Scots and Health inequality

In which I correlate census data about long-term illness and Scots speaking and try to find the hidden variable.

On the 17^th of June 2025 the Scottish parliament unanimously voted in favour of the Scottish Languages Bill, this legislation gave “official status” to Gaelic and the Scots language, and included provisions to support and promote these languages.

About a week later Public Health Scotland released the latest figures for Scottish cancer waiting lists, showing that these were the “worst on record”.

Some commentators on social media suggested that instead of spending time on Scots and Gaelic language provision, MSPs in Holyrood should have been sorting out these cancer waiting times.

It would be remiss of me not to ask – are cancer waiting times related to Scots language issues?

The census doesn’t include any data about prevalence of cancer, but does include several health related questions. We’ve previously noted that people who reported that they considered themselves able to speak or read Scots tend to report worse general health than the average and worse than non-Scots speakers.

The census contains data about “long-term illness” this might include cancer sufferers.

Here we see that on average 21.4% of the population (aged 3 and over) reported they had a long-term illness, among those with “No skills in Scots” 19.4% reported long-term illness. Those who reported any Scots skills had above average long-term illness – 24.9%, or 3.5% above average.

The worse suffering group here were those who could “Speak and Read Scots, but does not Write” with 28.4% reporting long-term illness, this is 7% above average.

It was suggested by social media commentators that this could be a class issue, as previously noted Scots speakers tend to have a lower standard of living than non-Scots speakers – lower paid jobs, living in deprived areas.

The 2022 census reports data in the National Statistics Socio-economic Classification (NS-SeC) which is a proxy for “class”, splitting respondants into seventeen categories. We can check the data and control for class of respondents.

And it reveals something odd about Class

We have already seen that people with tangible Scots skills make up a smaller proportion of the higher paid classes, but only from 38% to 27%, so less than 12% variation or the difference between two in eight people and three in eight people

L1: Employers in large establishments          27.0% with tangible Scots skills
L2: Higher managerial and admin occupations    29.5% with tangible Scots
...
L13: Routine occupations                       38.4% with tangible Scots
L14.1: Never worked                            32.5% with tangible Scots
L14.2: Long-term unemployed                    37.8% with tangible Scots

When we control for this socio-economic class, Scots speakers still report more long-term illness than non-Scots speakers.

When controlling for NS-SeC class people with tangible Scots language skills are 3.0% more likely to report long-term illness than the general population.

If we look at specifically those who reported they could speak and read Scots, but not write it, this group reports 5.3% more than average long-term illness

This suggests that even Scots speakers with high standards of living suffer more long-term illness than might be expected.

Is it geographic?

Commentators online further suggested that it could be something to do with geographic distribution, there might be more Scots speakers in areas with poorer health services.

We can use the census data to control for this.

When controlling for geographic location people with tangible Scots language skills are 3.8% more likely to report long-term illness than the general population.

If we look at specifically those who reported they could speak and read Scots, but not write it, this group reports 6.6% more than average long-term illness when controlled for geographic location.

Is it age related?

Perhaps the hidden variable is age, Scots speakers are on average older than non-Scots speakers, and older people are more likely to suffer from long-term health conditions.

Here we see the same effect, even when trying to control for age. Older people in general have a higher rate of long-term illness, but within each age group, people with tangible Scots skills are 2.2% more likely to have long-term illness than average.

And again those who specifically reported that they considered themselves able to speak and read, but not write Scots, have the worse long-term illness figures, with about 3.5% above average.

At this point we might note that between controlling for occupational Class, Geography and Age, the geographic variable has the strongest inequality and the age variable has the weakest inequality.

Perhaps its education

I’m just flailing, grasping at straws, the hidden variable is elusive.

In previous Substack articles I’ve laboured the point that Scots speakers don’t get the same educational outcomes as non-Scots speakers. So perhaps this would lead to a correlation with long-term illness, and if we control for the highest level of education using the census data then the Scots disparity will fall away.

But no. Here we see that for each category of highest level of qualification, Scots speakers worse than average rates of long-term illness (3.0% worse than average), and the people who can only speak and read Scots (but not write it) have the worst rates (6.1% worse than average)

But what about cancer?

Whilst we have used the census “long-term illness” rates, this article was kicked off with cancer waiting times.

Public Health Scotland reports data on cancer incidence and prevalence. But it as a google search that led me to MacMillan Cancer Support and their data broken down by local authorities.

If we look at the 20-year prevalence rate of cancer for each local authority and plot it against the proportion who reported they considered themselves able to speak, read and write Scots, we get the following graph.

This suggests that generally council areas with a high proportion of Scots speakers have lower cancer prevalence rates than council areas with lower proportions of Scots speakers.

The Pearson’s correlation coefficient it R = -0.53 which suggests a strong negative correlation.

This finding somewhat confounds the research above. Scots speakers have a higher than average rate of long-term illness, but a lower than average cancer prevalence 20-year rate - the long-term illness must therefore be disproportionately non-cancerous.

A brief literary diversion

At this point in the narrative I should mentioned that I just finished reading Denise Mina’s “The Less Dead”, its a novel where an adopted pregnant doctor tries to get in touch with her bio-family and finds her birth mother had been murdered and there’s some kind of conspiracy, a possible serial killer and stalking thing going on.

It was a compelling read, I rattled through it in about three sittings. Literally couldn’t stop myself reading it.

I have a secret plan to commission a handful of famous contemporary Scottish writers to write some short stories wholly in Scots, for an anthology to promote the language, and I reckoned I should read up on how much Scots these writers currently use in their predominantly English writing.

In “The Less Dead” the only Scots speaking characters were sex workers and corrupt police officers. The main character, her friends and the straight cops all spoke properly, spoke standard English.

Now its a realistic and moving story, but its a work of a fiction - not like Mina’s earlier book “The Long Drop” which was based on a true story. I mean this book is set in Glasgow which is a real place, and sex workers really were murdered there in the 1980s and 90s. But the characters and there manner of speech, that’s all on Denise Mina and her writing craft, rather than the fault of the real life sex workers and bent cops.

As we’ve seen Scots speakers make up a significant proportion of people in all walks of life, all age groups, all geographic regions.

And here is the epiphany…

Shitty service

Last year there was a newspaper story about the lecturers at Edinburgh University having to get special training to be told not to take the piss out of people with Scottish accents.

It wasn’t the fault of the Scots-speaking students, no character weakness on their part, this was down to the lecturers being shitty people, behaving in a shitty manner.

And perhaps this explains the long-term illness of Scots speakers. Its not about the people themselves, some genetic flaws or lifestyle failing, perhaps its down to shitty service from the Scottish healthcare providers.

This might explain why even the high social grade Scots speakers have higher incidence of long-term illness, because they use the same health service as the lower social grade Scots speakers, the same GPs, doctors and nurses - and end up receiving just a few percent worse service than non-Scots speakers.

It might be that medical professionals disproportionately struggle to understand Scots speakers compared to English speakers, the accent is too strong. It might be that the Scots speakers struggle to understand the medical professionals’ instructions. It might be that one party of the other chooses to ignore the explanations or instructions of the other.

I don’t know, but maybe some proper research on the linguistic friction between Scots speakers and the healthcare system would be illuminating.

Revisiting cancer

I have to confess, I’m not a healthcare journalist, I’m not a writer or a reporter.

I am a stock controller and production engineer. I work in factories, looking at manufacturing data, trying to figure out what’s going on and then fixing what I can.

So I don’t have innate practical experience or understanding of cancer issues or the difference between prevalence and incidence. A few paragraphs ago when I looked at the cancer prevalence 20-year rate, I mistakenly thought it meant that people had less cancer in Scots speaking regions.

The 20-year prevalence rate could also be interpreted as the 20-year survival rate. If the rate of people being diagnosed with cancer is pretty consistent across Scotland, and perhaps cancer doesn’t really care what language people speak, then the lower 20-year rates in Scots-speaking areas suggests that those areas aren’t as good at keeping people with cancer alive for 20-years or more, compared to areas with lower proportions of Scots speakers.

Which kind of explains the strong correlation between the rates over the longer 20-year time scale.

Just about to publish

This is a serious subject and I’m a little reluctant to publish this article.

What if I’m wrong?

I didn’t fake the census results, I didn’t meticulously fill in 1.5 million individual census forms with carefully planned correlations. The people of Scotland filled in those forms, I’m just trying to make sense of the data.

And I didn’t diagnose thousands of people with cancer at different rates in different regions. That was the health care system.

I didn’t write crime thrillers using personal prejudice to inform my views of what languages different characters should speak. Nor did I take the piss out of my students accents at university, that was the lecturers of Edinburgh university.

What I’ve brought to the discussion is pulling these threads together, suggesting how they could be linked in a manner that makes logical sense.

Is that enough?