by Peter Franklin
Thursday, 20
May 2021
Chart
13:27

Revealed: The most important political trend of our time

New research shows just how important the education divide is in western politics
by Peter Franklin

Want to understand the last fifty years of democratic politics? Then read this new paper by Amory Gethin, Clara Martínez-Toledano and Thomas Piketty.

It investigates a familiar hypothesis, which is that the most ‘highly’ educated voters are increasingly trending Left, with a corresponding shift of non-graduates to the parties of the Right. 

The remarkable thing about the research is just how far it goes in showing that this shift is not only true, but overwhelmingly more important than any other comparable development in voter preferences. 

For a start, the researchers use data drawn from 21 western nations — enabling them to demonstrate that the phenomena they describe are truly international. In a paper bursting with charts, two in particular stand out. 

The first shows the political divides of the 1970s — when, in almost every country, the parties of Right attracted higher levels of support from voters with the highest levels of income and education (with the parties of the Left doing the diametrical opposite).

Fast forward to the present day (or at least to the 2010s) and a remarkable shift has taken place. Though the parties of the Right continue to have an advantage among the voters with the highest incomes, it is the parties of the Left that now attract more support among voters with the highest level of education:

These charts have been doing the rounds on political Twitter — no doubt, because they tell such a dramatic story. It is worth pointing out a few details though.

Firstly, for both income and education levels, these charts are comparing the top 10% of the population versus everyone else. Thus these shifts are taking place among the elite and aren’t just a side-effect of expanding higher education to other parts of the population. 

Secondly, it’s important not to misinterpret the research as showing no shift towards the Right among poorer voters and no shift away from the Right among the richest ones. Graduates tend to earn more than non-graduates and therefore any move away from the Right in the graduate population will be reflected in the voting preferences of higher earners too. However, the researchers statistically control for this “mechanical” factor. As they say themselves, removing this control “displays a stronger decline in the influence of income on the vote.”

Thirdly, the shift shown by this research hasn’t occurred all-of-a-sudden. The charts above are for the 1970s and 2010s, but the paper also has charts for the intervening decades. Taken together, these show that the great shift has been underway across that entire period. Indeed, it’s truly extraordinary that it wasn’t spotted sooner. 

Looking back, we can see that the Tories before Brexit were wasting their time trying to appeal to young graduate professionals. In reality, a very different set of voters were waiting to drop into their lap — as Boris Johnson was to prove in 2019 and again this year. 

I don’t want to say that people like David Cameron and George Osborne were on the ‘wrong side of history’… but then again, they were on the wrong side of history. 

Join the discussion


  • There is a general misconception, particularly in the US, that credentials are a signifier of intelligence.

  • The theory ‘feels’ right of course; however, presenting a complex, multi-point graph to, in essence, display a, “look, simple proof! Everything is completely opposite to 50 years ago!” automatically makes me suspicious. A minor thing (but it has a big effect on the impact of the graphs) the data size scales are slightly different from one graph to the next. The income data range is reduced by 50%, the Education data range by 20%. What does that mean? By comparing two datasets together as if they had the same spread skews the story, magnifies it and makes it more dramatic. And we all know how these graphs will be used…Somewhere where nuance is unknown and dramatic simplicity is sovereign.
    To present a data representation as ‘proof’ of a simple, direct and huge societal change (:D) should not require the viewer to perform mental gymnastics to understand it. If it does, it has failed.

  • Good comment. To fully understand this study we’d have to review the original report and complete data set. We’d also require enough math knowledge to analyze the data and study design for ourselves. Most people don’t have that type of skill.
    I’m reminded of the endless debates involving science and covid. Do masks work? What is the real mortality rate for covid? Is growth of the infection exponential or does it follow a Gompertz (S-shaped) curve? Ultimately, we have to rely on someone, be it a scientist or a journalist with a science background, to answer these questions for us because it’s too hard to review all the primary data and figure out the answers ourselves.
    So who do we trust? Now, when presented with a study report like the one mentioned in this article, my first question is who conducted the study and what is their agenda? Does the study come from what appears to be an impartial research group, or is it tied to a political think tank pursuing a specific agenda. The answer to that question usually answers the question of whether or not I pay any attention to the study. Sad that science has become so politicized.

  • To get involved in the discussion and stay up to date, become a registered user.

    It's simple, quick and free.

    Sign me up