Choosing an Oxford College

I'm heading for Mansfield, which was my first choice but which regularly more or less props up the Norrington table. According to its latest newsletter, Mansfield thinks that the table calculation is unfair but feels that they need to get further up the table under the current system before they complain about it, for 'sour grapes' reasons. Not a mathematician myself, but think it is something to do with the points being skewed towards Firsts, particularly in the context of a small cohort. If the question was 'at which college are you least likely to end up with a 2.2' rather than 'at which college are you most likely to end up with a First', the results might be different and, it may be argued, more pertinent for most of us! But the maths is such that the table exaggerates what are really very similar results; half a dozen additional Firsts (which could of course mean half a dozen extra exam marks!) would sometimes have raised colleges many places further up the table.

The rapid journey of some colleges up and down the table must cast doubts on its reliability, yet it can't be denied that a handful are almost always at the top and bottom. For a number of reasons, I don't think that this is due to the intake (he says modestly) at each college, nor the quality of the tutorials. Funding may have some bearing, but I suspect that college ethos and culture would be the main factors. At Mansfield we have a new, high profile Principal starting soon (Dame Helena Kennedy) so it will be interesting to see if anything changes in this respect.

I feel incredibly lucky to be going to Oxford and have never really understood some of the TSR posts one reads, about not wanting to accept an offer from a 'pooled' and perhaps less well known college or PPH. All of the colleges will be presenting us with the chance of a lifetime and we will have no-one to blame but ourselves if our final result disappoints.

(edited 12 years ago)

Reply 1622

BJack

19

Original post by shoshin

According to its latest newsletter, Mansfield thinks that the table calculation is unfair but feels that they need to get further up the table under the current system before they complain about it, for 'sour grapes' reasons. Not a mathematician myself, but think it is something to do with the points being skewed towards Firsts, particularly in the context of a small cohort. If the question was 'at which college are you least likely to end up with a 2.2' rather than 'at which college are you most likely to end up with a First', the results might be different and, it may be argued, more pertinent for most of us!

Given that the typical degree classification breakdown can vary quite considerably from subject to subject (the percentage of scientists getting firsts is significantly higher than that of the artists, most generally), the breakdown of students' subjects must have some impact on how well a college can reasonably expect to do in the rankings.

That said, that the majority of Magdalen's finalists got firsts last year (for example) is really quite something & obviously can't be accounted for by subject bias.

Reply 1623

If you look at the pretty graph at http://en.wikipedia.org/wiki/Norrington_Table, only Magdalen, Merton, John's and New are consistently near the top; Harris Manchester, Mansfield and St. Hilda's towards the bottom.

I'm not aware of any colleges being particularly asymmetric in their arts\sciences split, though it is an obvious source of bias in the table. Keble have a stereotype of being predominantly male and scientists, but tend to be towards the bottom half than the top.

Reply 1624

Original post by Huw Davies

If you look at the pretty graph at http://en.wikipedia.org/wiki/Norrington_Table, only Magdalen, Merton, John's and New are consistently near the top; Harris Manchester, Mansfield and St. Hilda's towards the bottom.

I'm not aware of any colleges being particularly asymmetric in their arts\sciences split, though it is an obvious source of bias in the table. Keble have a stereotype of being predominantly male and scientists, but tend to be towards the bottom half than the top.

Mansfield is a good example of such an asymmetry, see:

http://www.mansfield.ox.ac.uk/prospective/studying/subjects.html

Edit: p.s. Stats at www.unistats.com highlight the issue, eg. Oxford Firsts in History 28%, Chemistry 39% etc.

(edited 12 years ago)

Reply 1625

cpchem

14

Original post by shoshin

I'm heading for Mansfield, which was my first choice but which regularly more or less props up the Norrington table. According to its latest newsletter, Mansfield thinks that the table calculation is unfair but feels that they need to get further up the table under the current system before they complain about it, for 'sour grapes' reasons. Not a mathematician myself, but think it is something to do with the points being skewed towards Firsts, particularly in the context of a small cohort. If the question was 'at which college are you least likely to end up with a 2.2' rather than 'at which college are you most likely to end up with a First', the results might be different and, it may be argued, more pertinent for most of us! But the maths is such that the table exaggerates what are really very similar results; half a dozen additional Firsts (which could of course mean half a dozen extra exam marks!) would sometimes have raised colleges many places further up the table.

The rapid journey of some colleges up and down the table must cast doubts on its reliability, yet it can't be denied that a handful are almost always at the top and bottom. For a number of reasons, I don't think that this is due to the intake (he says modestly) at each college, nor the quality of the tutorials. Funding may have some bearing, but I suspect that college ethos and culture would be the main factors. At Mansfield we have a new, high profile Principal starting soon (Dame Helena Kennedy) so it will be interesting to see if anything changes in this respect.

I feel incredibly lucky to be going to Oxford and have never really understood some of the TSR posts one reads, about not wanting to accept an offer from a 'pooled' and perhaps less well known college or PPH. All of the colleges will be presenting us with the chance of a lifetime and we will have no-one to blame but ourselves if our final result disappoints.

The Master of my old college has been a prat for years, and we've been all over the top half of the Norrington table (between 1st and 14th in the last eleven years). It does make me wonder - if they elect someone who's not a prat next time, will standards slip?

Reply 1626

Original post by shoshin

Mansfield is a good example of such an asymmetry, see:

http://www.mansfield.ox.ac.uk/prospective/studying/subjects.html

Edit: p.s. Stats at www.unistats.com highlight the issue, eg. Oxford Firsts in History 28%, Chemistry 39% etc.

Not massively clear from that without the numbers, but Chemistry\Biochemistry\Biology notable in their absence even if Maths\Materials\Engineering are there. I do believe you. With the college so close to the science area, too!

Reply 1627

Original post by BJack

...the breakdown of students' subjects must have some impact on how well a college can reasonably expect to do in the rankings

Good point. As the wiki entry notes:

The table is biased towards colleges which have larger than average numbers of students in science subjects such as Chemistry and Mathematics where a higher proportion attain a 1st Class degree compared to arts degrees

Reply 1628

Original post by Huw Davies

Not massively clear from that without the numbers, but Chemistry\Biochemistry\Biology notable in their absence even if Maths\Materials\Engineering are there. I do believe you. With the college so close to the science area, too!

Maybe I'm sounding a little desperate though, innit? The chap protesteth too much etc., as I'm off to Mansfield myself :smile:

Reply 1629

Original post by shoshin

Maybe I'm sounding a little desperate though, innit? The chap protesteth too much etc., as I'm off to Mansfield myself :smile:

Read the article by John Lucas linked to on the wikipedia page and feel much more satisfied that the NT is a Bad Thing. Ultimately, everyone sits the same exams and if you get a First you won't mind how many other people in your college did!

With such a big hole where various sciences should be it would be surprising if Mansfield didn't have a smaller total so I think your reasoning's pretty sound. Can't think of many reasons to underperform consistently other than admissions in which HMC is an obvious outlier and Hilda's was until very recently.

Reply 1630

Budgie

12

There is obviously going to be a difference in teaching quality across colleges - some random, some caused by differences in finances / prestige / what academics they can attract. However, it's difficult to make broad-brush statements - looking at consistent Norrington performance is the only way of doing that. For example, I know that the PPE tutors at Magdalen are fantastic but have no idea about any other subject, other than via the Norrington. So the best thing you can do is get a broad picture by combining the impression you get from the Norrington with speaking to students at the colleges you are considering who do your subject. It's perfectly possible that a great college will be bad at one subject. Also, remember that after first year a significant amount of teaching can be done outside college - 3 (of 8) of my modules are taught elsewhere.

Reply 1631

Original post by Budgie

...looking at consistent Norrington performance is the only way of doing that.

But, as we have been discussing, the Norrington Table (unlike its Cambridge equivalent, according to Wiki) doesn't take account of the fact that First Class degrees are (much?) more prevalent in Science subjects. So colleges that do not offer the full range of Science subjects (like Mansfield) are disadvantaged. A 'subject by subject' Norrington table would be a step forward, though this would only remove one of the problems arising from reliance on Norrington.

Edit: Sorry, my subject by subject suggestion above is not a good one, cos numbers per subject are statistically too small, particularly in the smaller colleges.

(edited 12 years ago)

Reply 1632

cpchem

14

Original post by shoshin

But, as we have been discussing, the Norrington Table (unlike its Cambridge equivalent, according to Wiki) doesn't take account of the fact that First Class degrees are (much?) more prevalent in Science subjects. So colleges that do not offer the full range of Science subjects (like Mansfield) are disadvantaged. A 'subject by subject' Norrington table would be a step forward, though this would only remove one of the problems arising from reliance on Norrington.

Edit: Sorry, my subject by subject suggestion above is not a good one, cos numbers per subject are statistically too small, particularly in the smaller colleges.

I wouldn't say that the numbers are statistically insignificant just for the smaller colleges - I imagine the largest subject-college year is about 15, which is pretty insignificant.

Reply 1633

Original post by cpchem

I wouldn't say that the numbers are statistically insignificant just for the smaller colleges - I imagine the largest subject-college year is about 15, which is pretty insignificant.

True. I was sticking limpet-like to my defence of Mansfield's academic credentials (as the smallest college, and with an Arts subject bias to boot). But, as you say, even the largest subject intake at the largest college would have its 'Norrington-by-subject' score varying wildly depending on a single First Class result.

Reply 1634

Budgie

12

Original post by shoshin

But, as we have been discussing, the Norrington Table (unlike its Cambridge equivalent, according to Wiki) doesn't take account of the fact that First Class degrees are (much?) more prevalent in Science subjects. So colleges that do not offer the full range of Science subjects (like Mansfield) are disadvantaged. A 'subject by subject' Norrington table would be a step forward, though this would only remove one of the problems arising from reliance on Norrington.

Edit: Sorry, my subject by subject suggestion above is not a good one, cos numbers per subject are statistically too small, particularly in the smaller colleges.

Agreed, but it's not as if there are many alternatives available at the moment. Looking at the Norrington and then trying to adjust for this factor is the best one can do.

Reply 1635

nulli tertius

21

Original post by Huw Davies

There's a lot of variation year-on-year in league table position, which suggests that chance is a big factor, though there are a few colleges consistently at the top\bottom. I think it's probably to do with (self)-selection of candidates and the overall culture of work rather than better tutorials. A lot of the finals-level stuff won't necessarily be taught at your own college.

It used to be the case, I do not know whether it still is, that a good Norrington Table position was reflected in a reduced number of applicants in the next admissions cycle. This was claimed to be one of the major reasons for the Yo-Yo effect in the Norrington Table seen by many colleges.

Reply 1636

Original post by nulli tertius

It used to be the case, I do not know whether it still is, that a good Norrington Table position was reflected in a reduced number of applicants in the next admissions cycle. This was claimed to be one of the major reasons for the Yo-Yo effect in the Norrington Table seen by many colleges.

A high position is more likely to be an attraction as a deterrent - if anything, the correlation is between lots of applicants and doing well.

I think the most obvious reason for yo-yoing is regression to the mean - the difference between mid-table colleges is probably almost entirely noise and so when they do well it is statistically more likely that they will do worse the following year than that they'll do better, and vice versa.

Reply 1637

nulli tertius

21

Original post by Huw Davies

A high position is more likely to be an attraction as a deterrent - if anything, the correlation is between lots of applicants and doing well.

That may be the case today. As I indicated, it was not in the past. Oxford seems no longer to publish statistics on the application to places ratio broken down by college.

Reply 1638