 2016 has extended or challenged this particular aspect of the modern era. Said the other way around, 21 of the 35 elections from 1946 produced split partisan control of the Presidency and the House, compared to only 15 of 59 during all the preceding years – where only one of the latter appeared to be a conscious preference for split control.
From those analyses, a modern electoral world begins to define itself. Yet the effort to extract it has ignored one crucial aspect of the separationist framework, while introducing another which has yet to receive the necessary attention. The missing element of the separation of powers is, of course, the Senate, while the effort to tease out genuine split results has implicitly introduced the question of the distinction, if any, between on-year (presidential) and off-year (purely congressional) contests. So the remaining questions for a proper analytic background are two: How does the Senate fit into this framework? And does the off-year/on-year distinction alter, refine, or simply map onto the resulting chronicle?
Figure 4A suggests a few lesser twists from adding the Senate to this overall chronicle. Yet the main story is that the Senate has largely tracked the behavior of the House across time. Like the House, the Senate does have distinctive structural characteristics that might alter the prospects for change over time in partisan control. For the Senate, this possibility is rooted in a long, 6-year term, longer than those of either the House or the Presidency. In principle, this might make the Senate less responsive to shifting electoral currents. In practice, it always means that only one-third of the Senate is up for election at any given point, so that a strange sample of partisan contests might generate more rather than less change opportunities.
But in fact, none of that appears to matter much in the long run. The House generates ever so slightly more shifts in partisan balance than the Senate: 23 versus 21 in the full period. Figure 4A and B confirm that this imbalance was greater in the early years, lesser – actually reversed – in the modern period, from 1994 onward. Yet the bedrock point is that the question of split partisan control is not much affected by adding the Senate to the institutional mix, with two exceptions that surface after modest smoothing. In passing, the complete chronicle, by not arguing for major differences between the House and the Senate, is an implicit argument that the XVIIth Amendment to the US Constitution, making the Senate directly elected, did not change its partisan place within the separation of powers in any major way.
Where two noteworthy exceptions to this common tracking do emerge, however, is when the chronicle is smoothed, dropping one-term interruptions from the House story and less-than-one-term interruptions from its Senate counterpart. Figure 4C and D do this, and those two exceptions emerge:
The first comes in the middle of that long period beginning with the Civil War, from 1860 to 1912, which brought a succession of new – and initially Republican – states into the Union. With automatic entitlements to two Senators but only one House member, this flow in effect insulated the Senate from the Democratic perturbations that were roiling the House. 
The other difference, previously acknowledged but more insistent in the smoothed picture, sets off the modern world, where – conversely – the Senate became clearly more changeable than the House. Beginning in 1980 and not faltering through the current moment, partisan control of the Senate changes much more than partisan control of the House.
Though note that both major exceptions do fit within a further generalization that says that while the Senate approximates the House more than it does the Presidency, the Senate has disproportionately joined the Presidency against the House when House and Senate were split for anything more than a single House term. This latter fact probably makes it unsurprising that there are three times when a one-term split between House and Senate – 1858, 1910, and 1930 – meant that the House was foretelling a change in partisan control of the Presidency at the next election, one that would sweep the Senate along as well.
On the other hand, this richer institutional arrangement does not suggest anything specific about the implicit prior question, about potential differences in on-year versus off-year elections, most especially their propensity to generate split partisan control. So that question must be addressed in its own right. Figure 5 does this for genuine splits only (as at Figure 3B), again eliminating elections in which split control is only the early arrival of a new partisan majority that will reach into the Presidency and restore unified control at its earliest opportunity. If these “lagging indicator years” were in the analysis, off-years would almost have to produce more split control than on-years, since the Presidency cannot shift with the House during off-years.
Yet with these inherently misleading years removed, there is still no real difference between the two types of election, when the focus is split versus unified control.  The previous temporal division remains. The years before 1946 produce almost no such splits – and would produce even fewer if we removed the “corrupt bargain” underpinning their appearance in 1874, 1876, and 1878. By contrast, the years from 1946 onward feature a majority of split outcomes. With nearly no instances of split control, the early period cannot in principle distinguish as between on-year and off-year elections. But now, with split outcomes as the dominant result, the years from 1946 onward are still nearly indistinguishable as between on-year and off-year contests.
That is a long chronicle of partisan fortunes. Its contents are sufficiently diverse that the analyst always needs to know which institutions are the focus, and where we are in historical time. Perhaps curiously, there is nevertheless a modern world that answers this latter question in a very straightforward way. Much of the preceding suggests that since 1994, we have lived in an electorally distinctive period, into which 2016 fits rather neatly. Seen one way, what distinguishes this period is its partisan volatility, especially by comparison to the years immediately before but actually by comparison to all those years before the very earliest ones in our chronicle.
Seen the other way, however, we nevertheless make this extreme volatility conform to a simple template. Consider just the usual atheoretical narrative for this period:
At the start of Bill Clinton in 1992, there was unified partisan control of the elective institutions of American national government, in Democratic hands;
Under Bill Clinton, there was also split partisan control of American national government, with the Presidency Democratic and Congress Republican;
At the start George W. Bush in 2000, there was unified partisan control of those elective institutions, but in Republican hands this time;
Under George W. Bush, there was also split partisan control in the direction opposite to the Clinton incarnation, with the Presidency Republican but Congress Democratic
At the start of Barack Obama in 2008, there was again unified partisan control of American national government, back in Democratic hands this time;
Under Barack Obama, there was subsequently split partisan control in the direction opposite to the Bush incarnation, with the Presidency Democratic and Congress Republican;
And at the start of Donald Trump, there is once more unified partisan control, of a duration unknown as this is written.
Yet at a slightly higher level of abstraction, some simple further generalizations can gather and systematize all of these individual contests:
Unified partisan control characterized the national elections producing each new President of the modern period;
Moreover, every one of these individuals – at least through Clinton, Bush, and Obama – were to be re-elected;
Yet unified control was always transient, followed by split partisan control in which Congress swung away from the party of the President, even though he was going to be re-elected.
On the one hand, this crash came earlier for the Democrats than for the Republican. On the other hand, that fact made their personal re-elections more impressive.
We have quite literally never seen this pattern to that degree at any point in American political history, and this singularity can be underlined by reverting to a less anecdotal form of presentation. Accordingly, Figure 6 attempts to capture the full complexity of aggregate options across time, while still confining the analysis to genuine splits only: solid red bars are unified Democratic control; solid blue bars are unified Republican control; striped red bars are a Democratic Congress with a Republican President; striped blue bars are a Republican Congress with a Democratic President; dotted red bars are a Democratic President with control of one house of Congress; and dotted blue bars are a Republican President, likewise with control of one house of Congress.
Smoothed in the fashion of previous figures, Figure 6 does call our attention to the mottled period after the Civil War, 1874–1894, when the Democrats could pick up the House but neither the Senate nor the Presidency. Yet the modern period still jumps out as most distinctive, offering all six variants of partisan control of American national government in one concentrated stretch. Beginning in 1990, all six of these options were to be realized in the space of fourteen congressional elections, yet none of them was to recur more than three times.
There is much for the disaggregators to learn about where this period came from and why it continues. Yet one crude aggregate outline that fits comfortably with it would say a) that the partisan balance in American society became very close, b) that the two parties within it became more homogeneous and simultaneously farther apart, c) that this made shifts in party control of the various pieces of American national government – all of them – easier than ever to accomplish, while d) making each subsequent iteration reliably unattractive to the general public, whose majority opinion was inescapably out of alignment with both parties, especially when they featured unified control.
Seen in this manner, 2016 is profoundly a reflection of the nature of modern American politics. In that sense, it was hardly a surprise at all. Given the opportunity to follow what has been the pattern of that politics for 25 years and counting, the American electorate followed it to the letter. Such a view has the virtue of taking the analyst away from the peculiarities of the candidates: the results of 2016 did not reflect the personal idiosyncrasies of either Donald Trump or Hilary Clinton. Such a view likewise has the virtue of taking the analyst away from campaign strategies: Hillary Clinton did not lose it because of incorrect strategic decisions, any more than Donald Trump won it by eschewing conventional campaign approaches. Instead and in the end, voters turned back to what they have been doing for more than a generation.
The now-superficial surprises that accompanied this behavior do cry out for efforts to understand the very real disjunction between disaggregated results – as tapped by the polls – and aggregate outcomes, especially state by state. Any and all clarifications will benefit our understanding of American electoral politics more generally. But let us close by leaving that task to those who specialize in it and ask instead what the next 4 years would have looked like had the polls been accurate, and then (more realistically!) what they can look like – this entire chronicle is sufficient warning that no one should say what they will look like – given the actual result.
Before the actual vote, what the polls were suggesting was a Democratic Presidency, a closely balanced Senate, and a Republican House. In that environment, the lead players for policy-making would almost inevitably have been President Hillary Clinton and Speaker Paul Ryan. In only slightly stylized terms, there appear to be three summary outcomes that could have followed:
Each main player decides that a simple “blocking strategy” is optimal, that is, try to see that no policy is made that would not be endorsed by their supporters while blaming the other side for the absence of any ultimate product.
Both main players, with personal histories as political leaders who have always aspired to make policy, decide to address major policy issues, despite the tensions that this will create with their own committed bases.
Both main players decide to seek to address major policy issues, but one or the other – or both – discover that they cannot bring their supporters along with them in a manner sufficient to bargain with their opposite number.
In the aftermath of the 2016 elections, all such strategic alternatives are just historical curiosities. What clearly did result in aggregate terms was unified partisan control of the elective institutions of American national government. Yet this was “unified control” of a very peculiar sort:
The House of Representatives looked much as it had before the election, indeed, much as it has since the dawn of the modern electoral era in 1994. This is the Republican Party of ideological activists.
The Senate, however, was left with a relatively narrow Republican edge – constituted from many Republican Senators who were dissident from their own party on one or more particular issues.
And the President was a thoroughly unorthodox Republican, standing to the right of his party on some matters, closer to the Democrats on others, but clearly not stereotypical of the Republican Party to be found in either the House or the Senate.
That left the pattern of policy-making for the next few years looking surprisingly fluid – a genuine challenge to systematic expectations. There is a set of major policy conflicts that surfaced during the 2016 election campaign, along with another set that will intrude more or less in their own right in the near future. The former might include trade, healthcare, taxation, infrastructure, fiscal repatriation, domestic security, and/or foreign relations. The latter – those items that may intrude on their own – will begin with the debt ceiling, but might well extend to the structure of Medicare and/or Social Security. As a collection, these policy realms align those “three Republican parties” differently, while leaving substantial policy choice – and strategic dilemmas – to the minority Democrats as well.
But if this peculiar version of “unified control” was simultaneously diagnostic of the modern world of American elections, can we not at least specify alternative electoral outcomes that would either conform to it or bring it to an end? In simple terms, yes of course. A loss of either House of Congress 2 years from now, or re-election of the President 4 years from now, would both be fully consistent with – mechanically parallel to – the electoral world stretching forward from 1992. Both would guarantee that it lasted until 2020, at a minimum. Conversely, the defeat of the President 4 years from now, or the extension of unified Republican control through his complete second term, would deviate from the pattern.
One-term presidencies have been common, albeit not typical, in American electoral history. Eight-year Congresses have been common, albeit again not typical, in that same electoral history. Rapid successions of all the available partisan mixes have been seen in the long run only in 1992–2016. If tomorrow is to be like yesterday, they appear to be here to stay. If tomorrow is to bring the dawn of a new electoral era, we at least know what to look for.
©2016 Walter de Gruyter GmbH, Berlin/Boston