“Traditional British values” in political science

by Chris Bertram on June 30, 2020

Yesterday, one of those reports was released purporting to reveal some things about British political attitudes. The take-home was that the public were closer to Labour than the Tories on the economic dimension but that things were reversed when it came to social attitudes, with voters being more authoritarian and traditional than their representatives and more closely aligned with the Tories. This, coupled with the claim that the social values dimension is gaining in importance compared to the economic one, looms large in some “political science” explanations of Brexit and Tory success.

Looking at the report, I noticed an odd thing: one of the questions was about “traditional British values” and whether respondents thought young people should respect them more. I imagined, naively that there must be a list somewhere of what these values are, given that they are purportedly what voters have opinions about. But no. Respondents are expected to interpret the question for themselves, so if a person thinks Britain is traditionally open and tolerant and thinks this has gone into reverse in recent years and says “yes”, their response will nevertheless count as a score for authoritarianism. I don’t know how likely my hypothetical case is, but it is at least possible, particularly, perhaps, from a disappointed immigrant who had been sold on a particular image of Britain.

There is, by the way, an official list of British values. It occurs in the case of the British goverment’s “Prevent” strategy, which is supposed to combat extremism. According to that policy, the British values are “democracy”, “the rule of law”, “individual liberty and mutual respect”, and “Tolerance of different faith and beliefs”. If you run an educational establishment, you have specific duties to encourage these. (A nursery for under-5s I’m acquainted with had a little official explainer on the wall, telling its charges that the “rule of law” is about “following the rules”.) I suppose, hypothetically, that a respondent with these values in mind, and thinking of the Brexiteer and Johnson record on them – plans for voter suppression, illegal prorogation of Parliament, frequent abuse of executive power in immigration, racist and Islamophobic diatribes by the man at the top – might regret their demise and say “yes” but be coded as “authoritarian”. Perhaps not so likely, I admit.

Chris Hanretty (@chrishanretty) tells me on twitter that this same set of questions has been in use in British politics since the early 1990s. Researchers back then, allowing respondents to interpret the question however they chose, found, neverthless that it was useful in positioning those same respondents on a libertarian-authoritarian axis. Still, there is something a bit fishy here. Responses to a question that has no determinate meaning are found to corellate with some quantum that isn’t itself independently observable but which the model postulates is “out there” on the basis of responses to this and other questions. So the responses are, supposedly, what one would expect, given other features of the model: coherence, in other words. But there’s obviously going to an inductive problem here: just because such a good fit between responses and presumed attitudes was observed when the model was originally devised in the early 90s, why do political scientists believe that correlation continues to exist in the vastly changed political and social environment of the 2010s?

There’s a reason why this problem of induction isn’t going to get pressed too hard within this kind of social science though. It is because sticking to the same questions, and implicitly the same model, is seen as a feature, not a bug. It is because sticking to the same questions generates a set of answers that can be compared to the answers reported in all the previous surveys. But unfortunately this can only be reliable if the questions have a consistent meaning for respondents over time. It isn’t hard to see that this often won’t be so. Years ago, when I sat on a committee looking at research proposals (details changed to protect etc etc …) we were looking at a project that proposed to ask teenagers about drug use. Having teenagers myself at the time, I could see that many of the questions would be incomprehensible or misleading to them. I queried this, only to be met with the answer that the psychologists had to stick to the same questions, because their predecessors had been asking the questions since 1959 and they needed to be able to compare …. See the problem?

One suggestion I heard on twitter was that the “British values” question tracks attitudes to authority and hierarchy. Those answering yes might be presumed to be more deferential. Well perhaps. But there’s another question in the “economic values” section, which seems not entirely unrelated, but which serves only to position people on that orthogonal “economic” scale. It asks whether there’s one law for the rich and another for the poor. Are people who think that society is a racket for the benefit of the toffs but who also think that traditional British values should be respected more by the young really best thought of as deferential to those above them? Maybe at least some of those who think it is a scam want a return to better standards in public life and say “yes” to British values for that reason? But people are complicated, and sometimes hypocritical, aren’t they?



Hidari 06.30.20 at 9:44 am

I really hope that this will not be perceived as ‘off-topic’ but as some wag on Twitter pointed out, you will really learn everything you need to know about ‘British values’ from this TikTok video, which is simply one of the best things I have seen in years.


Watch with the sound up.

I doubt Bojo would approve though (or, more worryingly, perhaps he would).


John Quiggin 06.30.20 at 10:25 am

Another example of the consistency problem: the Australian Bureau of Statistics continues to report the group aged 15-64 as the “working age population” even though people under 18 are strongly discouraged from entering the fulltime workforce and the pension age has been increased to 67. The result is to overstate concerns about the “aging population” (which, as I regularly point out, is aging 1 year at a time, as people always have).’


J-D 06.30.20 at 11:56 am

Even when the questions were first used, obviously there was going to be a good fit between responses to them and a model originally devised on the basis of those very responses. Even if the questions had a clear and consistent meaning to respondents synchronically (even setting aside diachronic consistency), some justification is still needed for choosing those particular questions as the basis for the analysis in preference to other questions with equally clear and consistent meanings. On careful examination, it generally emerges that what you’ve got in cases like this is that somebody decided that there were some issues which were of special importance to political divisions or political groupings, devised questions related to those issues, and then found that those questions could indeed be used to define political divisions or political groupings: in other words, the evidence that they were right was produced by first assuming that they were right. Two of the five questions about social values in this example specifically refer to the criminal law. Are attitudes to criminal law politically important? Well, possibly, but since the way this particular inquiry embeds that assumption in its methodology means that it can provide no test of it. On the other hand, none of the questions refer explicitly (for example) to religious beliefs. The way this inquiry embeds the assumption that religion (for example) is not politically significant into its methodology prevents it from testing whether it is more or less significant than (for example) attitudes to the criminal law.

It’s more likely than not that MPs and their voters have divergent views about some questions and not about others, but which are which? Maybe divergence in relation to the questions asked in this case is unimportant to the voters, or to the MPs, because they place more importance on their agreement about other questions, not asked by these researchers. But that’s the kind of possibility that this kind of study is methodologically blocked from exploring.


notGoodenough 07.01.20 at 11:28 am

It certainly is a bit puzzling as to what is believed to be gained from asking such ambiguous questions. One further wonders why it would not be more sensible to actually design a questionnaire which elucidates more useful information. It brings to mind the story regarding Howard Moskowitz and Prego.

Or, as the old axiom goes: garbage in, garbage out…

Though it would be nice if someone did clarify what “traditional British values” are – I have been asking for some years now!


bianca steele 07.01.20 at 5:53 pm

When an instrument doesn’t measure P (or doesn’t measure P well, or was intended to measure P and doesn’t any longer), an expected response, for someone who thinks it’s inportant to measure P, is to improve or replace the instrument. P is presumably important because of some action the government, etc., might take. It seems reasonable to conclude either that there is no such action, or that those who will take the action know P without the instrument.

That seems rather conspiratorial. Maybe it’s better for everyone if we conclude that the instrument actually does measure P.

The fact is, after all, that the original measures of “authoritarianism” were culturally and ideologically biased along lines that we’re now able to see we’re not entirely comfortable with. That they continue to be open to criticism is also just a fact; they probably always will be.


N. N. 07.01.20 at 9:55 pm

“But people are complicated, and sometimes hypocritical, aren’t they?”

Well, as Raymond Geuss wrote in the opening pages of Philosophy and Real Politics:

“People often have no determinate beliefs at all about a variety of subjects; they often don’t know what they want or why they did something; even when they know or claim to know what they want, they can often give no coherent account of why exactly they want what they claim to want; they often have no idea which portions of their systems of beliefs and desires – to the extent to which they have determinate beliefs and desires – are ‘ethical principles’ and which are (mere empirical) ‘interests.’ This is not simply an epistemic failing, and also not something that one could in principle remedy, but a pervasive ‘inherent’ feature in human life. Although this fundamental indeterminacy is a phenomenon almost everyone confronts and recognises in his or her own case all the time, for a variety of reasons we are remarkably resistant to accepting it as a general feature of the way in which we should best think about our social life, but we are wrong to try to evade it. […] [P]eople’s beliefs, values, desires, moral conceptions, etc., are usually half-baked (in every sense), are almost certain to be both indeterminate and, to the extent to which they are determinate, grossly inconsistent in any but the most local, highly formalised contexts, and are constantly changing.”


SusanC 07.03.20 at 8:28 pm

The phrase has something of the air of a dog-whistle about it.

There’s a certain sense in using dog-whistles in survey questions, because of the well-known effect that respondents tend to give socially acceptable answers. i.e. if you flat out ask, “Do you consider yourself to be a racist?” then most people are going to answer no. On the other hand, some suitable dog-whistle may serve to classify your respondents.

On the other hand, “traditional British values” also has a “motherhood and apple-pie” flavour to it, where everyone agrees to it without necessarily agreeing on what it actually means. This is less useful in a survey question.

The use in the context of “Prevent” has always struck me as self contradictory. “Tolerance of different faith and beliefs” is defined to be a traditional British value we’re supposed to uphold. Fine. But this does not sit well with a Prevent-type mechanism for reporting students to the police if they hold different values. A possible conclusion is that any teacher who suggests reporting a student to Prevent for what they say in class is themself a domestic extremist by the definition in the regs, and so ought to reported to Prevent instead of the student…


SusanC 07.03.20 at 8:39 pm

On the “stickiness” of bad metrics, I will confess to at least one paper in which we use a somewhat debatable metric because we wanted to do a comparison with prior work.

I also have on my “to do, if I ever get the time” list a paper where I take a certain metric that’s only being used because it has verified inter-rater reliability (so two raters will agree on what score to give an experimental participant, even if what they agree on isn’t really what you’ld like) and redo the reliability study for a newer and hopefully better metric. Someday. If I ever have time…


J-D 07.04.20 at 8:29 am

Well, as Raymond Geuss wrote in the opening pages of Philosophy and Real Politics:

Obviously both correct and significant, but I didn’t know anybody had put it in a book. Thank you for drawing it to my attention.


SusanC 07.04.20 at 10:08 am

A possible (but weak) defense of using venerable but flawed metrics: The referees of your paper (and the eventual readers) may suspect you of metric-gaming to get the result you want by choice of metric. By using the flawed metric from someone else’s paper, you can defend yourself against cjharges of metric-gaming. Of course, the author of the paper that originally introduced the metric might have themselves been metric-gaming to get the result they wanted, but that was a different experimental hypothesis, and if you’re lucky the earlier metric-gaming is orthogonal to the different thing you’re currently investigating.

If you’re unlucky, on the other hand, these things are statistically correlated and everybody is being led astray by a bad metric.


In the particular case of “British Values” the current rise of populism (of both the Corbyn style Brexity Left and the Boris Johnson style Brexty right) presents a challenge.

“Are you a populist?” is a reasonable thing for a survey to try to ask. But respect for rule of law, tolerance of multiple faiths etc. — which are legislatively defined to be British values — are something the decidedly non-populist faction like Tony Blair or Keir Starmer will appeal to. But “Traditional British values” is the kind of dog whistle you’re likely to hear from, at least, the populists of the right. So as a question, its really unclear which faction it will pick up.

Whether these actually are traditional British values, as opposed to being declared to be so by definition in the context of the Prevent program, is debatable.

I would gesture in the direction of various public statues that are currently being torn down by the Black Lives Matter movement as evidence that the traditional values might have included something else. At least for the people who thought it was a good idea to put up a statue of William of Orange, Edward Colston. etc.

William of Orange is well known for his support for a pluralistic society, in which people of all faiths, Catholic and Protestant, can live together in harmony :-)

Comments on this entry are closed.