For the regular boozer it is a source of great comfort: the fat pile of studies that say a daily tipple is better for a longer life than avoiding alcohol completely.
But a new analysis challenges the thinking and blames the rosy message on flawed research that compares drinkers with people who are sick and sober.
Scientists in Canada delved into 107 published studies on people’s drinking habits and how long they lived. In most cases, they found that drinkers were compared with people who abstained or consumed very little alcohol, without taking into account that some had cut down or quit through ill health.
People have been adjusting for that for decades. It’s nonsense.
“It’s been a propaganda coup for the alcohol industry to propose that moderate use of their product lengthens people’s lives,” said Dr Tim Stockwell, first author on the study and a scientist at the Canadian Institute for Substance Use Research at the University of Victoria.
But then Stockwell is a loon.
Keep an eye open for Chris Snowdon’s rebuttal of this. He’s the expert on this sort of tosh.
Keith Richard…
I know it’s anecdotal, and will not count, however twice I have had acquaintances, who frequented the same pub as myself. Both were diagnosed with bowel cancer, both given around 3-6 months to live. They were both in their late 60’s. They continued to drink commenting ” we are going out as we see fit”, I cannot in all honesty say it was the drink that kept them going, but both lived over 4 and a half years before shrugging off their mortal coil.
I think I read in bad science that is very hard to compare moderate drinkers with non drinkers because their life styles are so different.
If you think about why people don’t drink – religious reasons, very dedicated athletes etc I can understand why.
I don’t drink coffee I could easily imagine that there are people who have a similar lifestyle as me but drink coffee instead of tea, I can’t imagine a non drinker like that though.
Everyone dies. Life is 100% fatal.
So you might as well try and make the time you spend in this vale of tears as pleasant as possible.
At the more extreme end of the “alcohol problems” reporting… ISTR some years ago reading a study done by the University Hospital of Chicago on “Heavy Drinkers” – they’d essentially been studying “winos” – that indicated that it was only possible to identify actual physical harm done to the body once the consumption was north of 100 units per week.
21??!! Amateurs!
Seems like the science is still deliberating on booze, so to be on the safe side, maybe we should stick with hallucinogenic drugs.
It’s a very basic standard of scientific research that you have to control for variables that could produce misleading results (confounders). Srsly, it’s in “Science 101”.
So, it’s more than a little hard to believe this has gone unnoticed.
I would have thought that, if even moderate drinking shortened your life, the relevant statistics would make it really obvious, even when taking lifestyle differences into account. Another bit of anecdotal evidence. I have had a moderate alcohol consumption for most of my life, although I have sometimes abstained for fairly long periods when training for various sporting activities. My brother used to smoke but now uses vapes and has always been quite a heavy drinker. I’m 66, he is 62, he looks much older than I do. A while ago he was relating to us that, during a discussion about retirement that he’d had at work, a colleague had said, “Well you must be well past retiring age surely.” Brother was apparently a bit miffed.
If teetotalitarianism increased your life expectancy, you’d expect Muslims to live longer than most other population groups in the UK. So I had a Google.
ONS:
Today’s new experimental data show that between 2011 and 2014 people of White and Mixed ethnic groups had lower life expectancy at birth than all other ethnic groups, with the Black African group having significantly higher life expectancy than most other groups.
This is why I identify as a proud Black woman.
“It also found the White ethnic group were more likely to die of cancer than their Black or Asian counterparts, while for both sexes, ischaemic heart disease mortality was highest in the Bangladeshi, Pakistani and Indian groups.
Weird, eh? Is cancer racist?
These days virtually everything has been found, by one scientist or another, to be detrimental to your longevity. The only problem is that sometime later a different study finds the complete opposite…
So Steve, the poor old whites have a lower life span since the evil colonialist blacks and asians oppress them so viciously??
I think you have to look into the mechanics of science funding.
Can you get a research grant to investigate whether moderate alcohol consumption is health harming?
Apparently yes.
Well there you are then.
Life spans: I wonder how many Pakistani immigrants, for instance, have a recorded date of birth that can be trusted. Or, indeed, immigrants from Africa. I take it that people born in Hong Kong will have been properly registered, and probably anywhere in western Europe. Eastern Europe? Dunno.
With some immigrant groups could one even trust the recorded date of birth for babies born here? Or purportedly born here. Dunno again.
Do you think that the ONS devotes much critical thought to such questions? Neither do I.
Steve’s reference is to “life expectancy” rather than completed lifespan, I’ll grant you. But it also refers mysteriously to “new experimental data”. Experimental? How weird.
Bboy – Yarpsolutely.
If it was t’other way round, we wouldn’t hear the end of the Great British Racist Death Gap, instead of it being an obscure bit of numberwanging on the ONS website.
Inequality doesn’t get any more fundamental than the inequality between the living and the popped clogs.
PiP – Life spans: I wonder how many Pakistani immigrants, for instance, have a recorded date of birth that can be trusted. Or, indeed, immigrants from Africa
African chappies usually pretend to be younger than they are (see: any number of African “wunderkids” being touted to football clubs)
@Steve
Your remarks about Islam don’t prove that drinking alcohol is good for you – just that drinking is less harmful than marrying your cousin.
Some of the data on alcohol harms comes from including the whole of the world, where alcohol can be contaminated, even poisonous, where there are black markets.
I’m willing to accept though that asking your liver to work more than a 1/3rd of a week on the problem, say 55 hours or a 110 units (based on a unit every half hour) is going to harm you. Guess I’ll find out.
David – Definitely, my female cousins look like Joseph Stalin.
Bongo – funny thing about booze is that it’s far more dangerous to stop drinking (if you’re an alky) than it is to stop doing most illegal drugs.
Nobody ever died of cannabis withdrawal.
headline in Telegraph:
Paris 2024 fans cannot drink alcoholic beer inside Olympic venues – but VIPs can
@Dave Ward
Generally this “X is harmful” / “X is actually good for you” back and forth is a clue that the purported benefits and harms are relatively small in the first place, which is why the evidence has been ambiguous. If the effect is large, like smoking, there’s really no dispute. With alcohol, the back and forth has been about drinking a little rather than a lot. Heavy drinking and binge drinking is clearly harmful. For light drinking, you could get two world experts to argue the toss all day without making any headway with each other – these beliefs are very deeply held and each side finds the evidence base on “their” side convincing.
@Stonyground
“I would have thought that, if even moderate drinking shortened your life, the relevant statistics would make it really obvious, even when taking lifestyle differences into account.”
No, because any plausible dose-response relationship is going to make the benefits or harms of moderate drinking rather small, and trying to determine much about such small effects is bloody difficult. And taking lifestyle differences into account is very tricky. So this turns out to be not obvious at all, which is why people have been arguing over “the U curve” for such a long time. Anyone who thinks “I’ll just look at Mormons or Muslims and see how they’re different” is being naive, because groups like that are so different to the rest of the population (even socio-economic status alone would be enough to completely muddy the picture, but there’s also diet, family structure, social or community support, genetics, …) and all that will swamp whatever differences are due to alcohol consumption.
What’s important to remember is that on the left-hand side of that graph people are arguing over small differences though. It’s not a symmetric “U” and on the right-hand side (ie heavier alcohol consumption) the harms shoot up, which is why it’s also called a “J” curve (assuming you use a font where “J” slopes slightly down before rising again to a higher level). What’s undisputed is the really serious harms of heavy drinking – any heavy drinker who doesn’t want to go teetotal should at least consider drinking more moderately, which would produce much of the same benefits even if the U or J curve hypothesis is wrong. And there’s not much point teetotallers feeling smug they aren’t a very light drinker instead, or a light drinker feeling smug they might be benefitting from being at the bottom of the U while the teetotaller isn’t – whoever is right, the difference is pretty small either way. If you’re really concerned about optimising every little thing for health purposes, there might be bigger gains to be made from inspecting your exercise, diet, sleep and stress levels, making sure you are engaged in enough social activities, and so on.
Why are lifestyle variables such a pain? Well you often aren’t going to have all the data you would like to control for. See @David’s point that even the variables you have access to are correlated with each other, including with alcohol use – if you know anything about multivariable regression analysis, you’ll know this makes standard error (ie uncertainty) worse. What you do get is usually self-reported and therefore at best only approximate (such measurement error also increases standard errors) and at worst full of lies (which biases the result). Even in studies that do say they controlled for some variable X, usually it just means they chucked it into their regression equation as one of their potentially explanatory variables. Fair enough, but this only controls for linear effects of X, and it may be that X has non-linear effects too. It turns out to be difficult to control for X “correctly”.
To complicate things further, there’s little doubt that even tiny quantities of booze raise, for example, cancer risk. Again this is a dose-response thing, so a low quantity means a small effect, but if you’re a cancer-phobe you might want to steer clear of even very light drinking. For the sake of your own mental health you might also want to avoid ever seeing the list of things deemed carcinogenic in California. The dispute over the existence of the U or J curve for alcohol boils down to whether there are also benefits from light drinking which may counteract or even outweigh these small harms. And some of these purported benefits seem to be quite culturally specific – not something purely about the biochemistry of ethanol itself, but more about what you’re drinking and the way you’re drinking it. (Which mirrors the problems of quantifying the harms of heavy drinking – it’s not just the quantity of alcohol but whether there are “days off”, how much is being consumed in binges etc. So even a dose-response graph can’t tell the whole story.)
A classic example is whether there is some protection from heart disease from other chemicals present in wine. If so, then the net harm of a particular consumption level of alcohol is going to differ between cultures, or socio-economic groups, where most alcohol intake is wine vs beer. There may be social benefits to having a beer with your mates at the pub, or a dinner party where wine is served, and in turn there are health benefits from a good social life. (One of Snowdon’s points is that drinkers tend to earn more than non-drinkers – he wonders, and has been hoping someone does a study, whether there may be networking benefits to social alcohol consumption that can improve socio-economic status, which incidentally would produce some health benefits.) Sitting on your sofa and having a beer or glass of wine alone is a different proposition, even if the alcohol consumption is exactly the same. And in principle you could have got the social benefits a different way, but there’s something cultural about “going to the pub” or “wine with a nice meal” that means this is the way it’s done in British society, at least of a certain generation. Young people are far less likely to drink, and not just an effect due to ethnic minorities either, and they may be creating different cultural spaces for socialisation. “Going for a bubble tea” if you’re urban and trendy, I’m told. So in a few decades, this will be buggering up all attempts to figure out the health harms or benefits of bubble tea.
Stockwell is a good example of “you can prove anything with statistics”, in that he knows how to get what he wants to find from the data – and he is perfectly upfront about what he wants to find. If you want to avoid this kind of research, and are deeply interested in the truth of the matter, you ideally want a double blinded randomized controlled trial… but you’ll never get one. Even if you could get through all the funding, ethics, organisational barriers, there’s no placebo.
In econometrics there is an alternative to the RCT, where you look at observational data but identify an “instrumental variable” (IV) that serves a similar purpose to the way an RCT randomly allocates people into groups – something “random” that affects your explanatory variable of interest. You need the IV to only affect your outcome variable via its effect on the explanatory variable of interest, it shouldn’t have an independent effect of its own. If you can find an appropriate IV, then you can analyse a kind of “natural experiment”. This is a well established technique in economics. Freakonomics fans will have come across it.
Medical studies often use a genetic marker as the IV, in which context the method is known as “Mendelian randomization” (MR). There are genes which affect how well people process and tolerate alcohol, and variation in those genes accounts for a substantial part of the variation in alcohol consumption. In other words, you can make a reasonable guess of how much someone drinks from their genes (especially in combination with social or personal data about them) and the genetic lottery of inheritance works rather like an RCT. I’m afraid the bad news from MR studies is they don’t seem to find U effects, ie even low alcohol intakes have net harms compared to not drinking. But they do still find a dose-response relationship – the harms of light drinking are found to be smaller than the harms of heavy drinking, the question for a light drinker is whether they find such harms worth worrying about.
One of my few major quibbles with Snowdon’s takedowns is that he dismisses the MR studies since they are based on predicted, rather than recorded, alcohol consumption. But if they were based on the recorded alcohol consumption then you’re back to the standard observational study which is known to be of limited use for assessing causation. This is just how the math of IV methods works, and I hope he wouldn’t throw all econ IV studies out of the window for the same reason. There are things you can sensibly critique about particular IV studies, especially how well the instrument obeys all the necessary statistical assumptions, but I’ve not seen Snowdon do that.
Rather than aiming to “debunk” one or the other, there’s an interesting discussion to be had as to why observational studies with what look like appropriate controls tend to find a U or J (though the U effect can get smaller the better the controls are, the issue is whether it disappears entirely or not) whereas MR studies don’t. But people discussing the issue tend to be pretty partisan so I’ve not seen a discussion anywhere that goes much further than “because the other way is methodologically a total load of rubbish”, which isn’t very helpful.
Is the spam trap still consuming comments?
Fine, let’s have another go then.
@Dave Ward
Generally this “X is harmful” / “X is actually good for you” back and forth is a clue that the purported benefits and harms are relatively small in the first place, which is why the evidence has been ambiguous. If the effect is large, like smoking, there’s really no dispute. With alcohol, the back and forth has been about drinking a little rather than a lot. Heavy drinking and binge drinking is clearly harmful. For light drinking, you could get two world experts to argue the toss all day without making any headway with each other – these beliefs are very deeply held and each side finds the evidence base on “their” side convincing.
@Stonyground
“I would have thought that, if even moderate drinking shortened your life, the relevant statistics would make it really obvious, even when taking lifestyle differences into account.”
No, because any plausible dose-response relationship is going to make the benefits or harms of moderate drinking rather small, and trying to determine much about such small effects is bloody difficult. And taking lifestyle differences into account is very tricky. So this turns out to be not obvious at all, which is why people have been arguing over “the U curve” for such a long time. Anyone who thinks “I’ll just look at Mormons or Muslims and see how they’re different” is being naive, because groups like that are so different to the rest of the population (even socio-economic status alone would be enough to completely muddy the picture, but there’s also diet, family structure, social or community support, genetics, …) and all that will swamp whatever differences are due to alcohol consumption.
What’s important to remember is that on the left-hand side of that graph people are arguing over small differences though. It’s not a symmetric “U” and on the right-hand side (ie heavier alcohol consumption) the harms shoot up, which is why it’s also called a “J” curve (assuming you use a font where “J” slopes slightly down before rising again to a higher level). What’s undisputed is the really serious harms of heavy drinking – any heavy drinker who doesn’t want to go teetotal should at least consider drinking more moderately, which would produce much of the same benefits even if the U or J curve hypothesis is wrong.
And there’s not much point teetotallers feeling smug they aren’t a very light drinker instead, or a light drinker feeling smug they might be benefitting from being at the bottom of the U while the teetotaller isn’t – whoever is right, the difference is pretty small either way. If you’re really concerned about optimising every little thing for health purposes, there might be bigger gains to be made from inspecting your exercise, diet, sleep and stress levels, making sure you are engaged in enough social activities, and so on.
For the statistically inclined…
Why are lifestyle variables such a pain? Well you often aren’t going to have all the data you would like to control for. And see @David’s point that even the variables you have access to are correlated with each other, including with alcohol use – if you know anything about multivariable regression analysis, you’ll know this makes standard error (ie uncertainty) worse. What you do get is usually self-reported and therefore at best only approximate (such measurement error also increases standard errors) and at worst full of lies (which biases the result). Even in studies that do say they controlled for some variable X, usually it just means they chucked it into their regression equation as one of their potentially explanatory variables. Fair enough, but this only controls for linear effects of X, and it may be that X has non-linear effects too. It turns out to be difficult to control for X “correctly”.
To complicate things further, there’s little doubt that even tiny quantities of booze raise, for example, cancer risk. Again this is a dose-response thing, so a low quantity means a small effect, but if you’re a cancer-phobe you might want to steer clear of even very light drinking. For the sake of your own mental health you might also want to avoid ever seeing the list of things deemed carcinogenic in California. The dispute over the existence of the U or J curve for alcohol boils down to whether there are also benefits from light drinking which may counteract or even outweigh these small harms.
And some of these purported benefits seem to be quite culturally specific – not something purely about the biochemistry of ethanol itself, but more about what you’re drinking and the way you’re drinking it. (Which mirrors the problems of quantifying the harms of heavy drinking – it’s not just the quantity of alcohol but whether there are “days off”, how much is being consumed in binges etc. So even a dose-response graph can’t tell the whole story.)
A classic example is whether there is some protection from heart disease from other chemicals present in wine. If so, then the net harm of a particular consumption level of alcohol is going to differ between cultures, or socio-economic groups, where most alcohol intake is wine vs beer. There may be social benefits to having a beer with your mates at the pub, or a dinner party where wine is served, and in turn there are health benefits from a good social life. (One of Snowdon’s points is that drinkers tend to earn more than non-drinkers – he wonders, and has been hoping someone does a study, whether there may be networking benefits to social alcohol consumption that can improve socio-economic status, which incidentally would produce some health benefits.) Sitting on your sofa and having a beer or glass of wine alone is a different proposition, even if the alcohol consumption is exactly the same.
In principle you could have got the social benefits a different way, but would you have actually gone and done so? There’s something cultural about “going to the pub” or “wine with a nice meal” that means this is the way it’s done in British society, at least of a certain generation. Young people are far less likely to drink, and not just an effect due to ethnic minorities either, and they may be creating different cultural spaces for socialisation. “Going for a bubble tea” if you’re urban and trendy, I’m told. So in a few decades, this will be buggering up all attempts to figure out the health harms or benefits of bubble tea.
Snowdon’s discussion of a previous Stockwell article is https://snowdon.substack.com/p/moderate-drinking-is-still-good-for
Stockwell is a good example of “you can prove anything with statistics”, in that he knows how to get what he wants to find from the data – and he is perfectly upfront about what he wants to find. If you want to avoid this kind of research, and are deeply interested in the truth of the matter, you ideally want a double blinded randomized controlled trial… but you’ll never get one. Even if you could get through all the funding, ethics, organisational barriers, there’s no placebo.
In econometrics there is an alternative to the RCT, where you look at observational data but identify an “instrumental variable” (IV) that serves a similar purpose to the way an RCT randomly allocates people into groups – something “random” that affects your explanatory variable of interest. You need the IV to only affect your outcome variable via its effect on the explanatory variable of interest, it shouldn’t have an independent effect of its own. If you can find an appropriate IV, then you can analyse a kind of “natural experiment”. This is a well established technique in economics. Freakonomics fans will have come across it.
Medical studies often use a genetic marker as the IV, in which context the method is known as “Mendelian randomization” (MR). There are genes which affect how well people process and tolerate alcohol, and variation in those genes accounts for a substantial part of the variation in alcohol consumption. In other words, you can make a reasonable guess of how much someone drinks from their genes (especially in combination with social or personal data about them) and the genetic lottery of inheritance works rather like an RCT. I’m afraid the bad news from MR studies is they don’t seem to find U effects, ie even low alcohol intakes have net harms compared to not drinking. But they do still find a dose-response relationship – the harms of light drinking are found to be smaller than the harms of heavy drinking, the question for a light drinker is whether they find such harms worth worrying about.
One of my few major quibbles with Snowdon’s takedowns is that he dismisses the MR studies since they are based on predicted, rather than recorded, alcohol consumption. But if they were based on the recorded alcohol consumption then you’re back to the standard observational study which is known to be of limited use for assessing causation. This is just how the math of IV methods works, and I hope he wouldn’t throw all econ IV studies out of the window for the same reason. There are things you can sensibly critique about particular IV studies, especially how well the instrument obeys all the necessary statistical assumptions, but I’ve not seen Snowdon do that.
Rather than aiming to “debunk” one or the other, there’s an interesting discussion to be had as to why observational studies with what look like appropriate controls usually find a U or J (though the U effect can get smaller the better the controls are, the issue is whether it disappears entirely or not) whereas MR studies don’t. But people discussing the issue tend to be pretty partisan so I’ve not seen a discussion anywhere that goes much further than “because anyone who finds the result I don’t like is clearly methodologically doing a total load of rubbish”, which isn’t very helpful.
@Anon
Yeah i can see why you wouldn’t want to include Mormon data. You’d have factor in the numbers of smiling, friendly Mormons knocking at front doors end up buried at the bottom of people’s gardens. No doubt some of them the result of alcohol consumption. Or even lack of alcohol consumption. But not theirs. Jehovah’s Witnesses I imagine similar. A copy of the Watchtower can be a lethal weapon in some people’s hands.
A very non-technical intro to IV: https://freakonomics.com/podcast/you-make-me-feel-like-a-natural-experiment/
Note that the study to estimate the effect of military service on income is based only on the man’s Vietnam War draft lottery results, producing a kind of “predicted probability of military service” variable, and not whether he ended up serving in reality. This is a completely deliberate choice to avoid problems of confounding variables (like how the military selected from eligible to serve) and is at the heart of how IV works.
I’m not sure why Snowdon has been so vociferous about it being wrong to use the predicted rather than observed explanatory variables in IV studies – decades back it might only be something you saw at postgrad level but these days this is typically taught in the second year of an undergrad econ degree (I have a feeling Snowdon studied geography or something rather than econ) and it might be worth him hitting the books again. He’s typically very good at spotting statistical errors in other people’s work.
And for anyone who can bear to see some maths:
https://en.wikipedia.org/wiki/Instrumental_variables_estimation
https://en.wikipedia.org/wiki/Mendelian_randomization
@bis
There are lots of studies on Mormons, and Seventh Day Adventists who have similar lifestyle practices but also tend to be vegetarian. “Clean living” does seem to extend life, but I can’t imagine either of the studies below would persuade you towards religiosity however many extra years seem to be on offer. It’s just tricky to pin down how much of that is due to avoiding alcohol – and in a group with different social structure and culture, say one where alcohol plays a bigger role in socialising, it’s not clear that the results on light vs zero drinking would be transferable anyway.
https://pubmed.ncbi.nlm.nih.gov/17920112/
“Method: We examined 9815 religiously active California Mormon adults followed for mortality during 1980-2004 and 15,832 representative U.S. white adults enrolled in the 1987 National Health Interview Survey (NHIS) and followed for mortality during 1988-1997. The standardized mortality ratio (SMR) and 95% confidence interval (CI) was calculated relative to U.S. whites defined to have a SMR of 1.00.
Results: Active California Mormons practice a healthy lifestyle advocated by their religion, which emphasizes a strong family life, education and abstention from tobacco and alcohol. Unusually low SMRs occurred among married never smokers who attended church weekly and had at least 12 years of education. For those aged 25-99 years at entry, the SMR for all causes of death was 0.45 (0.42-0.48) for males and 0.55 (0.51-0.59) for females. For those aged 25-64 years at entry, the SMR for all causes of death was 0.36 (0.32-0.41) for males and 0.46 (0.40-0.53) for females. Life expectancy from age 25 was 84 years for males and 86 years for females. These SMRs were largely replicated among similarly defined persons of all religions within the NHIS cohort.”
https://pubmed.ncbi.nlm.nih.gov/11434797/
“California Adventists have higher life expectancies at the age of 30 years than other white Californians by 7.28 years (95% confidence interval, 6.59-7.97 years) in men and by 4.42 years (95% confidence interval, 3.96-4.88 years) in women, giving them perhaps the highest life expectancy of any formally described population. “
Hold on. This is a meta analysis of 107 studies.
So basically, none of these studies showed booze to be dangerous in small quantities, but if you’re *incredibly* clever you can combine the results to reach the opposite outcome.
That’s like saying that individually chickens can’t fly, but our computer model proves that if you attach 10 of them to an anvil you’ll find that they take off.
Everybody who drinks water dies.
There are lots of studies on Mormons, and Seventh Day Adventists who have similar lifestyle practices but also tend to be vegetarian. “Clean living” does seem to extend life, but I can’t imagine either of the studies below would persuade you towards religiosity however many extra years seem to be on offer.
I do find people like that make me contemplate committing murder. Now whether that affects their longevity…
my female cousins look like Joseph Stalin
Including the ‘tache?
Including the ‘tache?
Da, comrade. (shudder)
Btw, and on a completely different topic, have you ever noticed how many Hyacinth and Richard couples there are among Millennials? With a generously proportioned, yet stridently self-confident big round lass leading around her quiet, skinny husband? Fat girls need loving too, I suppose. And good on them, for landing a man and all that. But how exactly did Hyacinth and Richard conceive Sheridan, their probably gay son at the Poly?
“Riiiichard! That is not where you put your little marital implement, dear!”
That mental image was free. (Not free as in libre, free as in leprosy is free.)
So I think the lesson here is, don’t shag anybody from an official family reunion, except your wife. Also, you know how they made that Critters (1986) looking She-Yeti in My Big Fat Greek Wedding look presentable with the foul Eastern lies of makeupin order to trick a white man into marrying a girl who looks more like a 40 year old chain-smoking divorced gypsy fortune teller when not plastered in slap and flattering camera angles? Always beware of Greek trickery, even if they say something is free or touch your willy, I reckon.
Hope that helps. x
I would have thought that methodists would be better to compare to than Muslims or Mormons as they only don’t drink alcohol – they drink caffeine eat pork and don’t marry their cousins etc.
Sadly the church is declining.
IV analysis in the mud. And brilliant by Charlie Oaks.
“Keep an eye open for Chris Snowdon’s rebuttal of this. He’s the expert on this sort of tosh.”
Voila! There was a quite extraordinary degree of cherry-picking in this study. They started with thousands of pieces of evidence and whittled them down to FIVE! https://snowdon.substack.com/p/cherry-picking-the-evidence-on-alcohol