Sexism and data: When statistics hurt women | Crunched

Statistics are sexist. How dare you. Numbers are neutral. I mean, I don’t know about that. We’ve both been looking at
this book, Invisible Women by Caroline Criado-Perez,
which makes, I think, a lot of strong points that
averages, aggregate statistics, are understood by most people to
be representative of the whole. But they actually
contain a lot of biases, especially around
the topic of gender. I’ve been looking at
transport as a topic. When you look
beneath the surface, the way that people
travel around does actually have some quite
distinct gender-based patterns. So I’m going to use some
of our handy number blocks now to show the different gender
patterns you get in transport. So this is from
some Euro stat data. So this is representative of
the whole EU; that’s the EU 28. They looked at the
percentage of men and the percentage of
women who travel around on a typical day using
different forms of transport. They found that with cars,
59 per cent, ten and then I’ve got a nine
at the end, travel by car on a typical day
versus 49 per cent of women, quite a margin there. They then looked at
public transport. So this is including buses and
the underground trams, ferries, that kind of stuff, and they
found that 15 per cent of men travelled by public transport
on a typical day versus 22 per cent of women. If we then look at walking
around, 11 per cent of men walk on a typical day
any substantial distance and 17 per cent of women. So road use through private
driving, through cars, is male dominated. Public transport is
female dominated. Now, why is this an issue? Well, as highlighted
in the book, this becomes an issue when
people just think of transport as, oh, yes, it’s mainly roads. And so when the
government decides, all right, we’re going to
build loads of new roads, we’re going to
pile loads of money into the roads,
what they’re doing is they’re essentially most
of that money is going to men. It’s making men’s
daily lives easier. Most of that money is
not going to women. Now, when you view
that alongside the fact that over the last seven
years, the amount of money the government gives
to local authorities to spend on their bus
routes has almost halved, it’s gone from 375m which
I’m just going to write here. So spending on buses has gone
from 375m in 2011 to 200m. So what you’re looking at
there is spending on male dominated forms of transport
has continued to be strong, but these huge cuts
have come to buses which are part of
this form of transport that is predominantly
used by women. And we’re saying that this
disproportionate allocation of resources is because we see
the data as gender neutral, whereas it isn’t? Exactly. And so that creates
some inherent sexism. Right. This is the gender data gap. So people here, oh,
more money for roads, less money for buses. And they think this is just
a cars versus buses thing. Whereas embedded in that is
a male versus female thing. I can think of another example
of a sort of gender data gap, and it relates to poverty. When we look at
poverty statistics, we’ll say for example, in
the UK, I think roughly 22 per cent of the population
lives in poverty. And sometimes it’s
broken down by age. So we’ll know how many
children live in poverty, how many adults, and
how many pensioners. And those are considered the
most important statistics when it comes to poverty. But then, when you
dig in the data, you see that there is a
considerable difference between what share of the
population that is poor are men and what share of the
population is poor women. Let’s see some numbers. OK. So these are poverty statistics
from the UK’s Department of Work and Pensions. We’re looking at the percentage
of the population that is poor that is men and women. We’ll start in
1994, ’95, and we’ll end with the last financial
year for which the data is available. And we’ll start from 25 to 43. So this is what
happened with men. We started out at just 29 per
cent, 30 per cent actually, and then it went up and
up, financial crisis, and then it’s
gradually going down. And now we’re at
about 30, what is it? 34. 34 per cent yeah. For women, in ’94, ’95,
it started at 39 per cent. It sort of remained flat,
went up a little bit. But basically, it’s
been flat, and it’s now where it started at 39 per
cent but it’s consistently been higher than men. So again, I guess
the issue here is when we hear people
talking about, oh, poverty is a problem,
we need to reduce poverty, they’re not thinking
about how poverty affects the genders differently. Also, it brings us straight to
the core of the gender data gap because the way that this is
measured or at least the way that the UK government
measures poverty by gender, is that it looks at the
head of the household and look at that
breakdown by household. But we know from separate
statistics, collected by a different government
in India, that in that case, the majority of poor
women live in households that are not considered poor. So that’s because of very
complex gender dynamic. However, it does show
that it’s not necessarily an adequate form
to measure poverty. Instead, you should be
looking at individuals. So this is an example
of the gender data gap where apparently it looks
like we’re being neutral in the way that we
measure numbers, but actually we’re not taking
into account very important gender dynamics that
happen, for example, that the majority of the
head of households are men. Yeah. And even a more shocking
example of the gender data gap can be seen when we look
at health statistics. So for many, many
years in the US, health tests were run
with only male subjects. This happened because of
the thalidomide scandal in the ’60s, when
loads of pregnant women were prescribed medicines
for morning sickness that ended up causing loads of
health defects on their babies. So because of that, tests
were run solely on men, including tests for
ovarian and breast cancer. Now, this was eventually
worked out in the ’90s, but still, to this day,
many academic tests are only run on male subjects
or the gender breakdown is not provided. So this means that we will
get a lot of health advice that tends to be “gender
neutral” but doesn’t look at how male and female
bodies are different. And you were citing one earlier
that was really interesting. Yeah. I’m also just
curious about how you assess ovarian cancer in men. Right. Men can have breast
cancer, but ovarian cancer, I’ve never heard of an example. Interesting. But yeah, you’re right. So a really good example
of how this plays out, the effects of these
male dominated studies, is that in 2016, the
British Medical Journal found that young women were
almost twice as likely as men to die in hospital. Now, that obviously begs
the question of, OK, why would that be? What’s happening there? And a big factor
is the fact that separately a study by the
American Heart Association found that several risk
prediction models that doctors and hospitals use
to assess what’s wrong with the patient,
especially those with acute coronary issues, so
issues relating to the heart, were developed in
patient populations where the patients involved,
2/3 of them were men. So you’ve got studies being
done to see how should we deal with things
like heart attacks, where 2/3 of the
subjects are men. So the average of
that is obviously going to be skewed
towards men, and it leads to situations
such as those discussed in the book, where often the
symptoms for a heart attack present very definitely
in men and women, and so women with heart attacks,
that heart attack is being spotted later, if at all,
and then people are perhaps responding with treatments that
were specialised around men. So yeah, this has real outcomes
that affect women much worse than men. So we’ve looked at health. We’ve looked at a
key economic metric, and we’ve looked at transport. I think, in a way, we have
proven that, yes, data can be sexist in all walks of life. Now, the thing many
critics could say, well, if you break down data,
where do you stop? For example, when you
look at the pay gap, the ethnic pay gap is
very important as well and also the class pay gap. You could go on forever. Where do you stop? Right, and I guess especially
with the ethnic pay gap, even when we break
out, for example, black, Asian, and
minority ethnic groups, that’s still a
single number that covers people with a hell of a
lot of different experiences. And it doesn’t necessarily break
it down by gender in itself. So for example, looking at
Asian men versus Asian women, yes, you could go on forever. Maybe you should in
some ways, but until you get to a sample size
that is too small. Sure. But then, I guess,
what we’re also saying is even if you’re not going
to break the data down to that fine detailed level,
decision makers should at least be aware that when
they’re looking at an average for
a large group, they need to be aware that a lot
of people’s lived experience, a lot of groups’
lived experience, is very different
to that, and they need to know that a single
treatment might be privileging one group over another. And even the ordinary
neutral person, when they see a
statistic that supposedly represents an average,
they should think twice. Absolutely.

59 Replies to “Sexism and data: When statistics hurt women | Crunched”

  1. No you have not proofen anything, except that you don't have a clue of how statistics work, with all due respect.

    Most examples are not a sign of bias, but simple facts. Spending is NOT by gender. It's by NEED. Travelling by car is not a priviledge it's a hassle. Also if spend is on something that a majority of men use, it is still not by sex (a minority of women still use the same so its to their benefits too). Also most people live in families and contribute back to those families. So any spend to benefit one familiy member benfits all other members.

    The whole video is premissed on hogwash.

  2. the transport argument is slightly flawed as a counter argument could be businesses use roads, most likely using cars or vans, thus the government paying more into private road transport is more beneficial than public transport, it just so happened that men use cars more often than women…

  3. 1) No one is stopping women from using cars. Women have the same opportunity to use a car as men do, so it's fair.
    2) how is it that the majority of the poor are women yet the majority of homeless are men?
    3) I'm a little sceptical about those poverty statistics…
    4) we need segregation to solve all of these issues, we should have a women's hospital, and women only business departments, then the women can take responsibility for their own future.

  4. The issue here is more around data collection, processes to ensure good quality data and the breakdown of the sample rather than the data itself being sexist.

    Also, in society you aren’t likely to have equal numbers of men and women participating in all areas of life, so certain policies might slightly impact one gender more than the other, but no one is preventing more women from driving rather than commuting via buses. Perhaps there are also other reasons why there are differences, such as men working further from home than women. Ultimately the data shines light on these differences.

  5. This is some serious nitpicking.
    You do realise that busses use roads as well right? There are also statistics that favour women over men. One concerning one is the rate of suicide amongst men or drug and alcohol abuse. These statistics aren't helping get more funding for mens services. As a matter of fact they are non-existent compared to the amount that women have access to.
    Statistics being sexist is not mens fault. Don't pull out data examples that only prove how gender biased data is towards men and then say that it's sexism. I'm sure there's plenty of data in other areas to balance out the notion of who is perpetrating the sexism. There's plenty of female statisticians.

  6. Cuts in funding to bus services affects bus users who are disproportionately female.. and (for example) cuts to the Crown Prosecution Service affects accused criminals who are diprotionately male. Of all the interesting statistics to attempt to correlate with these trends (e.g. carbon emissions, quality of life metrics, urbanisation rates, mental health stats, crime rates etc) the gender angle is remarkably boring.

  7. Women are supposed to be mothers, givers of life, not second rate wage slaves. These feminists are misogynistic because they hate the idea of femininity and want women to become masculine.

  8. Heres another statistic: IQ spread. Female IQ gravitates towards the mean more than male IQ, which explains why there are far fewer female prisoners and also far fewer female geniuses.

  9. road transport is male dominated 'so' all money to build roads is benefiting men 'so' because its men its sexism. Flawless logic, I've cracked it. If men are involved, it is sexist 100% that's a statistical fact.

  10. Lmao…I have now seen it all. Buses use roads just like other cars. Also, he hasn't taken into the fact that their are other forms of transports that just buses. Take trams for example. This could explain the cuts in bus routes.

  11. There is not a more entitled majority in the history of the world than women. And, under-performing and higher polluting I might add.

  12. Is it possible that cars are prioritized over public transport because car owners contribute more to the GDP (and I'm saying this as a public transport user)?

    Also, making a title something like "Unpacking gender disparities in statistics" would have saved you from most of the dislikes.

  13. Its not the data/statics that are sexists… its the failure of the data collector or the limits of the collected data that restrict the usefulness of the conclusions derived from the current data set. Bias such as "sexism" is something that should be able to be mitigated through a thoughtful study and thorough discovery (data collection). There are loads of reasons why limitations might arise: limited access to availible data, lack of historic records, resistance to data collection (such as privacy concerns)… Then there are limits on the data analysis: data say nothing, sets are always interpreted.

    Numbers are numbers… data is data: without knowing how the data was collected and under what criteria it was gathered, we can't use that data correctly. Data are often abused by "showing" conclusions that the methods of collecting and the range sought can never justify.

  14. The was a statement on a BBC. The are lies, total lies and statistics. I can prove the roads are not built for men. On the road I live 70% of the drivers are women, but is that because there’s a school in my road. Also roads are not built for Joe/Jill public but for big corporations to transport their goods.

  15. Does not matter if the man or woman drives. A driver takes children to schools, wives or husband's to their jobs. It is only more common for men to drive in a historically predominantly patriarchic society. 59 vs 49 needs interpretation. Data is complex whereas your brain seems not to be.

    Also buses use roads. And road infrastructure attract investment, public transport only leads to real estate price exploding.

    Sexism and Feminism can not be copy pasted on all topics.

  16. Feminist grabs some statistics and starts interpreting.

    Guys, gender ratio using cars or public transport can vary but fact is investment into infrastructure like roads has a different economic value than investment into buses and sidewalks.

    One attracts foreign investment and the other attracts real estate gamblers making houses in affected area more expensive.

  17. We welcome meaningful comments and suggestions. We do not tolerate hate speech or personal attacks. Thank you for keeping this space respectful.

  18. On transport, you could split it drivers by religion or skin colour, and suddenly your investment in roads is stated as being "religiously" or "racially" biased. What utter dross!

  19. A lot of the comments are strawmanning the presenters here. All they are trying to say is "sometimes there are disproportionate gender splits in aggregate statistics that often get ignored", and then offering some examples showing how this crops up even in places you might not expect.

    They're not really saying "therefore changes to transport funding are sexist"; nor are they doing any detailed analysis or making any conclusions. Of course the real issues and the decisions that need to be made are complex and need to consider lots of other factors – all this is saying is "lets try to make sure this factor is considered".

  20. 1)the data for women doesn't even add up to 100% only to 88% which already questions the legitimacy of your data
    2)by your data more women use cars than other 2 categories combined meaning the most efficient way to benefit women is by investing in car area (whatever that means and assuming we even need to do that)
    3) buses use roads as well
    4)different areas require different amounts of money. To give you an example the amount of money the goverment spends on pencils will be lower than the amount spends on fuel. That doesn't mean working people are being discriminated in comparison to commuting people. There's just a certain amount of money you can spend on pencils for goverment workers efficiently
    5)roads are very important for the economy hence it improves everyone lives

    Now for the personal criticism:
    1) I only watched 2 minutes and during that time you managed to be so damn wrong in everything you say that it makes me question whether you're doing it on purpose as a form of art
    2) you clearly have no analytical skills, your statistical data is questionable and you have no understanding of correlation vs causation
    3)you find problems in areas where it doesn't exists creating more seperation between the 2 genders instead of bringing them together as just humans
    4)you should be fired for such an incredibly bad job
    Wtf is this FT you should be credible

  21. As much as I support equality, this isn't an informative video, nor does it do anything to promote any of the issues it seems to be attempting to supporting. It's just some bloke barraging a woman with a series of false equivalences and other logical fallacies. "This variable influenced X so X must be Y. If Y is the case then it must be due to <negative bias>." While the concept might have been worth a look, the actual argument falls apart, quite literally, in about two minutes. Which is impressive.

Leave a Reply

Your email address will not be published. Required fields are marked *