Non-Linear Relationships

While we are a bit on different planets in terms of how we interpret the world, I do think there are some opportunities for bridging the gap more.

It takes some time for me to get clear to myself and find a way to articulate the difference. But I think I have some better ideas now.

I agree that my worldview is data driven. What I would like is to find the causal environmental variables that explain test score gaps. What is it that drives the test score gaps?

Okay, looking at the various examples you bring up through the rest of your post regarding abortion, gun-crime etc, WHAT I see as the issue in how you approach all these issues is that you try to plot linear relationships against single variables.

Most of these issues I don’t think are linear in nature. For instance if everybody has a gun, doubling the number of guns are not going to double gun crime. We see the same with IQ score. For very bad environments, improving the environment will have a noticeable impact on the IQ score. But once you live in a good middle class home, doubling their income, increase length of education etc isn’t going to have much impact. That is why we see heritability for adopted kids are so high. Adoption families are usually high quality families. They reach a threshold. In other words we are not dealing with linear relationships.

I believe somebody also found that the chance of winning a noble prize, does not increase above the IQ of 130. Hence to solve really hard problems there is not clear linear relationship between IQ and achievement.

My World View is Model and Data Driven

I would say that my world view is also data-driven, but not the same way yours seems to be. For instance if you observe planets on the night sky and their position, then some simple linear plot against time would not clearly show the relationship.

You need a model that captures that we are in orbit around a sun, which creates a complex relationship between observation of positions of other planets.

I believe all interesting things about human society is the same. They involve complex models, where you cannot just dump data into a simple linear model. Before plotting data, you need to have developed some sense of the mechanisms which work behind the scenes. You need to have a sense of how a particular kind of input will produce a particular kind of output.

In most of the cases I reflect upon, I almost always end up concluding that a substantial number of the factors which will influence outcome the most, cannot easily be measured.

My take on your approach seems to be that you take the opposite approach. You don’t consider the model first. Rather you pick whatever data you can find and start to hack away at it looking for relationships.

You disregard things which doesn’t have clear measured values, because you deem this too subjective. The problem with this thinking is that a lot of the data you disregard while not crisp and clear is often some of the most important stuff. Your conclusions are not any more valid by taking crisp data values of limited influence on the phenomenon studied than if I take into account less crisp data which non-the-less can be assume to have a more significant impact on the phenomenon studied.

So the difference is perhaps, that we both care about data, but I look at much broader set of data, which cannot necessarily be easily plotted in a graph.

Models Are Complex Not Simple

For instance, if the black/white test score gap was a function of the number of racist Republicans in a city, then the gap should be larger in red states and smaller in blue states.

This is an example of what I am talking about. This is a far too simplistic and linear model of how racism works. It is not like some SimCity game where X number of republican citizens emanate racism points which then hit little black sims which cause their test scores to drop.

It is a long term process. It is racism working over long term shaping people and society. If you went to an area and used a magic wand to turn every white person non-racist, that would have minimal if any impact on African-Americans living there.

The structures built up by years of racism would still be there. The racial dividing lines of neighbourhoods. The poorly funded black schools.

Oh, and yes some people claim school funding doesn’t matter. Again people employ entirely linear models. If this was true a school with $1 in funding and would do the same as a school with $10 000 000 in funding. Nobody in their right might believes that. Again these things will be non-linear. You need to reach a certain level upon which point further funding is unlikely to improve the school markedly.

Often it isn’t even the quality that matters but the prestige. Why are Ivy league schools so insanely well funded if money doesn’t matter? Research shows these schools are not actually that good. It is because in particular in America society it is not about the skills you have but the pedigree. Being able to show an Ivy league school on your CV has tremendous impact on your job prospects. It is also a question of networking. You get in touch with a large number of important, rich and well connected families.

While funding as school well does not necessarily raise its quality much if anything, it does allow that school to engage in far more creaming. Better students and richer students are going to want to go there.

I’ve computed it county by county, on average the gap is the same, whether the voters all red or all blue.

The problem with such simple linear analysis using only a single variable is that you don’t take into account other important factors that tend to follow voting patterns and which will work in opposite directions.

Below is an example of how such a simple analysis can go wrong.

Image for post
Image for post
Simple example of abuse of statistics

Anyone looking at this graph the way you look at black-white test score gaps and voting habits would conclude that Democrat politics cause COVID-19 to spread. Or they may conclude that Democrats are more reckless in social distancing or mask wearing.

Here is where having a broad view of data matters. I take into account prior knowledge which is not necessarily crisp data. I take into account data you may dismiss as simple anecdotes.

I know from experience in terms of numerous articles written, my own travels etc that more densely populated urban areas tend to vote democrat while rural areas are more likely to vote republican.

Hence what this graph is really showing is actually that COVID19 spreads more easily in densely populated areas than in thinly populated areas. In other words how every other disease works.

We also know from news that Republicans are far less likely to wear masks and observe social distancing. That is important data, even if we don’t have clear crisp statistics of it. If you don’t take such less crisp data into account you will end up with the completely wrong conclusions.

Berkeley, California prides itself on being incredibly progressive and antiracist, only 3% of people there voted for Trump, I’d imagine that every house has a “Black Lives Matter” sign in the yard. But Berkeley has the largest test score gap in the country.

Without investigating this further, it doesn’t seem strange to me that you would find this. Berkeley is likely and area with some of the most highly educated and well off white people. That you see larger gap in test scores seems predictable. Blacks don’t necessarily benefit from any of this.

I have been to San Fransisco, and the abject poverty on display there and the homelessness is rather appaling. As a Norwegian I frankly find the notion of California as being highly progressive as a joke. If San Fransisco is this heaven of progressives then why are there all these run down buildings? Why shitty streets. Why all the homeless people? Why the worst most rundown subway I have ever seen in my life?

Although I suppose one of the key problems is that Republican states ship their social problems to more progressive states, down to the buying bus ticks. This will naturally discourage doing too much for poor people, as it will simply invite the social problems of other states. It is one of the reasons why implementing more social democratic policies in the US is very hard. Various US states will basically race each other to the bottom.

Tracking Groups the Right Way

More seriously, if you don’t track any of the statistics, it’s gonna be easier to ignore the problem. i.e. I started this conversation by saying that the crime ratios in London and Toronto are just as bad as the US. But less people are aware of the issues there.

I think you have a bias in your thinking here. You assume the US method of tracking people by race is the most appropriate one. If all your data is race based, it will only encourage science attempting to find racial difference, further advancing the notion of race.

I think nationality is in many ways a far more sensible way of tracking people. It captures better how cultural background affects your outcomes. People from Laos and China while both Asian have profoundly different culture which create very different outcomes as immigrants. Simply lumping these groups together and call them Asian, because they look somewhat similar is quite wrong IMHO. It just encourages a racial view of society. Rather than understanding the profound influence of culture.

And, as I showed with PISA scores, immigrants do about the same in most of western Europe as they do in the US.

And as I pointed out, I view those tests scores next to worthless, because the kind of people admitted to different countries is radically different. E.g. Norway as I pointed out admit a lot of refugees and people from culturally incompatible countries. Canada e.g. in contrast admit far more high skilled immigrants from culturally compatible countries.

IQ scores across the middle east come out around 85–90 in Lynn’s surveys.

I don’t see them fret quite as often about this in the US, but that may be because the US has few muslim immigrants relative to other groups. Thus I am not sure to what degree they deem this as genetic.

If Europe is taking a lot of refugees from the middle east, the “race realist” theory is that the people will underperform. And that seems to be happening across Europe.

That is certainly a concern among a lot of European racists. But I am not sure whether the more scientific racists are onboard with this. E.g. whether they think this lower IQ score is transient and a result of cultural factors or whether they believe it will remain low over generations.

Out of Wedlock Children

Conservatives talk about this because there’s a huge marriage gap by race

I am well aware of this. That is why I brought it up. Conservatives like to blame this as the key reason why blacks are worse off. The claim is basically that blacks can blame themselves. There is this moral finger pointing about Blacks not getting into proper stable married relationships.

E.g. if you plotted Norwegians in on your graph we would be at the same level as Latinos. Out of wedlock births in Norway is at 56%. On Iceland it is at 70% so the same rate as for blacks. According to American conservatives Iceland should for this reason be a society in complete social decay. The complete opposite is the case.

Again we are dealing with a complex model, where people assume we have some sort of linear relationship based on a tiny set of factors.

Abortion Statistics

That doesn’t sound like poor access to abortion, to me.

I would have to moderate my stance a bit. My perspective was based on the many places in the US where abortion clinics are far away. Such as described in this article. However I assume that the population dense urban areas generally controlled by democrats offer descent access to abortion.

It seems like you’re operating off of some stereotypes about abortion in America, though, rather than looking at data.

No, I don’t think that is a fair assesment. Because I would never have assumed that you could judge my position by the statistics you provided. You have to relate it to how many people get abortion relative to how many need or want it.

E.g. Norway tops world wide statistics on one-night stands. We have far more random sexual encounters than what is common elsewhere. Yet teenage pregnancies are not off the charts. Quite the opposite. That is the outcome of better sex education and easier access to contraceptives.

Hence that Norway would have relatively few abortions is not all that strange. Even our Christian party pushes contraceptives as a way of reducing number of abortions. Contrast that with Christians in the US primarily pushing abstinence.

Again we are dealing with my primary cricism of you approach to data. You don’t look at the nuance and complex relationship. Instead you hunt for simple linear relationships between two observations. You judge access to abortion simply by the number of people who have abortion.

I would never judge an issue like this by merely looking at statistics. I would want to hear people stories and people assessment of the situation as well. That is valuable data, even if it is less crip.

News Bias

Maybe because of selection bias in how police murders are reported?

People are going to look at this in different ways. I don’t think there is a bias making it look as if blacks are treated worse than they actually are.

Tony Timpa died in a way very similar to George Floyd, but only Floyd made national news.

I think you are missing the point. Police brutality is a general problem in the US, also affecting white people. However it is a bigger problem relating to blacks.

The outrage over George Floyd as far as I am concerned is not over a single incident. It is about the top of the iceberg. It is just capturing on camera something that happens all too often. Outrage sparked because of one incident like this is not really about that single incident. It is about what that incidient represent. People that see that incident know that it is just one of countless others.

An outrage about Tony Timpa did not happen because he was not representative of something that happens often to white people. I would say however he is typical of something I see frequently in American law enforcement which is absolutely terrible handling of mentally ill people. US police officers as far as I understand little training on dealing with mental illness.

We have had a similar incident to George Floyd in Norway that happened to a black guy. But it didn’t cause a major outrage because police brutality isn’t a significant problem here. Yes it occurs, but it is rare enough that minorities don’t consider themselves stomped on the way they feel in the US.

Like, stories about school shootings at white schools always dominate the news, even though they’re incredibly rare.

I don’t think it is odd that children getting gunned down makes the news and a gang shootout doesn’t. People want to know that their children are safe at school. Naturally it is a concern. And the US has an exceptionally high occurrence of school shootings.

Are you forgetting Anders Breivik?

He was not a representative of Norwegian authorities. I was remarking on the brutality of American authorities against black teenagers and school children.

Long Term Effects

It sounds obvious that historical oppression wrecked families and traditions and left people at a long term disadvantage. It’s suspiciously similar to the cultural decay and family breakdown among Native Americans.

At least we are somewhat on the same page ;-)

I’m not sure if that’s the source of the test score gap and how.

Not sure why you would rule it out though. It seems premature to me to consider race until we have been able to account for historical effects.

The test scores are even worse in Africa, but there a million other variables.

Some of the test scores used by people like Lynn are highly dubious, using rather small sample sizes and picking unrepresentative individuals. Anyway even if scores are lower we have two important factors to consider:

  1. Poverty and malnutrition is more prevalent in Africa than in the US.
  2. IQ score is rising much faster in Africa than in other countries. Pointing to likelihood that whatever African potential score is, they are far away from having reached it.

As far as I know Africans immigrating to the US does no worse than African-Americans. Quite the contrary. Yes we don’t know to what degree self selection matters. However we don’t see a regression to mean as the race realists would predict for a sample of people from a population with lower natural IQ. Hence the race realist claim about Africans having lower IQ for genetic reasons doesn’t really hold up to scrutiny. Their predictions don’t add up.

What are the Real Causes?

In any case, it seems hard to understand, change, or fix that situation if you can’t identify what the causes are. The gaps could be caused by anything from pollution (i.e. lead poisoning) to poor prenatal health (nutrition and drug use) to culture (less time spent on homework, focus on education) to genetics to overt racism.

I don’t see why we need to have this worked out in detail. There is an obvious advantage in tackling all these issue even if we don’t know exactly how important each factor is.

We don’t need proof that racism knocks off IQ points, to begin tackling the problem.

We know that blacks growing up in better neighborhoods do better. We know blacks raised by white families do better. So obviously we have a lot of opportunity to boost score.

The paper by Drew Thomas which gave the outcome below, was the newest paper I could find studying black-white racial gap. There isn’t a lot of pick from.

Child and Parent | IQ Score
White-White | 101.8
Black-White | 98.5
Black-Black | 89.4

Maybe this study isn’t perfect. I am not sure if we will ever know. However all sorts of other studies show that children from poorer families raised in solid middle class families tend to get an IQ boost. Thus whatever the accurate end result is, we know we can do a lot better than we currently do.

School Funding

A lot of things in this world are counterintuitive. I was surprised to discover that school funding doesn’t boost test scores.

So we are back to what I criticized initially in this post. You look at too few variables and attempt to make linear models out of non-linear phenomenon. Let me take one example you mentioned:

Utah only spends $7,000 a year, for each student in their public schools.
New York state spends $22,000 for each of their students. You might expect that students in Utah do poorly and children in New York do well. You’d be wrong, though, test scores are a bit higher in Utah.

I happen to have lived in Utah for about 3 months with a family there. I have moron relative, and I have visited New York in multiple occassions. Comparing these two places along a single variable does not do justice to either place.

These are just profoundly different places. Utah is a deeply religious state, where people live in something almost looking like 1950s America on a postcard style of lives. The Church in Utah works almost the same as Scandinavian welfare states. There is a strong moral imperative to help each other out in Utah. Partly due to the culture, religion and homogeneity of the state.

And I am not talking about homogeneity in a racial sense, but in a religious and cultural sense. I am from Norway right, which frequently gets labeled as a super homogenous place by Americans (who have never been to Norway.) Yet Norway cannot remotely compare to the homogenity of Utah. It is like Stepford wives place.

Children in Utah have a very strong support at home with their parents and in their local community. If there is anything we know, it is that parental involvement is crucial to how kids turn out in school.

New York in contrast is the urban jungle, with lots of social conflicts, poverty etc. It is a place with far more problems. Catering to these problems are going to require a lot more resources.

If you where going to make your plot about school funding in Norway, you would likely find that higher funding cause schools to get worse. Or rather it would seem like that because schools with more problems get more funding. They get extra teachers for help out with language problems of immigrant children e.g. Extra resources for kids with mental disability etc.

You could have done the same kind of plot with police relative to crime and I am sure your conclusion would have been that more cops cause more crime. Rather than concluding that more crime causes more cops to be hired to handle the problem.

You also have to take into account costs. Utah is a fairly low cost state. New York is not. You don’t need to pay teachers as well in Utah to offer them the same living standard as in New York.

Thus if your analysis should make sense, at least you should study average teacher salary as well as living cost in said state. You would also have to look at challenges different schools face which may require extra funding.

I have a friend who worked in New York schools. They had to have elaborate security checks and metal detectors. Something tells me they are not spending money on that in Utah.

Gun Ownership

I was surprised to see that murder rates don’t correlate with gun ownership. Knowing these things is helpful — it could let you direct money or policies that help more.

Interestingly I also wrote about this issue about 2 years ago and drew the complete opposite conclusions from you.

You can read the article, but I can point out some of the key reasons I draw different conclusions.

  1. I don’t consider US states and their legislation on guns as having much if any value statistically speaking, because they don’t represent actual borders. Guns can freely flow from a state with lax rules to one with strict rules.
  2. The entities which needs to be compared has to be places where there are actual border checks. Any place with a harder border is an entity worth considering the gun laws within. Any two entities without actual hard borders, are useless, regardless of different regulations.
  3. Hence I look at gun violence between actual countries. There the studies are pretty clear. More guns and laxer regulation leads to more gun violence.

If you read my article you will see that I think far more about developing a model of how guns are used in crime. I don’t just start hacking at statistics trying to tease out some correlation.

How to Government Can Drive Social Change

Groups differences seem to be based on factors within the home, whether that’s parenting or genes, and we don’t have a great way to direct government to change either.

I don’t think this is as hard as you make it out to. Frankly I think it is relatively easy. This is primarily about raising the social economic status of parents. You can do that by raising their living standard and giving them better education as well as giving them better child services, shorter work days etc to spend more time with the children.

The non-parent environment matters as well. Simply changing zip code to a better area makes a huge difference.

Basically many of these things are accomplished with a social democratic system. With universal free public goods such as health care and education, you give poor people a better standard of living and better level of education.

Changing zip code is a bit more complex. Nordic social democracies also have some problems in this area. But one can see contours of a solution. More mixed housing allows more diversity in a neighborhood. You need low socios-economic groups of people to be in more proximity to higher socio-economic groups. That is because you need these groups to pick up the powerful cultural memes that leads to success and prosperity.

But let me be clear what I mean when I say “easy.” I only mean that the methods to employ are not rocket science. It does not necessarily make it easy to implement. Political opposition may be massive. This is also why I view racism as a major force blocking black progress in America today. Racism is an important factor, making American negative to Social-Democracy.

Too many racists hate a social welfare state because they view it as a way of handing out freebies to “lazy black people.”

But to be fair, American liberals are a major part of the problem as well. Liberalism does not believe in universal public goods. Liberalism tends to believe in means tested programs. It is the same in Norway. I used to be member of a liberal party, but has since concluded their analysis of society is fundamentally broken.

What America has thus ended up in liberal cities, is various welfare programs that kick in once everything is in the toilet. The very time it is far too late to change outcomes. It is like trying to safe somebody’s health after they got diabetes rather than getting them on a healthier diet while they they where starting to get obese.

Universal public goods kick in long before you have a serious problem. They help avoid that you slide down to the abyss in the first place.

This gets us back to square one, regarding problems in America: The two party system. It completely blocks any ideological alternative to liberalism or conservatism. An important force for social change all over Europe e.g. such as Social democrats and other socialist ideologies are entirely blocked out of American politics. And it seems both liberals and conservatives in America like to keep it that way.

Explaining Different Outcomes

I’m really not okay with the discourse that considers prejudice as the only explanation for group outcomes, without considering all the other important variables (like single mothers, lower test scores, higher crime rates).

Yes, but as you can see with our discussion, while I do consider single moms a factor. I only consider it a factor in context of a conservative society with minimal support of single moms.

And I do consider poor memes circulating in various populations a problem. But the origin of these memes are in my opinion originating from prejudice. It is not just about prejudice that exists today, but about prejudice that existed in the past. In fact I think the legacy of historical prejdice against African-Americans likely have a strong impact of African-Americans today that the prejudice that currently exists. But that is just my speculation.

I see a lot of articles that seek to shame people for even pointing out crime rate disparities, when we talk about police brutality.

Generally I don’t like shaming, but I am quite understanding of people who would shame somebody for pointing out crime rate disparities. It just comes across as a cheap way of belittling or trivializing a serious problem. It is also somewhat circular logic. One uses the prejudice and racism that causes a lot more blacks to be arrested and charged as an excuse for further abuse.

Yes I don’t deny that crime rates are higher among African-Americans. But that doesn’t mean that African-Americans cannot at the same time be arrested and charged at a higher rate than what is natural for their crime level.

What I mean is that studies show e.g. that in two population, one white and one black using drugs equally much, you will find far more arrests and convictions for blacks.

When searches for drugs and weapons are done. Usually there is a higher probability of finding drugs and guns on white people than on blacks, suggesting that blacks are targeted by police far more frequently than their crime level warrants. I have written various articles about this issue, but cannot recall the headlines ;-)

Picking the Right Policies

I think we need empathetic people to see problems in society, but I think we also need data driven people to work out which problems are largest, which are solvable, and how to do it.

Well I could point to numerous Social Democratic solutions as working well but I always get dismissed by conservative Americans. Show them that we have lower recidivism rate with our prison system and I get hit back with “your criminals are much nicer than ours.”

Point out that we have lower crime or social problems, and I get hit back with “you got less black people!”

In my experience it is pretty much hopeless to discuss solutions with Americans. They will dismiss pretty much any working solution from abroad, by claiming American exceptionalism. They will cling to every broken solution they already employ no matter how much it has failed.

They whine about how nobody knows a solution to their problems. Yet whenever you present a solution, they reject it stating that it would not work in America because either: America is too big, or America is too diverse or whatever.

Reality is that when American conservatives complain, nobody can figure out a solution, what they really mean is that nobody has come up with a solution that agrees with their conservative free wheeling capitalist ideology. Socialist or progressive inspired solutions e.g. are reject outright without any serious consideration.

My impression is that a lot of conservatives in America are not really looking for solutions. Rather they are looking for ways of excusing the status quo.

It is part of the reason I am so skeptical towards a lot of the race science. It doesn’t really seem geared towards finding solutions, and move society forward. It seems more like an effort to find excuses to maintain society in its current bad state at best. At its worst it is a way of giving a scientific excuse for eugenics or some kind of apartheid policy.

If we somehow managed to prove that e.g. African-Americans where intellectually inferior for genetic reasons, you really think anything good would have come out of that? I highly doubt it. It would become a rallying cry uniting all the racist and fascist forces in the West and create a strong drive for a white ethnostate or apartheid regime.

Written by

Geek dad, living in Oslo, Norway with passion for UX, Julia programming, science, teaching, reading and writing.

Get the Medium app

A button that says 'Download on the App Store', and if clicked it will lead you to the iOS App store
A button that says 'Get it on, Google Play', and if clicked it will lead you to the Google Play store