Last week was “immigration week” on Sky News – but, to be fair, most weeks are immigration week across large parts of the media. In many ways, this is only right – immigration is an issue of huge national importance to the public, in the top three issues in our regular polling for many years, behind only the economy and unemployment.
And this has been driven by real changes in immigrant numbers, not a phantom concern: we were relatively unconcerned before numbers increased in the late 1990s as the chart below shows. (The images enlarge when you click).
But the nature of media focus is almost unique, driven to such a large extent by what the public think should happen - and how at odds that is from what has actually happened. Only welfare comes close to having such a public opinion-driven agenda, and even here there are nothing like the number and variety of measures of attitudes.
For many months now we have been conducting a detailed review of attitudes to immigration and how they relate to reality, for Unbound Philanthropy – and many times I’ve wished we hadn’t started it. The volume and variety of public opinion polling on immigration is frightening.
It’s not surprising then that new polls often tell us very little – or that they are often presented as the revelation of a new direction in public opinion when they’re clearly not. Take the poll Sky News has released to support its “Immigration UK” project. Their headline finding of 67% wanting drastic action to reduce immigration is deeply predictable, but presented as a sign of a “fundamental shift” where the public are “increasingly saying no more”.
Nothing could be further from the truth: while we weren’t raising it as a top issue in the late 1980s and early 1990s, almost exactly the same two-thirds were saying there were too many immigrants or that numbers should be reduced.
And there are many more traps in the gaps between opinion, reality and our interpretation - within an overall crystal clear picture that the large majority of people want overall immigration reduced. Here are just a few.
Extent and nature of immigration
We hugely overestimate the extent of immigration. When asked to guess at the proportion they make up of the population, the mean is 31% (median 26%), compared with an actual proportion of 13% (14% if you include the upper estimate of illegal immigration). Now this is not new, or unique to the UK: in fact all other countries we’ve seen measuring this find the same, although we tend to overestimate more than most.
We also have a very wrong idea of who immigrants are – our “imagined immigration” as Scott Blinder has termed it, is way off. When asked who comes to mind, we are much more likely to think of asylum seekers and refugees and much less likely to think of students. As the chart below shows, we get it the wrong way round: asylum-seekers are in fact the smallest of the main immigrant groups, students the largest.
It is no coincidence that our overestimate is of a group we’re worried about and we underestimate a group we’re relatively positive about. This is sometimes taken as evidence that we need to educate the public: if we knew the real scale and nature of immigration, we’d worry less.
Of course, this will be true to some extent – but there are other explanations. We need to recognise that cause and effect in these type of estimation questions run both ways. We have “motivated reasoning” when answering them: we don’t just have “accuracy goals” in mind, we also have “directional goals”. Whether consciously or not, we may be trying to express our concern about the scale of immigration or particular groups of migrants as much as we are concerned to get the right answer.
So it is arguable that our worry may cause our overestimation and focus on more problematic groups as much as the other way around – social psychologists call this “emotional innumeracy”.
The important practical point here is that “myth-busting” exercises are likely to have limited impact on people’s concerns. But equally, there is a significant danger in accepting that our inaccurate picture of immigration is fine because it partly reflects our concerns and emotional reactions: both are partial.
From looking across independent reviews of the economic, labour market and fiscal impacts of immigration, it seems fairly clear the aggregate effects are not huge (at a per capita level at least), but in general terms, the net fiscal impact is probably the most positive (if only because immigrants have tended to be younger and more economically active than the native population).
But public opinion is most negative on the fiscal impact: we’re most likely to be worried about the benefit and public service impact of immigrants from these sorts of topics.
But of course, this is also entirely understandable. First, people will not have a whole system perspective on the fiscal contribution of immigrants: the tax contribution of immigrants is invisible, but their use of services and receipt of benefits will be visible to many directly and especially through the media.
Second, people will not see supply of services as elastic: more money per head may come in as a result of immigration, but local services will not be seen to scale up to reflect the increased numbers, at least in the short-term. Whether rational or not, it is difficult for any policy-maker to win this argument.
Surveys are too broad
An important limitation of the large majority of survey data on attitudes to “immigration” is that they attempt to sum up views under a single and undefined label, leaving each respondent to answer on the basis of their own unstated conception of who “immigrants” are - which, as we have seen, will often be inaccurate.
A good illustration of this is seen in the fact that a majority of us believe that “immigrants” both take jobs from native workers and create jobs in the community. This is not because people are stupid, they will just have had a different mental image of immigrants when answering the questions.
And we do have very varied views. As Rob Ford’s analysis of the 2011 British Social Attitudes Survey illustrates, when migrants were described as professionals, net support for settlement in the UK is very positive, regardless of the migrants’ origin or motive for migrating. When migrants were described as unskilled labourers, net support was negative, in each combination with region and motive.
There is some concern about an immigration “arms race”, where political parties out-do each other in their toughness on immigration. Some of this may be due to a further perception gap: that people just do not know what’s already been put in place by this and previous governments, as the table below from Lord Ashcroft’s recent polling shows.
Overall, then, any government or political party has real problems on immigration: concern is high, views ill-informed, government is not trusted, they have limited policy levers they can pull, and the areas in their control are the ones people are least concerned about (such as students and highly-skilled non-EU workers).
These issues are not that new or unique: Gary Freeman developed the “policy gap hypothesis” in 1994 to explain why immigration policy was consistently less restrictive than the public seem to demand, across a range of countries. But the scale of the disconnect between opinion and perceptions of policy in the UK is so huge that, while we can be critical of the quality of the debate and particular policies, it is difficult to blame the government for trying to get closer to where most people are. The real challenge is to do this without reinforcing ill-informed prejudices and shoring up misperceptions.
Hard Evidence is a series of articles in which academics use research evidence to tackle the trickiest public policy questions
Bobby Duffy is the managing director of the Ipsos MORI Social Research Institute and Visiting Senior Research Fellow at King's College London. Ipsos MORI receives funding from a wide range of government departments, NGOs and other funding bodies. This research was funded by Unbound Philanthropy to review published data on attitudes to immigration.