January 2021 UK Politics Thread - sunlit uplands are here

Imagine you have a line of 11 people, arranged from shortest to tallest.

Get the shortest to leave the line and stand to the left.

Then get the tallest to leave the line, and stand to the right.

Keep going until there’s one person in the middle, that’s mr. Median Average.

There will also be 5 people standing to the left who are all shorter than the average person, and 5 people to the right who are all taller than them. The taller ones could all be 6ft11 and the average person is only 5ft 9

1 Like

Half the people.

Take all the people, put them in a row ordered by salary, take the person in the middle of the row and ask them what they earn. That’s your median.

1 Like

Seven numbers in order:

0 1 1 2 9 9 9

The median is the one in the middle (2). The mean would be to add all those up and divide by seven (4.3).

Or something like that.

1 Like

Right, but as @harru says, the people below the median could be on £11k, and the people above the median could be on £50k? Like, there’s no assumption that it’s a straight correlation from lowest to highest salary?

That’s right, there are different averages used for different things - the median is useful if you have some bellend like Bezos in the room earning 10bn per second, because it just ignores him. It’s not useful though at pointing out outliers or discrepancies.

Professional stats nerds will spend all day (their whole life) working on stuff like this to try and get averages that are realistic and take all different things into account.

Yes. Generally speaking with earnings the median is more useful than the mean, partly because as you say the highest earners skew the results a lot because there’s no top end to potential earnings, but there very much is a lower limit that people can live off.

Once you enhance the median into deciles and percentiles it becomes even more informative.

1 Like

Yeah you end up with a distribution that crudely looks like this

Where the median is a more useful midpoint with as many people earning below as above it, and the mean being skewed by the long tail off and someone on 100k having 5x greater impact on it than someone earning 20k

edit: mainly for @anon34135029 sorry

2 Likes

Mega, thanks everyone! I think that, at the age of 34, I’m figuring out that I’m a visual learner because this graph makes everything much clearer

Also, to my question about inequality, does that not mean that the median also acts to reinforce inequality statistically as the long head of high earners is disproportionate compared to the big hump of folks below the median?

That’s a harder question I think. Anything that tries to distill national earnings into one average figure is going to inevitably miss lots of nuance. Median is the most useful in this sort of question because if you take a sample of full time workers theres a 50:50 chance that they’ll earn above or below that figure.

It doesn’t take into account geographic differences, cost of living, how much above or below that line they are, dependents, single/multiple earner households or anything else that you’d need to examine inequality in any detail.

Not quite sure what you mean by this. the median line means there are exactly the same number of people above as below the line, and no extra weighting is given to any individual. The mean will be skewed by that long run-off, a median won’t

2 Likes

Yeah, this is where I start to fall down! Thanks for clarifying.

I think, as you’re pointing towards, I’m trying to obtain meaning that’s not really doable without more granular analysis.

1 Like

Assange has been denied bail and will remain in prison, while the US appeals the decision to turn down the extradition request.

another interesting example that works out the same way as salary is that there are more people with a super size willy than with a super small willy

so the median size is smaller than the mean size

ergo, most people have a smaller than average willy

@ me you coward

4 Likes

Yup, this is why in Stats we have multiple different values we can use to describe a set of data.

The different forms of average (median/mean etc) give you a middle value, but tell you nothing else about it - other values would be used to start to describe the shape of the data and get into talking inequality.

1 Like

https://twitter.com/PippaCrerar/status/1346828526608125952?s=19

https://twitter.com/PippaCrerar/status/1346828625836986369?s=19

G’wan Pippa

17 Likes

Tickled that this post has no likes.

1 Like

WTF at the answers to this poll?

Like earning £31k is like unrealistic dream level stuff that’s never going to happen, and you’re telling me that 60% of people here are getting more than that?!?

2 Likes

There’s something in the middle tbh - MPs represent blocks of people but also distinct communities/places, particularly in our system where you have winner takes all / relatively small constituencies. So there’s a solid argument for dedicated MPs for distinct rural areas or isolated places being good for democracy, even if per head they’re representing fewer people. This is why some places (the Scottish islands, as an example) have it set out in legislation that they have to be a distinct constituency at Westminster, and can’t be subsumed into a bigger one.

Or: London doesn’t gain much from getting a couple more MPs. Wales would lose a lot with a couple fewer.

4 Likes

Like everything, it depends on what you do for a living?

There’s an over-representation of graduates, office workers, and people in their late 30s (or later) and in London on here, which probably explains a lot of it.

(But then the starting graduate salary in my field was mid-20s, so it’s only a few promotions above that :man_shrugging:)

1 Like

18 Likes