Why political analysis needs probability and history to address uncertainty

by Helen Thompson

Thompson oct12 trump

Political punditry has had a bad couple of years. Indeed, the failure to predict the dramatic political outcomes of a number of elections and referendums in 2016 and 2017 has itself become part of the political landscape. Nothing illustrates this fact better than the rise of Donald Trump to become American President.

The inability of the commentariat class in the United States to take Trump seriously when he entered the fight for the Republican nomination in June 2015 became political ammunition for Trump in his charge that America is run by cosy cartels looking after their own vested interests – whilst in the heartland of the country, so the argument goes, the American dream is dying and the insiders scarcely notice.

In some ways the blunders of many electoral pundits over the past few years to foresee political change is par for the course. Electoral shocks are nothing new. Notoriously the Chicago Daily Tribune published an edition on 3 November 1948 with the headline ‘Dewey defeats Truman’. Plenty of column ink was sunk in the 1970s and 1980 as to why a man like Ronald Reagan could not become President. Before Barack Obama came to represent the era of normal politics as a contrast to the disruption of Trump, he was once the young upstart with a largely empty political CV and trailing twenty to thirty points in national polling behind the supposedly anointed Democrat nominee, Hillary Clinton.

Contingencies have always changed the course of elections that looked like they were going one way, long before Francois Fillon’s bid for the French presidency was scuppered by allegations of paying his wife for fake employment, or James Comey re-opened the FBI investigation into Hillary Clinton’s email use because of material uncovered in the investigation of Anthony Wiener.

As such, electoral predictions not infrequently go awry because they are not sufficiently attuned to questions of probability. The occurrence of contingencies ensures there will be a number of possible outcomes to any election, even on the day itself, and recognising that fact should be a fundamental feature of electoral forecasting. Elections in most western democracies have for some time turned on relatively small margins, nowhere more so than in the United States where 33 years have passed since the losing candidate got less than 46 per cent of the vote (without the presence of a third candidate who left the winner with less than 50 per cent). Moreover, the four preceding presidential elections to 2016 were largely determined by the relative turnout of different demographic groups rather than any significant change in party coalitions.

Of course, grappling with the probability of what in the case of the American presidential election must produce one of two outcomes, except for the extreme case of a tie, is far from straightforward. Those election analysts who did offer probability-based predictions of the 2016 presidential election offered radically different chances of Trump winning. Even Nate Silver, the forecaster who in this respect was closest to recognising the real possibility that Trump could win (discussed in Glen O’Hara’s post in this series), had a model that showed an extraordinary degree of fluctuation about the probabilities of a predominantly binary outcome, even whilst a number of constants remained in place.

However, there is also another issue at work in the quality of recent political analysis: the general paucity of historical understanding in academic political science.

For too long much of political science has focused on the theoretical categorising of micro-level political phenomena in an atemporal manner, at the expense of meaningful explanation of the change in actual serious matter of the political and economic world. If we are going to explain significant political change of the kind that disrupts existing political orders, then we have to situate the political and economic world in historical time by which we can see what once underpinned these orders, and how these foundations erode.

In analysing political orders, international economic structures and their geo-political corollaries must be central. This imperative arises not because political life can be reduced to material issues; clearly, when values are contested and culture and identity are constitutive parts of existing political orders, it cannot. Rather, these economic and geo-political structures are the context in which the conflicts inherent to politics take place.

When in the early 1970s the Bretton Woods monetary order broke down, and the oil production of the United States began to fall as it rose outside the West, the domestic politics of the United States and Western Europe was transformed. Without the inflationary and geo-political fallout of this structural change, there would have been no shift to the political right in 1980s in the United States, Britain and Germany, and without the problems created by the attempts to manage European currencies outside Bretton Woods, there would have been no European monetary union.

The deep structural forces that are presently playing themselves out through Western politics have their origins in the slow breakdown of the post-1970s economic and geo-political order that became visible in 2008, and that have been amplified since by the diminution of American power and the rise of Russian influence in the Middle East, the rise of debt-financed shale oil production in the age of quantitative easing, and China’s turn to its one belt, one road economic strategy.

The absence of enough history in political science also risks us turning our back on sources of political prescience. Cyclical history has long been out of intellectual fashion but there is considerable insight into the times which we live from reflecting upon past temporal sequences of politics. If we were to learn from the last century of the Roman Republic, the rise of a man like Donald Trump to the American presidency would become a recognisable political phenomenon. As the political, economic and geopolitical predicaments for the Roman Republic mounted from the last third of the 2nd century BC, Roman politics produced a succession of outsider members of the oligarchic class seeking to win power in a factional struggle by politically mobilising the discontent of non-elite Roman citizens, starting with Gaius Marius.

Of course, history does not itself determine. Whatever the fate of republican Rome there is no inevitable civil war, or termination of democracy, at the end of our present political predicaments. But long history is a central means of understanding that intermittent crises of political order that produce political turbulence is an inherent part of the political world.

This post was originally published on speri.comment: the political economy blog at http://speri.dept.shef.ac.uk/2017/10/09/why-political-analysis-needs-probability-and-history-to-address-uncertainty/

About the author

Responsive image
Helen Thompson is Professor of Political Economy at the University of Cambridge and a regular panelist on the Talking Politics podcast. She also tweets @HelenHet20.
You May Also Like