Do differences between men and women disappear in more gender equal societies?

There has been discussion in some media articles recently about gender differences between men and women, and whether those differences are inherent or created by society. I’ve tried to summarise below the evidence from psychology about which differences between men and women are consistent across cultures, and which persist or are increased in more gender equal societies.

The first link below is a journal article by Mac Giolla which takes measurements of personality differences in men and women in several characteristics (agreeableness, contentiousness, openness to new experiences, extraversion, and neuroticism, the latter defined as variability of moods and tendency to worry) – qualities that are commonly used in psychology descriptions of personality, and are referred to as the “big 5.”  If you scroll down to Figure 2 in that article, you’ll see five scatterplots. Each scatterplot has a set of light points (females) and dark points (males) with a line of best fit for both. The vertical axis is the average score for that personality trait in a country, and the horizontal axis is a measure of gender equality for that country. Each dot is the average score for males or females in that country. As you move from left to right, the line of best fit for each gender separates, i.e. the average difference between men and women gets greater, not smaller.

I can’t lift the graphs one at a time, but here is a screen shot of two:

scatterplot gender differences

From Mac Giolla 2018.

A paragraph in the results says, “The results indicate that women are typically more worried (Neuroticism), social (Extraversion), inquisitive (Openness), caring (Agreeableness) and responsible (Conscientiousness) than men, and that these differences are larger in more gender equal countries.” (Emphasis added.)

This is the reverse to what most of us have been led to expect over the last 40 years. Most of us have been lead to believe that personality differences are bought about by social conditioning, and if we lived in a more gender equal society, these differences would be reduced. This is not what the data shows.

The discussion shows correlations between higher gender equality in a country and increases in the differences between the sexes in the relevant variable. The “mean” is another word for arithmetic average. The degree of correlation is shown by “r” where zero would mean no correlation and 1 means a perfect correlation.[1]

The article by Khazan (2018) includes a scatterplot of women in STEM graduates vs gender equality for several countries.

scatterplot 2

The more gender equal the society, the fewer women take STEM degrees. Why? She says that one possible reason is that in very gender unequal societies, taking a STEM degree is a sure path to a higher income for girls, whereas “Countries with the highest gender equality tend to be welfare states, [with] a high level of social security… It’s not that gender equality discourages girls from pursuing science. It’s that it allows them not to if they’re not interested.”

The article by Brooks summarises a book by David Schmidt, a professor of evolutionary biology. Schmidt compares cross-country data on gender differences for 28 characteristics. In 20 of the characteristics, the differences between men and women grow bigger in more equal societies. Two characteristics narrow in more equal societies (the tendency to value resources in a mate, which become less pronounced for women in more equal societies, and interest in casual sex, which increases for women in more equal societies, both bringing women closer to men.) Six characteristics don’t change. Brooks says, “Likewise, men score higher than women for the “Dark Triad” traits of Machiavellianism, Narcissism and Psychopathy. Gender equity has the salutary effect of reducing each of these three rather nasty traits, but it does so more for women than for men, resulting in wider sex differences.”

The article by De Bolle looks at personality differences in adolescents in 23 cultures. The differences are consistent across cultures and emerge around age 12 and converge to adult levels around age 17. The differences emerge around puberty.

Overall, moving to more gender equal societies doesn’t reduce differences between men and women. It increases them. This is hard to understand if differences are produced by ‘the patriarchy’ and social conditioning. On the other hand, it’s easy to understand if there are actual differences between men and women that incline them towards different fields of work or study. More gender equal societies are usually higher income societies, with social safety nets, which means there is a lower cost to doing what you actually want to do.

Now to anticipate some objections. Someone is likely to say, “These are only averages, it makes it sound like all women are more agreeable than all men.” No it doesn’t. Take height (and I’m using height because no one has an ideological position on height, so it’s easy to explain something about graphs without an ideological battle.) We all know that men are on average taller than women. We all know that there is a spread of heights around the average. The next diagram shows the average heights for men and women in the US. Men have an average height of 70 inches with a standard deviation [2] of 4 inches. Women average 65 inches with a standard deviation of 3.5 inches. Where the two graphs cross at 67-68 inches, there are roughly equal numbers of men and women. But look what happens as you go out to the right or left.

scatterplot 3

 

Source: https://www.usablestats.com/lessons/normal

Nobody claims that all men are taller than all women, and nobody accuses anybody else of thinking this. However there is a point to be made here. If you have two groups of people where the average measurement is different, and the standard deviation (the degree of spread) is even roughly similar, out at the far left and far right of the graph, there will be a large discrepancies between how many members of each group have (or lack) that quality to a very strong degree. At and above 75 inches, tall men vastly outnumber tall women. At and below 63 inches, short women vastly outnumber short men. This will be similar in anything that can be measured along a scale where the two groups differ on the average and which have the usual bell shaped curve. In the case of height, an eight percent difference in the middle (5 inches over 65) results in differences of about 20:1 in the ratio of tall men (over 75 inches) to tall women.

Evolution may have selected for women to be nurturing, since that maximises the chance of a child surviving. This is not the same as saying, “all women are more nurturing than all men.” But it means that there will be large differences in the number of women and men who are very nurturing or very indifferent to the needs of others. If men are on average more narcissistic than women, the will be many more very narcissistic men than very narcissistic women.

One aspect in which men and women differ is “orientation to things” versus “orientation to people.” Exposure to androgen – a male hormone – in the womb (because of a mother’s medical condition –  having congenital adrenal hyperplasia) makes girl foetuses become girls with a higher than average orientation to things, so there appears to be a hormonal basis for this difference (Beltz 2011).  If there are some jobs which call for a certain quality or aptitude, and that characteristic has a different mean (average) for men and women, there will be a large differences in the number of people who have (or lack) that quality to a strong degree (i.e., out at the left and right of the curve). If men score higher on “interest in things” and women score higher in “interest in people,” and there are some jobs which suit people who have that quality to a strong degree, (e.g., engineering vs child care or social work) then men more will be attracted to some jobs and women to others. It’s not a binary, and no sensible person would claim it is.  You can (and should) remove obstacles to girls pursuing engineering (if they want) but creating a more gender equal society doesn’t bring about equality of outcomes.

Above, I showed the actual graphs for male and female heights in the US.  The average male height is about one and a half standard deviations above the females. Below I’ve show two bell shaped curves moved apart by one standard deviation. The two graphs both have a standard deviation of two, and the second curve is one standard deviation to the right of the first. If you move one standard deviation to the right again (to the value 14), draw a line, and look at the total number of people (the area under the graph) the number of group 2 members will be six times the number of group 1 (compare the areas under the graph to the right of the black line). To the right of 15, group two will outnumber group 1 by 9 to 1 (draw a line at 15 and look at the areas to the right).  If the two genders differ on some quality by only one standard deviation, there will be a very large difference in the number of women and men who display that quality to a very large degree.

scatterplot 4

In the case of many jobs, for example, this doesn’t really matter. Most jobs don’t require you to have some personality characteristic to an extreme degree. But some do. Engineers tend to be very interested in things, rather than people. Social work, psychology and medicine tend to attract people who are more interested in people rather than things. For a discussion of how different personality types are attracted to different university courses see Vedel 2016, or Ali, which is a non-mathematical discussion of Vedel. Law and economics appear to attract people who score low on trustworthiness and concern for others.

To talk as though apparent differences between men and women are “binaries” which we should reject because ‘binaries are wrong’ is to criticise a straw man. Nobody in psychology claims they are binaries. However even when there is a large overlap, there will be an imbalance in the number of men or women exhibiting a characteristic to a high degree. There is no contradiction between saying that men and women can both be narcissistic (or agreeable or tall) and saying that there are gender difference in whether those qualities are exhibited strongly.

Costa, Terracciano and Mcrae (2001) summarise differences in gender across 26 cultures in surveys of 23,000 people.

To sum up: men and women score differently in various aspects of personality. Those differences appear to be consistent across numerous countries and cultures. The qualities involved are not binaries. They are measured along a scale. The differences are greater, not less, in more gender equal societies. This is the opposite of what most of us have believed for the last 40 years. If patriarchal oppression makes girls be more agreeable, contentious etc, it is unclear why the differences between men and women would be greater in more gender equal societies. If men and women score differently on average for a certain quality, then there will be a large difference in the number of men or women who possess that quality to a very large degree. There is no contradiction between saying “men and women can both have quality x” and saying that “there are many more men or women who are very x.” Gender equal societies reduce the disadvantages to women pursuing classes or jobs that actually interest them, so if a job attracts or requires a very strong degree of some quality, the sexes may, and do, select differently.

-0-0-0-0-

Ali, Aftab, 2016, ‘New study finds link between ‘Big Five’ personality traits and which subject students study at university’ The Conversation, June 21, 2016 https://www.independent.co.uk/student/student-life/Studies/new-study-finds-link-between-big-five-personality-traits-and-which-subject-students-study-at-a6846996.html#r3z-addoor [This article is a discussion of Brooks, Rob, 2016, ‘Gender equity can cause sex differences to grow bigger’]

Beltz, Adriene et al, 2011,  ‘Gendered Occupational Interests: Prenatal Androgen Effects on Psychological Orientation to Things Versus People’ Hormones and Behavior Volume 60, Issue 4, September 2011, Pages 313-317 https://www.ncbi.nlm.nih.gov/pmc/articles/PMC3166361/

Charles, Maria, and Bradley, Karen. 2009 ‘Indulging our gendered selves? Sex segregation by field of study in 44 countries.’  American Journal of   Sociology Vol. 114 pp. 924-976. https://www.ncbi.nlm.nih.gov/pubmed/19824299

 

Costa, P Terracciano, A and McCrae, R 2001, ‘Gender Differences in Personality Traits Across Cultures: Robust and Surprising Findings’ Journal of Personality and Social Psychology, Volume 81 pp 322–331

De Bolle, et al, 2015, ‘The Emergence of Sex Differences in Personality Traits in Early Adolescence: A Cross-Sectional, Cross-Cultural Study, Journal of Personality and Social Psychology, ‘Volume 108(1), January 2015, p 171–185

Giolla  Erik Mac  and Kajonius  Petri J. 2018 ‘Sex differences in personality are larger in gender equal countries: Replicating and extending a surprising finding’, International Journal of Psychology, September 2018 https://doi.org/10.1002/ijop.12529

Khazan, Olga, 2018, ‘The More Gender Equality, the Fewer Women in STEM’ The Atlantic, Feb 18, https://www.theatlantic.com/science/archive/2018/02/the-more-gender-equality-the-fewer-women-in-stem/553592/

Vedel, Anna, 2016,  ‘Big Five personality group differences across academic majors: A systematic review’ Personality and Individual Differences Volume 92, April 2016, Pages 1-10

A newspaper article which discusses the topic without the maths:

https://www.thetimes.co.uk/article/patriarchy-paradox-how-equality-reinforces-stereotypes-96cx2bsrp

 

[1] For those who have forgotten their stats, r is a measure of correlation, i.e., when one thing increases, does the other thimg increase or decrease? A positive value for r means that as x increases y increase, a negative value would mean that as the x thing increases the y thing decreases. (In this case, x is the measure of gender equality in the society – the bottom axis – and y is the measurement of the personality characteristic, the vertical axis.) a value of one means a perfect correlation. Zero means no correlation.

[2] The standard deviation is a measure of how spread out the bell curve is. In a normal distribution, 68 percent of the cases occur within one standard deviation of the mean (the average), and 95 percent of the cases occur within 2 standard deviations of the mean. So if men are 70 inches tall on average, with a standard deviation of 4 inches, 95 percent of the cases will fall between 62 inches and 78 inches.

Advertisements

Where Does Moral Thinking Come From?

Jonathan Haidt tackles a big question in his book ‘The Righteous Mind.’ Haidt, a psychology professor who has done stints as a speech writer for democratic politicians, looks at why democrat and republican voters in the US seem never to agree on almost any moral issue.

Book cover righteous mindAfter doing extensive opinion polling and interviewing voters to get their reactions to hypothetical moral situations, he finds that people seem to take five factors into account in forming moral judgements. These are: (i) Is anyone injured by an action? People generally avoid doing harm to others, and see it as wrong. (ii) Fairness. – not cheating, not taking more than you are entitled to.  (iii)  Loyalty to a group. This comes from our early origins as tribal creatures.  People who show disloyalty to  organisations, sporting teams, the army or their country are often condemned. (iv) Respect for legitimate authority. (v) Showing respect for ‘sacred’ objects, such as national flags, or religious objects, and avoiding ‘dirty’ or contaminated things.  A person who places a lot of weight on this factor will disapprove of burning national flags, and, for example, disapprove of the photograph ‘piss Christ’, by Andres Serrano, which shows a crucifix in a jar of urine.  More details are available here  here  here.

Interestingly, when Haidt asks American voters to rate themselves on a seven point scale where 1 represents a very liberal (in the American sense of progressive or democratic voter ) and 7 represents very conservative, he finds an interesting result. Voters who rate themselves as 1 on that scale use criteria (i) and (ii) in their reasoning, but give almost no weight to the other three. And as you move across the political spectrum, the emphasis on  the last three factors rises steadily People who rate themselves as very conservative give all five factors roughly equal weight.

Some of this jells with other things I’ve read elsewhere. For example, conservatives are usually strongly opposed to pardons to people like Chelsea Manning who leaked documents from the army. This relates to criteria of loyalty and respect for authority. I’ve read that having strong reactions of disgust at photographs of ‘unclean’ objects such as meat with maggots, or a person treading in dog poo is a strong predictor of voting conservative – this relates to the criteria about avoiding things that are seen as disgusting.

Haidt claims one reason why democrat politicians fail to appeal to conservative voters is that their speeches and advertisements only ever appeal to criteria (i) and (ii), and leave out the other two. He tried to persuade democrat politicians to include more of the last three factors in their speeches, with limited success. Progressive and conservative voters are, in a very real sense, speaking a different languages.

Haidt also discusses the role of religions in moral thinking. He believes they act as reinforcers of moral values by giving  a stamp of approval to rules that promote social cohesion, such as don’t kill, don’t steal, don’t lie or cheat, or commit adultery, and so on. All societies have these rules, because you need them to keep harmony ain a primitive society that relies on group harmony to be able to function. All religions promote these rules because in Haidts view, that’s what religions are for.

It’s an interesting read, and I strongly recommend it.

Political Viewpoints are not Just Like Food Preferences

Recently on a social media thread I was commenting on, I said that I said that I was bothered at how people seem to adopt political opinions in ‘bundles’ without thinking individual policies through based on evidence. They buy the centre-left bundle, or the libertarian bundle, or the conservative bundle. Someone else commented, “An opinion is just that an opinion. It is not truth or fact and doesn’t need to be. I hate okra. My opinion is based on the “fact” that it is slimy and tastes bad to me. What if I said okra is bad, that’s a different story but still it’s still based on my personal view. ”

Soup from pixabay dot com
Not a political opinion. 

The notion that political opinions are just opinions like food preferences lies at the heart of what is wrong with much of political discussions. When it comes to describing political problems and proposing solutions, there usually are relevant facts. They may be sometimes be hard to discover or hard to weigh up, but some relevant facts usually exist and we should make the effort to know them before we advocate positions.

There are actual facts available on police shootings of black people in the US, and those statistics can and should influence your view on whether there is a real problem that needs to be fixed. If you deny the existence of a problem when the stats say there is a problem, then you won’t take action when it should be taken, and if the stats were to show that there isn’t a problem then you would be making major changes and spending money to fix a problem that didn’t exist. (Note here, and below that I’m talking about the relevance of facts to the policy you do or don’t advocate, I’m not – here- adopting a position on BLM, I’m saying that relevant facts usually exist.)

When environmentalists advocated removing CFC gasses from refrigerators and aerosols in the 1970s and 80s because CFC gasses were damaging the Ozone layer, those claims about CFC damage were either factually right or they weren’t. If the factual claims were wrong, then an incorrect policy was being advocated. to solve a problem that didn’t exist. Refrigerators and aerosols would cost more to consumers when perhaps they didn’t need to. If the factual claims were right then the policy being advocated was correct and the need to fix the problem was real. The facts actually mattered.

In relation to abortion, there are many factors that might influence your position. One of them might be when a fetus becomes pain capable. If this is correctly claimed to be at 24 weeks, this may be a factor you want to take into account on when, if ever, you allow abortion.  If someone falsely claims that a fetus becomes pain capable at 10 weeks, this may make a difference to your position, and you would then be adopting a position on the basis of incorrect claims of fact.  There might be many other factors you want to take into account, but if you are influenced by a false claim of fact that’s really not a good thing.

In relation to NAFTA and whether you oppose further free trade pacts, it actually matters whether the people who have lost their jobs to cheap imports would have lost them anyway due to robotisation. A belief (for example) that free trade benefits countries as a whole but that free trade agreements should only go ahead with generous retraining packages for workers who will be displaced, and incentives for new industries to be established in affected areas is a position that involves millions of dollars of expenditure. If such policies are advocated or opposed, it actually matters whether factual information exists and whether it supports one side or another. It’s not the same as ‘I like pumpkin soup but I don’t like potato soup.’

If you adopt the position that opinions are all just like food preferences, then politics just becomes a process of tribes screaming each other about soup, resulting in whoever screams the loudest getting to spend billions of dollars on policies that might or might not work at fixing problems that might or might not exist.

Note that I’m NOT starting a debate here about BLM, abortion, or NAFTA. My point is that on almost all political and social issues, relevant facts exist. A disregard for facts leads to mere tribalisation of politics followed by bad policy and expenditure on wrong policies when that money could be spent on policies that really would fix real problems. If we think that social policies are just opinions about soup, we are in a lot of trouble, and that thinking needs to be challenged.

 

 

Edit

WHY AUSTRALIAN GUN CONTROL JUST WON’T WORK IN AMERICA. 

In 1996 a lone gunman killed 35 people at Port Arthur, a popular tourist site in the Southern-most state of Australia. A wave of national revulsion swept through the country, and shortly after, a conservative Prime Minister and the six Australian states (lead by both conservative and labor state premiers) agreed to massive restrictions on gun ownership, and commenced a taxpayer funded $500m buy back of guns from Australian residents.
Now to be clear, Australia did NOT ‘ban all guns’. If you see someone on Facebook claiming that, they have no clue what they are talking about. You can still buy a non-self reloading shotgun and go duck hunting in Australia. You can still buy single shot long barrel rifles and go rabbit, fox, kangaroo or pig shooting in the bush. You can still join a target shooting club and have a handgun of the sort used in competition target shooting (and that includes semiautomatics, although you have to use the club’s guns for the first six months.) Australia does send target shooters to the Commonwealth and Olympic games so we have to have such clubs, right? About five percent of Australian households possess a forearm, compared to about thirty percent in the US, and let me repeat, because this is the essential point.it won’t be a semiautomatic.
Yes, all gun owners need a permit and each individual gun is registered. The biggest change that Australia introduced was the banning of semiautomatic rifles by civilians with only a couple of  exceptions: farmers (who periodically need to destroy livestock after fires and floods) , and professional shooters (many of whom do feral animal culling). The rifle magazine sizes are limited to ten. There s a couple of other provisions for gun dealers and collectors but they affect few people . Semiautomatic rifles are what you normally need to carry a mass public killing. The important point is that the “average person” can’t just go down the street and buy a semi-automatic gun.) It’s natural that people in the US, on both sides of the debate, simplify and distort what we did.
Australia hasn’t had a mass public shooting (normally meaning four or more dead in total) since. So why can’t the US just be sensible and do the same? For several reasons.
Australia has only six states and there is relatively greater degree of social homogeneity and similarity between the states.  The greatest difference between states and major cities in Australia is nowhere near the difference between, say, San Diego and Louisiana or between Arizona and New York. (I’ve been in all four places as a tourist.) Being more homogeneous, Australia is a more cohesive society, and reaching consensus is easier here. One of my American friends once said, “The thing you have to understand is that America isn’t one country. It’s fifty different countries.” Australia has numerous migrant groups, but our immigration is predominantly skills based, and no one migrant group dominates. We don’t have a dominant non-Caucasian group in the way the US has Latinos and African Americans.
The power balance between the state and federal governments is also different. The state governments have less power relative to the feds compared to states in the US. Australia isn’t like Britain, where regional governments are the creation of the national government and really can’t resist it. . Australian states exist in their own right, but the list of powers reserved to the federal government by the constitution is more extensive than in the US. If we made a scale with Britain on 1 at the left hand end and the US on 10 at the other end, Australia is towards the US end of the spectrum, but not all the way there. Maybe it’s a 7 or an 8. American states regularly take the national government to the Supreme Court; Australian states do, but not so often.
Australians don’t have the mind frame that Americans have where the national government is seen as something that is going to oppress us, and that we might have to overthrow by violence. Another American friend once said, “We are a nation of extremists. We had to be, or we wouldn’t exist. We were born of a revolution in the wilderness.” The mind set that you need arms to potentially overthrow your own government is something Australians don’t understand, because Australia wasn’t born that way. In the 1890s, the Australian colonies said to Britain , Hey, we feel like making our own country,” and Britain said, “Oh yeah, alright, hold yourselves a conference and write a constitution.” So everybody had a nice cup of tea, and we did. The American attitude of loathing and distrust of the national government looks incomprehensible to Australians,
The Australian voting system (preferential voting (or ‘instant runoff’ as Wikipedia calls it) in the lower house and proportional representation in the senate and state upper houses) forces major parties to stay relatively close to the political centre, and forces major parties to negotiate with each other or with minor parties to get things through the senate, and compulsory voting means Australian major parties focus on persuading the middle twenty percent of the population. The American system forces parties to focus on getting their base to physically turn out and vote, and cater to the fringe, and this results in different strategies and postures. In a normal senate election, we elect six senators per state, and each party gets one senator for each 14.29 percent of the vote (a “quota”) in that state. Normally both the conservatives and labor get two whole quotas each and hence get 2 senators each. There will then be several minor parties and independents with fractions of a quotas each. There’s then a method to determine who gets the last two places where nobody has a quota in their own right. Currently a quarter of the senate seats are held by minor parties and independents. This makes cooperation between parties necessary.
I feel much happier living in a country where I have only one fifth the chance of being murdered than a similar person in the US. But telling Americans to do what Australia did just isn’t going to work.
And here are some links to articles that anyone should read if they are not familiar with this issue.   The first is the article “What it’s like to own guns in a country with strict gun controls” from Time magazine, here. which gives the best factual account I’ve seen of Australia’s gun laws. The second is “It took one massacre: how Australia embraced gun control after Port Arthur” here ,from the Guardian, which describes what happened in the six weeks after Port Arthur. The third is Wikipedia’s article on “instant runoff voting” here. The fourth is “The rate of all suicides and homicides in Australia has declined since the gun buyback” from the Sydney Morning Herald which gives accurate stats on homicide rates before and after 1996 here .The fifth is Wikipedia’s article on “list of countries by intentional homicide rates” here.  (A lot of inaccurate nonsense is talked about murder statistics in Australia, compared to other especially by non-Australians who don’t understand the stats, and are believing what they want to believe.)

Finally, Some will claim that most gun deaths in the US are from suicides (this is correct). But the figures I quoted above (one fifth the chance of being murdered in the US) are based on intentional homicides so the suicide data is irrelevant. Some claim that British and Australian statistics are complied such that to die from murder, you have to die at the scene, but if you die at the hospital later you died from the wound, so it doesn’t appear in murder stats. This is incorrect. British law used to say that a violent act was murder if the victim died from the effects of the attack within a year and a day after the attack, so obviously if they died a year later they didn’t die at the scene. The year and a day rule was abolished in 1996, so the victim might now die after two years in a coma, but it’s still counts a s murder. Attackers are routinely charged with murder in Australia when the family of the victim turn off life support in hospital, so obviously the victim didn’t die at the scene. It’s still a homicide. I don’t know why these false claims keep circulating on the internet, other than confirmation bias.

Just another WordPress.com site