Faith and demographics: unveiling seemingly paradoxical trends in the UK
The Policy Institute’s analysis of the 2022 World Value Survey reveals something puzzling about Britain: while Gen Z are the generation most likely to say they have no religion, they are also the most likely to say they believe in Hell. Is this due to a rise in people with unorthodox combinations of beliefs, shunning organised religion but believing in eternal damnation? That may be an intuitive solution – but it’s not the answer. Much of this apparent paradox can be explained by a simple demographic trend, and the key is British Muslims. The pattern we see in Gen Z emerges not because of changes in the combinations of beliefs held by Britons, but changes in the composition of who Britons are.
To explain the seemingly-paradoxical rise in belief in Hell, the first thing to understand is that although the proportion of religious people in the UK has been falling across the generations, the proportion of Muslims has been rising, and it is at its highest in Gen Z. You can see this pattern in the charts below, which show the proportion of each generation belonging to each faith in the WVS dataset and the 2021 UK Census.
Fig. 1. While Britons have got less religious in general, the proportion of Muslims has increased.
So while most of Gen Z have no religion, they are the generation with the largest Muslim minority[1]. This is important because Muslims are much more devout than other groups in terms of their belief in Hell, as well as Heaven, the Afterlife and God. As Figure 2 shows, virtually all Muslims in the WVS believe in all four, which is not the case for Christians.
Fig. 2. Muslims are much likelier than Christians to believe in Hell.
An important caveat to this analysis is the small number of Muslims in the sample – 19 in Gen X, 31 among Millennials and 12 in Gen Z. This means there is potential for considerable sampling error. But given the consistently high levels of belief among Muslims across the generations and all four topics, and the size of the distance to Christians, it seems safe to conclude that Muslims do indeed have higher levels of belief.
Putting it all together, the increasing proportion of Muslims together with their high levels of belief in Hell means that Muslims on their own can explain why Gen Z has higher levels of belief in Hell than Millennials and Gen X. As Figure 3 shows, when Muslims are excluded from the calculation, Gen Z’s belief in Hell drops by 10 points and is actually lower than the two preceding generations. In fact, it is lower than preceding generations for belief in Heaven, the Afterlife, and God too, with the non-Muslim portion of Gen Z returning to Baby Boomer levels of belief.
Fig 3. Gen Z are less likely to believe in Hell than Gen X and Millennials when Muslims are excluded
So, the increasing proportion of Muslims can account for much of the seemingly-paradoxical pattern of beliefs among Gen Z. The growth of this epistemically-devout portion of the population allows belief in Hell to rise even as more young people today than ever before say they have no religion.
Can compositional changes across the generations like this also explain a notable trend visible in Figure 3, namely, that levels of belief in Hell, Heaven, and the Afterlife have increased among Christians over the generations, at least from Baby Boomers onwards? One reason for this could be the large number of immigrants the UK has welcomed from devoutly Christian countries like Jamaica, Nigeria, and Poland, among others, since the Baby Boomer era (1946-64).
As Figure 4 shows, later generations do include a higher proportion of people from immigrant backgrounds (a person was classified as having an immigrant background if either they described themselves as an immigrant or described both their parents as an immigrant), though this has actually levelled off in Gen Z, which is probably only because the Gen Z cohort is still too young, with birth years between 1997 and 2012, for many Gen Z adults to have immigrated to the UK from abroad, compared to preceding generations.
Fig. 4. Later generations contain a higher proportion of people from immigrant backgrounds.
And, as Figure 5 shows, Christians from immigrant backgrounds tend to have higher levels of belief in Hell, Heaven, the Afterlife and God.
Fig 5. Christians from immigrant backgrounds are more likely to believe in Hell, Heaven, the Afterlife, and God.
Putting it all together, immigration has seemingly boosted levels of belief in God, Heaven, Hell, and the Afterlife among Christians in recent generations – as Figure 6 shows, when those from immigrant backgrounds are removed from the calculations, levels of belief for all four topics drop among Gen Z, and to a lesser extent, Millennials.
Fig. 6. Christians from immigrant backgrounds have boosted levels of belief in Hell, Heaven, the Afterlife and God among younger Christians.
But Figure 6 also shows that even among non-immigrant Christians, belief in Hell, Heaven and the Afterlife has increased since the Baby Boomers. Why might this be? The answer may be linked to another puzzling trend you may have spotted in Figure 3 – belief in Heaven and the Afterlife has increased in recent generations among those with no religion, at least up until Gen Z.
Changing social norms may provide a unifying explanation for both trends. Suppose there are two factors which contribute to whether a person identifies as Christian: firstly, sincere belief in Christian teaching, including about the existence of Hell, Heaven, the Afterlife, and God, and secondly, the strength of the social norm to be a Christian. These factors might act additively meaning a person with low-to-moderate levels of belief in Christian teaching might identify as Christian if the norm to do so is strong, but would identify as having no religion if the norm is weak.
Crucially, the norm for identifying as a Christian is likely to be stronger for older generations than for younger generations, since it has been more common throughout the lifetime of older people for British people to be Christian. Therefore, people with weak-to-moderate beliefs in Christian teaching would be less likely to identify as Christians in younger generations than older generations. This has two implications. First, it means that younger generations of Christians contain fewer people with low levels of beliefs compared to older generations, and so the total level of belief among younger Christians is higher – whereas in older generations, both the uncertain and the devout might have classified themselves as Christian, now only a devout core remains, relatively speaking. The second implication is that younger generations of the non-religious contain more people with some level of religious belief, and so the total level of belief among the younger non-religious people has increased – whereas in older generations, those with weak-to-moderate Christian beliefs might identify as Christians despite their uncertainties, considering it normative to do so, in younger generations, these people may identify as non-religious despite harbouring some religious beliefs, because it is more unusual to identify as Christian.
This is of course speculation, but it presents a plausible explanation for these two puzzling trends. In fact, a formal model of this scenario predicts both trends. I simulate five generations of agents, with each generation containing 1000 agents. I label the generations 1-5, going from oldest (1) to youngest (5). Every agent has a degree of belief in Christian teaching, which is sampled from a normal distribution with mean 5 and standard deviation 1. Each agent also is influenced by the norm towards Christianity – to reflect that perceived and actual norm strength will vary within populations, the norm strength is also sampled from a normal distribution with a standard deviation of 1, but the mean of the distribution depends on the generation. For the oldest generation, the mean norm strength is 5, but this decreases by 1 for each successive generation, reflecting the waning strength of the norm. Agents are asked two questions. Firstly, “Are you a Christian, or do you have no religion?”, to which they answer “Christian” if the sum of their degree of belief and perceived norm strength equals or exceeds 10, and “No religion” if it is below 10. Secondly, “Do you believe in X?”, where X is a generic aspect of Christian teaching, such as Hell or Heaven, to which they answer “Yes” if their degree of belief equals or exceeds 5, and “No” if it is below.
Figure 7 shows the percentage in each generation of the agents who identify as Christian and those with No religion that believe in X, the aspect of Christian teaching. As it shows, despite the mean degree of belief remaining the same in every generation, the proportion of believers goes up in both groups over the generations, as the norm becomes weaker.
Fig. 7. Results of simulation
Of course, this account doesn’t explain why there are some differences between the four topics – belief in God has gone down for Christians over the generations, for example, and among the non-religious, levels of belief for all four topics appear to be falling again from Millennials to Gen Z. While we should remain cautious about how much differences like this just reflect sampling and measurement errors, this heterogeneity leaves something interesting for further analyses to explore.
Technical Note. People were classified into generations by their birth year according to the following scheme: Before 1946 – Pre-War, 1946-1964 – Baby Boomers, 1964-1980 – Gen X, 1981-1996 – Millennials, 1997-2012 – Gen Z. All percentages are weighted percentages of non-missing responses, using the ‘S017’ weight. Analysis script available upon request.
[1] Quite why the Census reports a much higher number of Christians and fewer people with no religion in each generation than the WVS is not clear. To speculate on the reason for this discrepancy, the fact that the Census is often filled out by one person on behalf of their family members could lead to, for example, parents wrongly assuming their children would describe themselves as Christian. However, this doesn’t explain why there are large differences even for the older generations. Another possibility is that the Census allows people to describe themselves as simply ‘Christian’, whereas the WVS in fact does not, and instead only allows people to identify with a specific denomination, like ‘Roman Catholic’, ‘Protestant’, or ‘Orthodox’. Therefore, people who might describe themselves culturally as Christian, but do not actively participate in Christian worship or are unsure of which denomination to select, might choose Christian in the Census but ‘No denomination’ in the WVS, and hence have their religion classified as None. Nevertheless, both sources demonstrate the key trend of a rise in the proportion of Muslims across the generations.