Did the Supreme Court's Article 50 decision matter?
People (mostly Remainers) were more likely to support parliamentary involvement after the decision
Tomorrow the Prime Minister will notify the European Union of the UK's intention to leave.
This satisfies the deadline set out in the Prime Minister's party conference speech of last year, when she promised notification by the end of March.
Since the timing of Article 50 notification hasn't changed, and since Parliament placed no restrictions on that notification, it's reasonable to ask whether the Supreme Court's judgment mattered for anyone other than public lawyers.
In a draft article, I argue that the Supreme Court changed people's minds about whether or not Parliament should be involved in triggering Article 50. Most of this effect was concentrated amongst Remainers — people who were already favourably disposed to a role for Parliament — but the case shows the potential for the Supreme Court to affect public opinion.
The survey and results
I asked YouGov to survey people before and after the Supreme Court's decision. I asked them the following question:
"In order to leave the European Union, the UK needs to activate Article 50. This will formally notify the EU of the UK’s intention to leave. Some people have argued that Parliament should be involved in this decision. Others have argued that the government does not need to involve Parliament. Which of the following comes closest to your view?
a) The government should have the power to activate Article 50, rather than Parliament
b) Parliament should have the power to activate Article 50, rather than
We can compare answers to this question in the pre-decision wave (fielded 12th to 16th January, or eight to twelve days before the decision) and in the post-decision wave (fielded 27th to 30th January, three to six days after the decision).
Here's what that looks like across different groups of respondents:
Support for parliamentary involvement, although it remained a minority, moved from 41% to 47%. Support was always higher amongst Remain voters than Leave voters, and amongst those who heard about the decision compared to those who didn't.
That last distinction is important. We should only expect the Supreme Court to have an effect on people who could recall how the Supreme Court decided — people who didn't know how the Supreme Court decided, or who thought the Supreme Court had ruled in favour of the government needn't be affected in any way.
Using coarsened exact matching, I investigate the differences between those who heard about the decision and those who didn't. Matching does what the name suggests — it matches someone who heard about the decision with someone who didn't, but who is alike in many other respects. Here, I matched people on the basis of
- what they thought about parliamentary involvement before the decision
- how familiar they were with the UKSC, and how much they trusted it
- their attention to politics
- how they voted in the EU referendum and the 2015 general election
- their preferred daily newspaper
When I do this, some of the estimates change, and we have to be careful how we talk about the results. Here, I show four estimates of the effect. The FSATT is the effect that hearing about the decision had on those for whom we could find matches; the ATT is the attempt to generalise that to everyone in the sample. "Default" is the result you get out of the box, "User" is my attempt to coarsen variables manually to achieve more useful matches.
This shows that our best estimates of the effect of hearing about the Supreme Court decision range from increasing support for parliament by 4 percentage points (Default/FSATT) to around nine percentage points (Default / ATT). My preferred estimates are the figure from the user-coarsening — so either a five or seven percentage point boost for parliament being involved.
The great thing about matching is that it enables you to average treatment effects over different groups. Here, I show the effect averaged across voters depending on how they voted in the EU referendum:
This figure shows that our best guess of the effect of the decision upon Leavers was almost exactly zero. Conversely, we can be very confident that the effect on Remainers was positive, and somewhere north of ten percentage points, taking support for parliamentary involvement from already high levels to an overwhelming majority (roughly 66% to 78%).
How do we explain this? One interpretation is that whilst people generally trust judges, those who heard about the decision recognized its political implications, and revised their views about the court after hearing how it decided. Imagine someone who follows this reasoning:
- When a trusted source endorses something, that's usually a good reason for endorsing that same position
- The Supreme Court is a trusted source
- The Supreme Court ruled that Parliament must have a say in Article 50
- There are good reasons for thinking that Parliament must have a say in Article 50
That reasoning, for some people, conflicts with another chain of reasoning
- Brexit should be supported
- Parliament having a say would increase the chances of Brexit not happening
- There are good reasons for thinking that Parliament should not have a say
Where a conflict of this kind exists, we can either deal with the cognitive dissonance, or minimize it by abandoning one of the premises above.
In this case, Leavers chose to trust the court less (and Remainers to trust it more). The following graph shows levels of trust before and after the decision, according to vote behaviour in the referendum. This data is taken from a subset of respondents who, in the second wave, were explicitly told how the court had decided, and were only then asked how much they trusted the court.
Leave voters (in blue) were already less trusting of the court — and they became less so after they heard how the court had ruled.
I'm sure that public opinion regarding parliamentary involvement has now changed again — but this example shows the potential of the UK Supreme Court to change minds, at least for those prepared to listen. This example may not generalise to other cases — there's a reason I shelled out to test the effects on public opinion in this case, and not a hum-drum tax case. But given some occasionally mixed findings for the US Supreme Court, and the total lack of any evidence for the UK Supreme Court, this finding does matter.