please empty your brain below

"It's because they're not trying to be representative, they're trying to be inclusive"
And why just pick the categories of race, age and sex? What about disability, wealth, attractiveness, interest in trains?
This is a fascinating visualisation. Thank you.

How easy would it be to include income, or social class (A, B, C1 perhaps)?

I think you would need at least eight people before there would be more than a 50% chance of one having attended a private school (6.5% or about 1 in 15). And fifty or more to get one that attended Oxbridge (less than 1%).

Think about that, and sex and age and ethnicity, when you are looking at corporate CEOs, or judges, or MPs.

Another consideration might be that our little group of 6 characters is not going to be the only thing that anyone sees. It will take its place among the thousands of images already out there. Historically these were (arguably) predominantly white and male. So perhaps we may wish to make a tiny effort to redress this balance.
Properly interesting, and thought provoking. And beautifully presented!
Thank you.
> What about disability, wealth, attractiveness, interest in trains?

Fortunately for the public bodies' graphic designers, most other characteristics aren't as immediately visually apparent. Though there is a trend to include some form of disability (necessarily a visible one), presumably on similar grounds.

Compare & contrast with housebuilders' artists' impressions of finished developments, which tend to reflect the people they want to sell to (always young, latte-holding...).
Answering my own questions on social class and income:

By reference to NRS social grade in 2016, which looks at the occupation of the "head" of the household, 1 in 25 are A (higher managerial, professional), 2 in 9 are B (intermediate managerial, professional), 2 in 7 are C1 (supervisory/junior managerial, professional), 1 in 5 are C2 (skilled manual), 1 in 7 are D (semi- or un-skilled) and 1 in 10 are E (state pensioners, casual, unemployed).

So of the six, three are ABC1 (no As, one rounded down B, two rounded up C1) and three are C2DE (one rounded down C2, and one each rounded up D and E).

And from ONS incomes data for 2018, the median annual income for each sextile would be £14k, £20k, £25k [national median of £29k in here] £30k, £38k, and £54k.
The problem with concentrating on "inclusiveness" is that you end up with a visually distorted view of what society looks like while still not covering every group.

And do people just look to see their "skin colour" portrayed anyway? For example, would an 80 year old Sikh man see a 20 year old Muslim woman as representing him better than say an 80 year old white man?
Skin colour. DG has implied that human skin comes in three colours - black, white and Asian. Odd, because my colour chart doesn't have the third one of these.
Indeed, Malcolm, and white people are in the main some variety of pink at this time of year.

But if you look at the population statistics that DG linked, you'd see the census data has many sub-categories for ethnic origin which are grouped into five main categories: white, black, Asian, mixed, and other. The sub-categories for Asian are described as Indian, Pakistani, Bangladeshi, Chinese, and other, and about two thirds are in the first three boxes, i.e. South Asian, rather than East Asian, or something else.

That said, ethnic origin maps poorly onto skin colour. If it helps, please imagine red, orange, yellow and green people. I'll let you decide which is which.
Scandinavian white, Irish white, southern European white, Eastern European white, north mainland European white, Antipodean white, half-caste white and any other, west African Black, Southern African black, North Eastern African black, South East African black, Central African black, Any African half caste with any other, Mongolian, North Chinese Asian, South Chinese Asian, West Chinese Asian, Central Chinese Asian, East Chinese Asian, Red American Indian, North Indian Asian, Central Indian Asian, South Indian Asian, Arabian,et al et al et al.

I look forward to a time when it is not necessary to be specific about the colour used to represent a human just as it is not necessary to specify a breed of dog when a image of a dog is used to signify "for all dogs".
I work in a charity and am involved in producing their publications. We have this sort of discussion about photos of people and it's a basic principle to try to show at least one Asian and one black person.
But the proportions of pictures we show of different age groups don't have to be politically correct or correspond with national averages, because our services are mainly for older people.
"As for age, we need one character representative of the youngest half of the population and one representative of the oldest half. In statistical language we need the lower and upper quartiles of the UK's age distribution."

See, I'm not sure you do - or, more accurately, you only need those if you're trying to do one particular thing, which is not the only thing you could do.

As an alternative, instead of picking representatives at 25% and 75% of the distribution, you could instead pick representatives at (with rounding) 33% and 67% of the distribution. Similarly, with n representatives, instead of picking people at 1/2n, 3/2n, .... 2n-1/2n, you could pick people at 1/n+1, 2/n+1.... n/n+1.

Is this better? It depends what you're trying to achieve. 50% of the population will be between 25% and 75% and people in the middle may not feel well represented by either alternative. On the other hand, with the more even distribution that I propose, more people will feel somewhat close to either of the alternatives proposed. On the downside, people at extremes of the distribution (e.g. the very old and the very young) would not be so well represented.

You can't say one is definitely better than the other, but you can come up with different reasonable ways to define what better is, and optimise accordingly.
Intriguing. Certainly the notion of inclusiveness is one that only the most curmudgeonly of hard core racists could possibly object to. Of course it may not accurately reflect society but that is not the intention.

An prime example is the recent visibility on UK television advertising of mixed race couples. If a company is trying to sell its cars/food delivery/mobile phones/banking services or whatever, there is good chance that their commercial will feature a loving couple with different racial heritages.

As part of a mixed race couple myself, I am amused that advertisers now see us as cool and trendy but I also recognise that, outside of London at any rate, partnerships of this kind are still pretty rare. It's nice to be "included", even if there aren't actually many of us. Still, if it stops people outside London staring at us as though we'd just landed from Mars, then it must be a good thing.
I chose ages to minimise the difference in age between individual members of the population and my selected representatives.

You can, of course, do it any way you like.
I've just sat down and drawn a bunch of triangles - a bunch of \/\/ triangles for your way and a bunch of \n/ triangles for my way. You can measure the area beneath them easily. Your way comes out better, by 9/72 to 10/72 where lower is better, so I'll shut up. (Trying to recognise that age distributions are not uniform, I think your way comes out marginality better still.)
Minimising what exactly? Minimising the total of age differences is all very well, but it is not clear to me that age differences are sensibly additive. Are twenty people with an age difference of one year just as bad as one person with an age difference of twenty years? Or better, or worse.
Maybe it's a bit of an angel-pinhead-dance thing.
I see there is no actual acknowledgement of the source of the picture that you have taken other than a possible indication if you actual click on it. As someone who is so vocal about people taking his pictures without due acknowledgement and/or permission this would be surprising other than it is part of a continuing pattern of this poor behaviour.

dg writes: If you hover over the picture, its source is given.
Duly hovered - revealing Engage Barnet.
64.1% are classified as some sort of white in the London Borough of Barnet (Wikipedia).
From the website "Engage Barnet", DG has reproduced only a section of a strip that runs along the top of the home page. I counted 16 figures out of which 9 seem to be white i.e. 56%. And selecting a few more pages I counted 20 out of 30 i.e. 66%. So I conclude this web site is doing a good job of reflecting the ethnic distribution of the population it serves.
Because we're basically choosing representatives from a population, I thought of using the D'Hondt method of recent fame for figuring out who to choose.

If I use the broad ethnic categories in the census ("White", "Mixed", "Asian", "Black", "Other") I get the same first 6 as you (Asian being 4th and Black being 6th). You'd need 18 representatives to get the first Mixed person, and 27 for Other.

Using the narrower ethnic categories, which split for example Asian into "Indian", "Pakistani", "Bangladeshi", "Chinese", and "Other", the first Black (African) representative comes in at number 8, and Asian (Indian) at 9, with an "Other White" at number 4.

You could do a similar thing with ages if you accept that the picture is basically going to differentiate 3 categories of "young", "adult", and "elderly", and pick some thresholds.
But which one is transgender...?

dg writes: Read the post.
Thanks for your no~nonsense approach to the topic. I admit to clicking on the comments section with some trepidation and was stratospherically delighted to see it swimming in discussions about maths, stats and interpretation methods rather than my imagined tirade of abuse.
Can it be true? Are we still a civilized community? Thankfully yes. Egad!

TridentScan | Privacy Policy