Genuinely wondering here: why even “adjust for…” if we’re talking about discreet events (crimes)?
We’re not trying to find out if Nigerians are inherently more criminal than Germans, we’re trying to find out which origin groups (if any) have higher criminality rates and why the German nation has to tolerate increased criminal presence for the greater good.
Why adjust for age? Why adjust for sex? If they are more young and more male they will commit more crime — does it therefore make sense for the German economy/society to have to absorb more criminals?
That's what they are trying to find out, whether 'foreigners' (by which they measure those without German passports) are more criminal and thus contribute to crime rates in areas, even if one accounts for their sex and age distribution and whatever effects of the areas themselves are.
From a political perspective, if you are interested in short term effects, then using raw data is of most interest. If you are interested in long-term effects, the age and sex adjusted values are of most interest.
Any serious person who has spent time analyzing behavioral data to understand a complex problem knows this: we can easily mislead ourselves with numbers.
Here's how we mislead ourselves:
1. There is never a single thread of analysis. There are always many. Each thread tells a different story. How do we pick the right thread?
2. Each approach to the model leads to fresh questions.
3. If you don't interrogate and clarify the problem you want to solve, you won't know the questions.
4. The mathematicians often confuse themselves with complexity.
5. If the analysis is not easy to grasp, the decision makers will not be able to agree on what it means. It must be powerfully simple
All of this assumes we do not have motivated reasoning. If we have motivated reasoning, we should assume the analysis is rubbish.
And none of this addresses the data collection and validation questions.
Does this put the whole "a homogenous society will have lower crime" argument to rest? I grew up in a high crime and highly homogenous society and hence never found the argument convincing.
As your data points out Japanese and likely other many East Asian migrants commit fewer crimes regardless of where they go. This can be either due to cultural or genetic reasons.
The same case can probably be proven for Arabs and Africans. They will be a high crime ethnicity regardless of where they go.
The problem with this study is, that it is so confiscated in its findings and no clear error is on display. For a lefty being “German” really means having a piece of paper with your name on it and therefore any crime statistics can be shifted at will.
The mainstream now has a study they can cite while whites have to endure statistically expected knife stabbings and other crimes because a 3x over-representation is somehow no over-representation.
These estimates are very sensitive to whether you look at whether a criminal is foreign born, the child of foreign-born mother, or simply a foreign citizen. This gives a lot of researcher degrees of freedom.
But the ecological fallacy is the main problem as you note. First-generation migration is highly location-specific, often down to the level of the street.
I really don't understand why you would need any controls other than age and sex? Seems like a rather clear attempt at manipulation by the study authors.
Normal sociological theories of crime consist of things like unemployment, which they ascribe to local factors, say whether some capitalist decided to put a factory there or not. So they think factors of neighborhoods cause crime, not people.
Never believe anything the current German government posts.
Where is Dracula when Europe needs him?
Had to chuckle hard at this :-D .
Genuinely wondering here: why even “adjust for…” if we’re talking about discreet events (crimes)?
We’re not trying to find out if Nigerians are inherently more criminal than Germans, we’re trying to find out which origin groups (if any) have higher criminality rates and why the German nation has to tolerate increased criminal presence for the greater good.
Why adjust for age? Why adjust for sex? If they are more young and more male they will commit more crime — does it therefore make sense for the German economy/society to have to absorb more criminals?
That's what they are trying to find out, whether 'foreigners' (by which they measure those without German passports) are more criminal and thus contribute to crime rates in areas, even if one accounts for their sex and age distribution and whatever effects of the areas themselves are.
From a political perspective, if you are interested in short term effects, then using raw data is of most interest. If you are interested in long-term effects, the age and sex adjusted values are of most interest.
Behavioral Rubbish.
Any serious person who has spent time analyzing behavioral data to understand a complex problem knows this: we can easily mislead ourselves with numbers.
Here's how we mislead ourselves:
1. There is never a single thread of analysis. There are always many. Each thread tells a different story. How do we pick the right thread?
2. Each approach to the model leads to fresh questions.
3. If you don't interrogate and clarify the problem you want to solve, you won't know the questions.
4. The mathematicians often confuse themselves with complexity.
5. If the analysis is not easy to grasp, the decision makers will not be able to agree on what it means. It must be powerfully simple
All of this assumes we do not have motivated reasoning. If we have motivated reasoning, we should assume the analysis is rubbish.
And none of this addresses the data collection and validation questions.
Questions we might ask:
1. Do we have patterns of criminality?
2. What variables are available to interrogate the patterns?
3. How reliable is the data?
4. Repeat
Does this put the whole "a homogenous society will have lower crime" argument to rest? I grew up in a high crime and highly homogenous society and hence never found the argument convincing.
As your data points out Japanese and likely other many East Asian migrants commit fewer crimes regardless of where they go. This can be either due to cultural or genetic reasons.
The same case can probably be proven for Arabs and Africans. They will be a high crime ethnicity regardless of where they go.
Controlling for shoe size, men are not taller than women.
Controlling for size, pitbulls are not more likely to kill people than dachhunds.
Thanks for looking into this.
The problem with this study is, that it is so confiscated in its findings and no clear error is on display. For a lefty being “German” really means having a piece of paper with your name on it and therefore any crime statistics can be shifted at will.
The mainstream now has a study they can cite while whites have to endure statistically expected knife stabbings and other crimes because a 3x over-representation is somehow no over-representation.
As per Edward Bernays statistics are a propaganda tool.
These estimates are very sensitive to whether you look at whether a criminal is foreign born, the child of foreign-born mother, or simply a foreign citizen. This gives a lot of researcher degrees of freedom.
But the ecological fallacy is the main problem as you note. First-generation migration is highly location-specific, often down to the level of the street.
I really don't understand why you would need any controls other than age and sex? Seems like a rather clear attempt at manipulation by the study authors.
Normal sociological theories of crime consist of things like unemployment, which they ascribe to local factors, say whether some capitalist decided to put a factory there or not. So they think factors of neighborhoods cause crime, not people.