Wild Zontargs wrote: ↑2017-12-05 01:00pm
K. A. Pital wrote: ↑2017-12-05 12:31pm
Sorry if I missed anything, did the Nazis leak the police reports or database itself or did they
only provide their own "conclusions"?
If the former (report was leaked and was/is still accessible to others as well), then one needs to challenge the data itself. But if the latter, one has to see the data first.
A translation of the page with the data is available here.
Translation of the information page:
Information on comparisons between Swedes and non-Swedes sentenced to imprisonment between 2004 and 2014
Published November 27, 2017
Content: 168,000 sentencing decisions from the prosecution service between 2004-05-10 and 2015-01-08, of which approximately 141,000 are for a unique ruling. The filter requires that the social security number, court and destination number are uniquely unique in the data covered by the study. A few sentences of imprisonment from foreign courts where the penalty period is over 14 years has been removed manually when a separate Swedish decision is available. The information has been developed in cooperation with Nordfront, which has donated the source data.
Each decision document contains the decision date, first name and surname separately, social security number, judicial authority who assessed the penalty, the number of the case, the date of judgment, the period of imprisonment, the region and the place of crime for the prosecution and verification email and more.
comments:
A very small number of the estimated penalties may contain a small number of penalties that come from forfeited conditional release for other offenses.
Around 26,700 punishment decisions of the 168,000 are for existing decisions. These 26,700 decisions have therefore been removed from the analysis and only the last decision is included. This means that the prosecution estimates about 16% of their penalties due to errors, reassessments or the like, and it is therefore most reasonable to rely on the latest.
Some rulings concern life imprisonment. In the analysis, the life sentence was converted to 24 years in prison.
Because nationality or ethnicity is lacking in the decisions, an assessment has been made of the convicted names.
Full references are available for each penalties decision included in the study.
Calculation and Analysis Modules:
Search for crimes by date of birth or coordination number
Most common and surnames of Swedes and non-Swedes
Name variation for Swedes and non-Swedes
Violence for Swedes and non-Swedes by birth year
Comparisons of all crime and coarse (more than 4, 10 and 20 years)
Comparisons between Swedes and non-Swedes after punishment period
Comparisons by age of the convicted person
Occupational criminals
Violence in Sweden after the prosecution area
Distribution of Swedish and non-Swedish wrestling by unit of investment
Analysis of fake birthdays for non-Swedes
Emphasis mine. Now, there's where we can debate the methodology used. They're going by Swedish names and non-Swedish names. Non-Swedish names are over-represented compared to demographics.
1. The only way to do that search is going to be to use machine learning techniques, and it is very very easy to build a racist classifier by accident, let alone for Nazis to do it.
2. You're damn right I'm going to question the methodology of using people's names as an ethnic classifier. The reason I object is should be obvious to everyone with a fucking brain, but because I'm pretty sure yours is defective here we go: People who are from Sweden might name their child using a foreign name, while immigrants might name their child using a swedish name. Then there are the mixed ethnicity families. And the fact that Sweden is in the EU and Schengen Area, which means that there's everyone from brits to russians hanging around in these data as well.
3. If they have these data, why the hell don't they break it down by crime? Your original link seems to suggest they do, what with the specificity of gangrape as a crime. Yet there we are, with these links not seeming to give that capacity. That seems suspicious to me (see below). The same goes for year of crime.
4. Just because these people have repeatedly violated swedish privacy laws and may have obtained some data does not necessarily imply that they haven't cooked their books. They don't seem to be giving anyone direct access to their metadata or have any documentation whatsoever, so there is no way to authenticate the data they do have available. No way to replicate their techniques in searching and analyzing it. They're Nazis. Lying is something they do and so they ought not get the benefit of the doubt when it comes to the veracity of their claims or the integrity of their data.
5. All of the above notwithstanding, even if we assume that every person listed as foreign born is brown, and that they are being completely honest... a simple numerical comparison is shitty shitty statistics. Why? Because crime is influenced by factors other than ethnicity. For instance, the first thing you need to do is statistically control for socio-economic status and break crime down by category (Drug-related, property crime, subtypes of fraud, violent crime, sex crime), otherwise all you are getting is noise. You're not actually getting real relationships out of it, and it's basically useless. For instance, it may well be the case that immigrants and native born swedes are--when other factors are controlled--identically likely to commit various crimes, but because the immigrants are more likely to be poor, and poor people are more likely to commit crimes, immigrants are more likely to commit crimes.
6. THEN you have account for factors related to criminal investigation itself. In the US, and I have zero reason to think Sweden is any different, brown people are more likely to be stopped by police and investigated in the first place, independent of the actual frequency at which crimes are committed.
Long story short, your data is untrustworthy, the methods are crap, and it is thus useless.
