35

I saw on Reddit that a site called Sputnik News reported that one-fourth of people named Ali in Sweden have a criminal record. I researched a little bit but couldn't find any other source validating this claim.

Is this true?

Source: https://sputniknews.com/society/201911111077277162-one-fourth-of-people-named-ali-in-sweden-have-a-criminal-record--report/

Samhällsnytt's search found that nearly a quarter of people named Ali residing in Sweden have a criminal record.

A search in Statistics Sweden revealed that 40,790 people have Ali as either first name or surname. A search in Lexbase resulted in 9,742 hits, meaning that roughly 24 percent of all Alis in Sweden have a conviction.

Reddit post: https://www.reddit.com/r/tucker_carlson/comments/dzup5e/imagine_my_shock/

Oddthinking
  • 140,378
  • 46
  • 548
  • 638
OddDev
  • 2,139
  • 5
  • 14
  • 14
  • Comments are not for extended discussion; this conversation has been [moved to chat](https://chat.stackexchange.com/rooms/101363/discussion-on-question-by-odddev-is-it-true-that-one-fourth-of-people-named-al). If it isn't about improving the question, go there. – Oddthinking Nov 22 '19 at 13:47
  • 34
    Given that this question is related to blatant propaganda, propaganda that relies on its headline message being spread far and wide, propaganda that doesn't hold up the second you scratch the surface; is it really a good idea that this question is in HNQ? Putting it in HNQ seems to be an act that works _for_ the propaganda rather than against it (because more people will see the headline in passing and not read further, leaving them with the message implanted in their minds). Or, rather, does being in HNQ actually encourage deeper inspection? (as it has for me) Or is this a question for Meta? – Aaron F Nov 23 '19 at 13:20
  • 5
    @AaronF That sounds like an excellent question for Meta. – Llewellyn Nov 23 '19 at 18:41
  • 3
    @Llewellyn OK, [I've asked it](https://skeptics.meta.stackexchange.com/questions/4470/does-featuring-a-question-regarding-propaganda-on-hnq-help-do-its-work) – Aaron F Nov 23 '19 at 21:41
  • In the name of improving the question, I note that "residents of Sweden" (the subject of the claim) isn't the same as "Swedes" (the subject of this question). In particular, I suppose that a number of residents of Sweden who bear the name Ali are not Swedish. – phoog Nov 25 '19 at 06:23

4 Answers4

131

The short answer: the post uses invalid figures to support their claim "roughly 24 percent of all Alis in Sweden have a conviction". Furthermore, they are misrepresenting what their figures actually mean.

The first figure is the number of people residing in Sweden who are named "Ali". The source for this figure is the Swedish governmental website Statistics Sweden. They explain what their figures represent:

Name statistics are based on information about people registered in the Population Register submitted by the Swedish Tax Agency to Statistics Sweden.

The way the population register works in Sweden, anyone who wants to open a bank account, work in a regular job, or get medical treatment will need to register. It is, however, still possible to live in Sweden without being registered.

If you enter "Ali" in the Statistics Sweden web-site, you get (as of today) the following results:

14 914 persons have Ali as a main first name

First name:

14 896 men have Ali as a main first name

A total of 26 115 men have Ali as a first name

18 women have Ali as a main first name

A total of 6 474 women have Ali as a first name

Last name:

14 675 persons have Ali as last name

In order to come up with the figure of people named Ali from the original post (40 790), you have to add the total for men with Ali as any first name (26 115) and the total for people of any gender with Ali as a last name (14 675).

Note how the figure quoted in the original post conflates male and female name-bearers. Women who have "Ali" as their second name are excluded, but women with "Ali" as their last name are included. Using this figure to represent "people have Ali as either first name or surname" is therefore incorrect. According to Statistics Sweden, the correct number of that is 47 282 people.

The second figure comes from Lexbase, a highly controversial private service that makes use of (or exploits, depending on your personal view) the fact that most official documents need to be available to the public in Sweden. This includes court documents. There are several criticisms of the site, but the one that is perhaps most relevant is that the database contains data of anyone who has been under legal investigation in Sweden, but the outcome of a trial is only visible after users pay a registration fee to Lexbase. The Swedish Wikipedia article explains:

Även personer som friats eller vars dom inte vunnit laga kraft ger sökträffar och markeras på kartan. Typ av rättegång framgår inte av kartan, utan allt från trafikförseelser och vårdnadstvister till grov brottslighet indikeras på samma sätt. I resultatlistan från sökningar anges numera måltyp (tvistemål eller brottmål), namn, ålder och nuvarande folkbokföringsadress. Den som kostnadsfritt prenumererar på e-postuppgifter om personer i sitt närområde som förekommer i databasen får dessutom veta deras medborgarskap. Detaljer om rättegången, inklusive rättegångsutfall, framgår av de domstolshandlingar som Lexbase tillhandahåller mot en avgift.

Google translation:

Even people who have been freed or whose judgments have not gained legal force give search results and are marked on the map. The type of trial does not appear on the map, but everything from traffic offenses and custody disputes to serious crime is indicated in the same way. The result list from searches now specifies the target type (dispute or criminal case), name, age and current census address. Anyone who subscribes to e-mail information about people in their immediate area that is included in the database will also know their citizenship. Details of the trial, including trial results, are set forth in the court documents provided by Lexbase for a fee.

This means that being listed in Lexbase is not the same as having a criminal record, at least as far as the term is generally understood. People listed in Lexbase have been in contact with the law for whatever reason, but the charge may have been a trivial neighborhood quarrel just as well as a violent crime, and they may have been acquitted of their charges, or they may have been sentenced – they all will be listed as entries in Lexbase.

This means that unless a fee has been paid to Lexbase so that each entry could carefully be checked, the number of people listed in Lexbase does not correspond to the number of convictions. The original post misuses the figure that they claim to have obtained from Lexbase.

Note also that the Lexbase search doesn't distinguish between male and female entries – the reported number of entries will be (unlike the Statistics Sweden figure reported in the post) both men and women who have Ali anywhere in their name.

Regardless of all this, I couldn't even reproduce the number of entries that the original post reports (9 742 entries in Lexbase). As of today (November 22, 2019), if you enter "ali" as a name in their person search, the database reports 4 961 entries, as the screenshot shows:

Screenshot of a Lexbase search for the name "ali", showing 4 961 entries as of November 22, 2019

This is clearly much smaller than the figure reported in the original post. It may be possible that the number retrieved from Lexbase was larger at the time the original author made the post, but given that old entries are not purged from Lexbase, this isn't very probable.

So, to wrap this mess up: The conclusion that "roughly 24 percent of all Alis in Sweden have a conviction" is wrong:

  • The number of people called Ali is, according to Statistics Sweden, 47 282 and not 40 790, as the post claims. Their figure approximates the number of men who are named Ali, but it also includes women with Ali in their last name.

  • Entries in Lexbase do not represent people with criminal records, but people who have been involved in legal proceedings. The entries include both men and women.

  • The number of entries in Lexbase for the name Ali is currently 4 961, not 9 742 as the post claims.

If you still want to relate the number of entries for the name in Lexbase to the number of people with the name Ali in Sweden, this will give you a value of 10.5%, which is way off the "nearly a quarter" that the post claims.

Schmuddi
  • 9,539
  • 5
  • 44
  • 46
  • Comments are not for extended discussion; this conversation has been [moved to chat](https://chat.stackexchange.com/rooms/101503/discussion-on-answer-by-schmuddi-do-one-quarter-of-swedes-named-ali-have-a-cri). – Jamiec Nov 26 '19 at 09:02
33

Schmuddi's answer notes that the Lexbase database only contains 4961 persons with Ali as one of their names. But if we read the Samhällsnytt article, we find that the number 9742 comes from the Verifiera database.

The Verifiera database contains cases not only from criminal and civil courts, but also from a number of administrative courts and similar bodies. This includes cases regarding:

  • Disputes between landlords and tenants
  • Decisions by the Swedish Tax Agency
  • Decisions by municipal social services
  • Decisions by the Swedish Social Insurance Office
  • Involuntary care for addiction, psychiatric illness and minors
  • Decisions by municipalities and county councils
  • Decisions by The Swedish Migration Agency
  • Various other decisions by government bodies

I would assume that most of these cases were brought to the courts by a private person to challenge a previous decision made by a government body.

I think the Migration Court cases are particularly noteworthy in this context. Anyone whose application for asylum or residency was declined can appeal their case to a Migration Court. The Migration Courts saw a big surge in cases during and after the European migrant crisis. According to this article 39 929 cases were brought to the Migration Courts in 2016 alone. The largest group of refugees arriving during this period was Syrian and Ali is an extremely common name in Syria. Since people appealing an asylum decision have already been declined once, it's likely that many of the people that these cases concern are no longer living in Sweden and would not be included in the quoted name statistics.

jkej
  • 2,048
  • 1
  • 8
  • 11
17

There are so, so many statistical issues with this (as is expected with a release from a Russian propoganda outlet), but others have covered most of the most important ones....except one. The quote is:

Samhällsnytt's search found that nearly a quarter of people named Ali residing in Sweden have a criminal record.

We break this down to:

  1. Almost 25%
  2. Of those people who are named Ali
  3. Of those people who live in Sweden
  4. Have a criminal record.

That means that if we had 1000 people living in Sweden named Ali, 250 of them would have a criminal record. Schmuddi and Dave have shown that the numbers themselves are wrong and that they're stealing from a list of indicted people, not convicted ones, but I'd like to point out that the wording doesn't match with their methodology, either.

What they're doing is to take a census data point (number of Ali's in Sweden) and take a crime data point (number of people indicted who are named Ali) and compare them. These cannot be compared without additional data! What if one of those Ali's in my example was indicted 250 times? That would suddenly mean that .1% of Ali's have (a criminal record/indictment/whatever). In fact, if all the other steps were 100% correct, just one Ali having two entries in that list messes up their numbers significantly.

Bottom line, this isn't science, it's propaganda.

Carduus
  • 2,096
  • 10
  • 15
  • 11
    You seem to be assuming that they counted indictments in Lexbase, not unique people. – Barmar Nov 22 '19 at 20:22
  • 3
    @Barmar well, if they didn't count indictments, their numbers are even more way off than they are now, so it would make their conclusion even more wrong – eis Nov 23 '19 at 16:18
  • 1
    Can names of non-Swedish-resident people appear in Lexbase? Extra non-Swedish Alis would also skew things. – Peter Cordes Nov 24 '19 at 02:42
  • Why did you mention somethig you called "a Russian propoganda outlet"? – exebook Nov 24 '19 at 02:52
  • 4
    @exebook Because that's what https://en.wikipedia.org/wiki/Sputnik_(news_agency) obviously is. – user3445853 Nov 24 '19 at 11:46
  • @exebook RT and Sputnik are the two main Russian government-run news agencies. Everything they print is either censored-by or sourced-from the Russian government. – Carduus Nov 25 '19 at 13:44
8

I'm not sure why you asked the question here, because the claim contains everything you need to verify it for yourself.

At the time of this posting, searching for the name "Ali" on Statistics Sweden (https://www.scb.se/en/finding-statistics/sverige-i-siffror/namesearch/Search/?nameSearchInput=ali) produces the result

14 914 persons have Ali as a main first name

First name:

14 896 men have Ali as a main first name

A total of 26 115 men have Ali as a first name

18 women have Ali as a main first name

A total of 6 474 women have Ali as a first name Last name:

14 675 persons have Ali as last name

Searching for the name on Lexbase (https://www.lexbase.se/personsok?Search[fullname]=ali) gives

'ali' gav 4961 träffar i databasen

followed by a list of those 5000 Alis.

Note that, in addition to the Lexbase results being only about half the number given in the claim, Lexbase includes all court cases. These 5000 Alis are not all convicted criminals. Some of them were found innocent. Some of them were brought to court for minor violations, such as speeding tickets. Some of them got divorced and sought a ruling on the division of marital property or custody of their children. And so on. Lexbase is a database of court cases, not a database of criminals.

Dave Sherohman
  • 852
  • 1
  • 9
  • 11
  • 1
    +1 for reproducing results, that's a good first step. Now, this raises a couple of questions: does Lexbase search by "main first name"? or by full first, last and middle names? or by the substring "ali" anywhere in the name? does it filter results by gender? – default locale Nov 22 '19 at 08:58
  • 8
    This is where we hit original research problems. Why Ali? Why not Mohammed, or Ahmed, or other common Muslim names? Maybe they don't provide a nice outlier, so someone kept trying names until they found a 'good' one? Pick 10 common Swedish names - what is the range of % of them in the Lexbase records? Can any attempt be made to account for some ethnicities being more or less like to be charged versus cautioned, or convicted versus acquitted on the same evidence? Are there economic factors involved (poor people more likely to commit common crimes, immigrants might have lower average incomes). – PhillS Nov 22 '19 at 09:07
  • 1
    I've experimented a bit. It appears that Lexbase tries to search for a full name (first, last or whatever, see results for [Abdal](https://www.lexbase.se/personsok?Search%5Bfullname%5D=abdal) and [Abdalaziz](https://www.lexbase.se/personsok?Search[fullname]=abdalaziz)) and also includes hyphenated names (Abdal-Amir counts as both Abdal and Amir). So, the better estimate would be 4961/(26115+14675) ~ 12% – default locale Nov 22 '19 at 09:17
  • 3
    @defaultlocale - If you're going to do that calculation, it's actually 4961/(26115+6474+14675) = 10.5%. (You left out the 6474 women with Ali as a first name.) However, this ignores the possibility of some people having Ali as both a first and a last name; the actual total number of people with the name Ali could be anywhere in the range from 32589 to 47264, which is why I didn't include this calculation in my answer. – Dave Sherohman Nov 22 '19 at 09:26
  • 2
    We had a [similar question](https://skeptics.stackexchange.com/questions/43978/) where another approach taken by answers was to point out the *relative* distribution of first names in populations with a strong islamic influence -- i.e., that "Mohammed" (or, in this case, "Ali"?) is *much more common* among immigrants than e.g. Sven or Michael is among Europeans. – DevSolar Nov 22 '19 at 11:49
  • @DaveSherohman Also, would a case where *both* parties had 'Ali' in their name count once or twice? Probably a statistical outlier, but... – Shadur Nov 22 '19 at 15:02
  • @PhillS That would be an interesting exercise.... what is the most common Swedish (male) name... wonder what the "percentage of criminals" of that is. (understanding that Lexbase doesn't give 'criminals' but anyone with any form of legal proceeding) – CGCampbell Nov 22 '19 at 18:20
  • 2
    @CGCampbell: I'm pretty sure that the percentage will be lower than Ali. Chances are high that the most frequent name in Sweden will be fairly evenly distributed across Swedish society. Thus, involvement in legal proceedings will be average for them. In contrast, Ali is strongly associated with socially disadvantaged groups. It's well-researched that members from these groups have an above average chance to be accused of illegal activities. The interesting exercise would be to find a native Swedish name with the same frequency and which is associated with similar socioeconomic groups as Ali. – Schmuddi Nov 22 '19 at 20:38
  • @Schmuddi instead of saying you're 'pretty sure' why don't you actually test it instead of spreading more hearsay? Isn't that the whole point of this stackexchange? – par Nov 25 '19 at 00:46
  • 1
    @par: And why don't you tell the same for *any* of the other comments that poses a question or hypothesis, i.e. all of them? – DevSolar Nov 25 '19 at 12:25
  • @DevSolar stating you're 'pretty sure' about something isn't posing a question or hypothesis. Also, looking at the comments above I can see that many of them look at data, to back up their claims for example the comment by default locale mae on November 22nd. So your vauge implication that 'all of' the other comments are of the nature of that posted by Schmuddi is wrong - in other words an exaggeration you made for no purpose other than to prove a point. An example of posing a hypothesis is the first comment (default locale Nov 22) - notice the difference in tone to that posted by Schmuddi – par Nov 27 '19 at 15:34
  • @DevSolar and while we're on that point, questions have been raised by others about how the 'ali' search string is being used in LexBase. Many names contain 'ali', for example 'alison', 'alice', and so on. So perhaps the original commenter can be less than 'pretty sure' that the 'ali' which is being serached in LexBase truly represents the subset of the population which they think. – par Nov 27 '19 at 15:35
  • @DevSolar (continued from previous comment) Schmuddi's doesn't provide evidence for their second assertation that sweedish minorities are more likely to be accused of illegal activities, having checked a few sources myself, he is correct on that point. My issue isn't with Schmuddi being right or wrong, it was with the tone of their assertation. One simply cannot be 'pretty sure' of anything without evidence or good reasoning. Although this is a relatively minor point in context of their entire comment. Hope this answers your question. – par Nov 27 '19 at 15:44
  • @par: As you have found out yourself, SE comments are limited in length. Demanding full sources for something I consider a pretty uncontentious ("Ali" being associated with socially disadvantaged groups, socially disadvantaged being overrepresented in crime statistic) in reply to a different user, in the tone *you* used, strikes me as being rather belligerent. – DevSolar Nov 27 '19 at 16:26
  • @DevSolar I haven't used any belligerent language and if I have it isn't intentional, nor would I demand anything on a volunteer based knowledge exchange. Apologies if I have done so. It is possible to include sources by way of hyperlink, you can see examples of this in a comment above. This would be a space efficient way of including sources. Also as we discussed lexbase isn't a crime statistic (read the answer above for the reasoning behind this) and furthermore the search term 'ali' may not refer purely to those with the name ali, but any name which contains the substring 'ali'. – par Nov 28 '19 at 19:29
  • @DevSolar I appreciate you are more experienced with the topic and the data than I am. I hope you won't interpret my questioning as being belligerent. Im not intending to be rude or offensive. You mentioned: ""Ali" being associated with socially disadvantaged groups, socially disadvantaged being overrepresented in crime statistic". Probably true. Based on your knowledge can you give a sense of how much over-represented the name Ali is 'crime statistics', would it be 2 times more, 5 times, 10 times? Your insight on this would be really interesting and relevant to the OPs question – par Nov 28 '19 at 19:36