Putting aside other parts of the @Policy_Exchange report out today, I'm going to focus on this Figure 2 (on page 52) for a mo. It attempts to say academics have become more left-wing over time, using % of academic party ID at certain time points. 1/n

Let's look at this in relation to vote share in nearby elections, and the spatial positions of parties from this figure at https://doi.org/10.1111/1467-923X.12638. In this thread, Fig2 refers to the policy exchange one. 2/n

1964: Fig2 puts Lab ID at 47%, Con at 35% and LDs around the 18% mark. At that election, Labour got 44% of the vote and Con 43% -- it was close -- with LDs at 11% of vote share. 3/n

Spatially, all parties are close ish too, all leaning toward the left, from -35 to -10 on scale above. So in this election, parties were close (left leaning), academic support was close, and the election result was close. 4/n

1976: Here we will use the 1974o election for comparisons. In Fig2, Lab & Con support in academia have both declined to 40% and 29% respectively, with LDs rising to around 27%. At the election, Labour got 39% of the vote, Con 35% and LDs 18%, not quite as close as last time. 5/n

Spatially, though, there is more of a gap with Lab around -30 still, but LDs at 0 and Con at +10. Here, the election was less close but the ID gap is wider. 6/n

The 29% of academics supporting Con might be more right-wing than the previous 35%; the 27% LDs might be strictly centrist, and the 39% Labour may have become a bit more left-wing. 7/n

1989: Here we use the ‘87 election. In Fig2, Labour and LDs are almost joint at around the 37% academia party ID point, whereas Con have fallen to 18%. In the election, the Conservatives won with 42% of the vote share, Lab had 31% and LDs 22%, so this one wasn’t close. 8/n

Spatially, it’s also not close. Con are at their peak right-leaning at over +30, Lab are at about -15 and LDs at -10
So the majority of academics almost jointly support the two centre-left parties at this time, and the right-wing party loses favour. 9/n

2017: So we get to the final time point of Fig2, where labour have a majority of academic support at 53%, Con remains at 18% and LDs are just below at say 17%. Greens and Other are around the 5% mark (but let's not get into that yet). 10/n

The election was close between the two main parties, with Con at 42%, Lab at 40% & the LDs at 7%. LDs have lost public favour, Lab have gained some. Spatially though,Cons are the centrist party placing around 0 on the scale, Lab come back to around -30 & LDs are in the -20s. 11/n

Now the right-wing party has moved left (to the centre) but kept the same amount of academic support, LDs have moved left but lost support, whereas Lab have become slightly more left and gained support in academia. 12/n

But, this mirrors the public support where LDs have lost out to Lab’s gain and Con have remained the same.

Does this really say that academics are more left-wing than in 1964? 13/n

In 1964, 47%+18% = 65% of academics supporting parties at -35 to -20 (but *all* parties are left leaning at this time)
In 2017, 53%+17% = 70% supporting parties at -30 to -25 (and Con are in the centre)

Doesn’t appear to be a huge jump to me? If at all? 14/n

There are other things going on here, such as new parties emerging, Con and UKIP combined in Fig2, and all the missing data points where support undoubtedly bounced around. Nvm the fact that left-right is probably not a helpful dimension on its own anymore. 15/n

BUT, the point is that we cannot use these figures to say that academia has become more left wing over time.

1. The whole party system has changed.

2. These are distinct points in time (with govt policy on academia changing). 16/n

3. These are different generations of academics, and people in general!

4. A Conservative today is more centrist than a LibDem (if you use the comparison they are trying to).

5. Principally, to me, this Policy Exchange report is just bad science. /end