I can’t find evidence for the luxury beliefs hypothesis

The luxury beliefs hypothesis has achieved considerable fame. The work of Rob Henderson, a writer with an impressive backstory, the hypothesis postulates a relationship between social justice values and status,

‘In the past, upper-class Americans used to display their social status with luxury goods. Today, they do it with luxury beliefs… We feel pressure to display our status in new ways. This is why fashionable clothing always changes. But as trendy clothes and other products become more accessible and affordable, there is increasingly less status attached to luxury goods. The upper classes have found a clever solution to this problem: luxury beliefs. These are ideas and opinions that confer status on the rich at very little cost, while taking a toll on the lower class.’

The theory is important. Social scientists have long explained social behaviour in Bourdieusian terms, i.e. with reference to the underpinning role of status, and ideology may well have such a function. A few years ago, Henderson did a fascinating podcast on this Substack and was extremely nice. Recently, others have elaborated the hypothesis. This week, Yascha Mounk and Ruxandra Teslo had an interesting discussion about the theory, debating the motivations of elites (among other things).   

But five years after the formulation of the hypothesis, further refinement might be beside the point; primarily, we need empirical data on its performance. Commendably, one can test the theory with various methods. There are more specific ways of doing this (see below), yet one would expect to observe signs of the hypothesis in the general relationship between social justice values and socio-economic status. Admittedly, the hypothesis is nuanced; would one expect those with social justice values to have higher status, reflecting luxury beliefs, or lower/declining status, reflecting the need for luxury beliefs? Notwithstanding these ambiguities, data should reveal some sign.   

In a forthcoming book*, we examine the relationship between social justice values – these are distinct from liberal values and involve an emphasis on identity and direct action and an extended conception of harm – and socioeconomic status in five datasets. These include the American National Election Study (ANES), British Election Study (BES), US General Social Survey (GSS) and two of our own surveys of US and UK voters which use the Progressive Values Scale (PVS). The datasets contain diverse measures of socioeconomic status (including personal economic situation, socioeconomic status, evaluation of the general economic situation, vulnerability to unemployment and poverty, job satisfaction, career standing, childhood economic precarity, childhood socioeconomic status and expectation that the young generation will have a better life than their parents) and various measures of social justice values**.

To my surprise, there was little sign of a relationship between these variables and social justice values. Few variables achieved statistical significance and those that did went in conflicting directions. General economic evaluation was an occasional exception; in ANES and BES, negative economic evaluations are associated with social justice values. However, controlling for government approval (at the time of the surveys, conservatives were in office) and political values reduces this effect. As one might expect, negative economic evaluations are downstream from these variables.

Having undertaken this investigation, I am sceptical that a meaningful relationship exists in datasets. Admittedly, this might be non-linear – for example, it could be that the relationship holds among a subset of high-education and low-income voters, many versions of the theory focusing on these groups – but my attempts to uncover consistent evidence of such an association have also been fruitless. If others are able to demonstrate such a relationship, I would be pleased to hear from them.

Alternative tests reach similar conclusions. In a new paper, the UCD economist Margaret Samahita uses a signaling game to test whether low-status participants identify with such beliefs to gain credit among high-status participants. She finds that ‘luxury beliefs are not strongly associated with status: they are only perceived to signal college attendance and negatively correlate with income and perceived income; and ii) there is no evidence of signaling using these beliefs in a (close to anonymous) online setting’.

Such results are consistent with other trends in the social sciences. In literature on radical right populism, economic interpretations have an underwhelming record and, in recent decades, the influence of the economic cleavage has declined. In other areas which I research, economic variables underperform and, generally, I am sceptical of their influence.

Caveats are necessary. Absence of evidence is not (as the saying goes) evidence of absence. A link between status and ideology is plausible and, in well-designed qualitative studies, proof of the luxury beliefs hypothesis might emerge. Quantitative tests of traditional Bourdieusian hypotheses could be much more extensive (for exceptions, see here) and critics must beware of selective demands for rigour.    

More broadly, such findings do not disprove the existence of social justice ideology, as distinct from liberalism. Rather, they suggest that the ideology is associated with alternative variables and, in our book, we find evidence of a relationship with demographic factors such as female gender and age.

Yet such findings raise questions for those who advocate the luxury beliefs hypothesis. In 2019, the hypothesis was fresh and coverage of Henderson was important. I have been glad to see the hypothesis gain influence; Bourdieusian explanations should not only be wielded against conservatives.

But years have passed and hypotheses need evidence. Admittedly, the lack of empirical investigation reflects conditions within universities – as I have written before, academics tend to be reluctant to examine social justice ideology – yet the onus is also on heterodox scholars, i.e. those scholars who are critical of established academia. Increasingly, such researchers are organized in institutions such as the University of Austin and investigation of their own arguments should be a priority.

If the community cannot find compelling evidence for the luxury beliefs hypothesis or a more plausible version, they should not use it.

*Incidentally, this accounts for my lack of recent Substack posts. Sorry! The book is nearly finished and, very soon, I shall write much more regularly and reveal details of the book.

**On request, I will share my code for the (publicly available) ANES, BES and GSS datasets but cannot do so (yet) for our own surveys of US and UK voters.

---

If you enjoyed reading this, do think about subscribing! Subscription is free – all it means is that you’ll receive a weekly email. But every new subscriber makes me very happy 😊 😊 😊