im still curious how those (few) courses which are oversubscribed select then.
There are loads of courses which are oversubscribed. The point is that (a) if a course isn't 5:1 oversubscribed then the natural action of the UCAS form and then clearly means that it isn't (ie, if you imagine a world with 5 universities each with 100 places and 500 applicants, every place would be 5:1 oversubscribed but in the end everyone would get a place) and (b) standard offers are set so as to balance applicants against places, with the offer usually being an indication of how desirable students think the course is.
The elephant in the room for admission is A Level grades: the offer is a measure of how popular the course is, and measures a mixture of the subject, the institution and the location. Why mess about picking through unreliable personal statements when you can simply raise the standard offer, which is a virtuous circle? The achieved tariff is a key input to a lot of league tables, and a lot of students are attracted by high offers, thinking (not entirely wrongly, but not as much as they hope) that it says something about how good the course is. We raised our offer and got more applicants and more conditional firms: what's the beer that used to advertise itself as "reassuringly expensive?" People are reassured by high standard offers even if you end up taking large numbers of people who actually miss that offer, because you can keep quiet about that and it often doesn't show up in the tables because you include in the achieved tariff qualifications that you didn't use for admission.
Most people deal with high or low numbers by raising or lowering their offer until they are happy they have about the right number of conditional firms to get about the right number of students come October. It's safer to err slightly on the side of under-recruiting if you're popular, as you can pick up any shortfall by accepting some of the conditional firm students who miss the offer (I've heard of students with AAB offers achieving ACD and still getting their first choice). The statistic no department is going to tell you without a gun to their head is "if your standard offer is AAA, what is the lowest set of grades you actually accepted prior to results being published, and what is the lowest set of grades you accepted in clearing?" And the statistic there is no way to calculate is "of the applicants, how many were realistic applicants whose qualifications were remotely suitable?" Popular, high-profile courses attract a lot of applicants who are never going to get in whatever happens, but they bolster the impression of those courses being wildly over-subscribed.
The places that for various reasons can't admit just by raising the offer use other stuff (references, aptitude tests, etc) as input to an interview process and commit a lot of resources to that, up to interviewing every plausible candidate. They may have very clearly stated work experience requirements, too. What I don't believe is that there are any significant number of courses which admit by looking at personal statements and then using that to either make a conditional offer lower than they would if they offered to everyone, or a no offer decision, and I don't believe that the personal statement is a significant input into the decision to interview (ie, if two candidates differ only in the quality of their personal statement, I think most/all places would interview both).
The key point is that if universities are operating a hidden set of criteria which they cannot objectively defend as being necessary to either (a) ensure people are suitable for the course or (b) select the stronger candidates from an excess of people who meet criterion (a) then they do so at serious risk to their OfFA credibility. That's why I am deeply sceptical about the "ah, they may say that, but..." or "well, behind closed doors they think..." claims. A university will publish its admission requirements, including listing the weight if any placed on particular things in the application, and if that isn't the truth, they are taking a substantial risk.