Skip to main content

LUP Student Papers

LUND UNIVERSITY LIBRARIES

Sanningssökare - En bayesiansk bevisvärderingsanalys av ej styrkta rekvisit i friande mål om människohandel

Mårtensson, Alexander LU (2025) JURM02 20251
Department of Law
Faculty of Law
Abstract (Swedish)
Syftet med arbetet har varit att, med hjälp av bayesiansk sannolikhetsteori som bevisvärderingsmetod, kritiskt analysera svenska människohandelsmål. Detta har gjorts för att utröna om det förekommer tecken på bristande bevisvärdering från domstolarna som bidrar till att rekvisit ej anses styrkta i dessa mål där någon blir friad, och i utsträckning bidragit till bevisproblematik kring brottet. Syftet uppnås genom en kritisk, kvantitativ rättsfallsanalys för att identifiera potentiella tankefel i samband med rättens bevisvärdering kring ej styrkta rekvisit i mål där en åtalad människohandelsgärning friats. Termen tankefel används i vid bemärkelse i arbetet och omfattar typiska fall av kognitiv bias såväl som bristande... (More)
Syftet med arbetet har varit att, med hjälp av bayesiansk sannolikhetsteori som bevisvärderingsmetod, kritiskt analysera svenska människohandelsmål. Detta har gjorts för att utröna om det förekommer tecken på bristande bevisvärdering från domstolarna som bidrar till att rekvisit ej anses styrkta i dessa mål där någon blir friad, och i utsträckning bidragit till bevisproblematik kring brottet. Syftet uppnås genom en kritisk, kvantitativ rättsfallsanalys för att identifiera potentiella tankefel i samband med rättens bevisvärdering kring ej styrkta rekvisit i mål där en åtalad människohandelsgärning friats. Termen tankefel används i vid bemärkelse i arbetet och omfattar typiska fall av kognitiv bias såväl som bristande trovärdighetsbedömningar utifrån uppsatsens bevisvärderingsmetod. Frågeställningarna kopplat till syftet har handlat om hur många rättsfall med tankefel som förekommer, vilka olika typer av tankefel som förekommer samt vilken betydelse eventuella tankefel har haft. Betydelsen har avgjorts utifrån om tankefelen har haft ett kausalt samband med att rekvisitet de hänfördes till inte ansågs bevisat.

Det föreligger omfattande problematik kring att utreda och lagföra misstänkta gärningspersoner för människohandel. En del av problematiken består i att faktiskt lyckas bevisa gärningen. En egen statistisk undersökning gjordes på samtliga 45 rättsfall mellan år 2018 och 2025 där människohandel ingått som åtalad gärning, för att utröna hur många åtalade gärningar av detta slag som blev friade eller dömda. 58,5 % av samtliga åtalade gärningar i tingsrätten, och 28,7 % i hovrätten, ledde till friande. Den genomsnittliga andelen friande gärningar över instanserna låg på 45,7 %. Jämförelsevis blev cirka 32 % av de åtalade våldtäktsbrotten – ett notoriskt svårbevisat brott - under 2017 friade.

För att läsaren ska förstå terminologin som används under analysen och be-tydelsen det har, presenteras först de teoretiska utgångspunkterna. Arbetet har redogjort för de bevisrättsliga och straffrättsliga utgångspunkterna, samt grundläggande information om bayesiansk sannolikhetsteori som bevisvärderingsmetod och vittnespsykologi. Den kritiska, kvantitativa rättsfallsanalysen har omfattat samtliga fall av åtalad människohandel mellan år 2018 och 2025, där någon blivit friad för gärningen.

I analysen har identifierade tankefel gradindelats enligt en skala på låg, medel och hög säkerhet utifrån vilken säkerhet tankefel har kunnat bekräftas med. Resonemang har förts utifrån om tankefelen i ett rättsfall haft ett kausalt samband med rekvisiten som ej bevisats. I det fall ett kausalt samband har bedömts föreligga, har detta även gradindelats utifrån om det kausala sambandet konstaterats med låg, medel eller hög säkerhet.

Totalt har 30 rättsfall där den tilltalade frias för en människohandelsgärning analyserats. Cirka 79 % av tingsrättsfallen, och cirka 82 % av hovrättsfallen innehöll minst ett tankefel. I cirka 26 % av tingsrättsfallen, och cirka 18 % av hovrättsfallen, förekom tankefel av sådan betydelse att de bedömdes ha ett kausalt samband med att rekvisit i dessa mål inte ansågs styrkta av dom-stolarna.

Den slutliga bedömningen är att det finns starka tecken som tyder på att det hos domstolarna förekommer bristande bevisvärdering som bidrar till att rekvisit ej anses styrkta. Bristerna består främst av bristande trovärdighetsbedömningar - särskilt hos tingsrätterna - vilket spelar stor roll i mål om människohandel där de tilltalades och målsägandenas muntliga uppgifter ofta är kärnbevisningen. (Less)
Abstract
The purpose of this paper has been to critically analyze Swedish court cases of human trafficking where the accused has been acquitted of such a criminal act. The critical analysis has been executed with the aid of Bayesian probability theory as an evidence evaluation method. This has been done to determine whether there are signs of flawed evidence evaluation from the court itself which has contributed to necessary conditions not being fulfilled for the crimes in these cases, and in extension contributed to the problems regarding proving this type of crime. This purpose is achieved through a critical, quantitative court case analysis to identify potential signs of cognitive bias regarding the necessary conditions for the crime in these... (More)
The purpose of this paper has been to critically analyze Swedish court cases of human trafficking where the accused has been acquitted of such a criminal act. The critical analysis has been executed with the aid of Bayesian probability theory as an evidence evaluation method. This has been done to determine whether there are signs of flawed evidence evaluation from the court itself which has contributed to necessary conditions not being fulfilled for the crimes in these cases, and in extension contributed to the problems regarding proving this type of crime. This purpose is achieved through a critical, quantitative court case analysis to identify potential signs of cognitive bias regarding the necessary conditions for the crime in these cases that are not considered proven by the court. The term cognitive bias is in this paper used to encompass not only typical cases of cognitive bias, but also by the court lackluster credibility assessments regarding oral statements and the person giving them. The research questions for the purpose of the paper have been regarding how in many cases the court’s evaluation of the evidence appears to be affected by cognitive biases. It also includes which types of different cognitive bias that are present as well as the significance of these potential cases of cognitive bias. This significance has been determined from whether a cognitive bias has had a causal effect on a prerequisite of the crime not being considered proven by the court.

There is a substantial difficulty regarding investigating and prosecuting suspects of human trafficking. Some of this difficulty stems from proving the criminal act before a court. I did a lesser statistical analysis based on all the 45 cases of human trafficking in the country from between 2018 and 2025 to determine how many charged acts of human trafficking were acquitted or convicted. 58,5% of all charged acts in the district courts, and 28,7% in the Courts of Appeal, led to an acquittance. The average percentage of acquitted acts across all courts of instance were 45,7%. In comparison a third of all judgements regarding cases of rape, a notoriously hard crime to prove, in 2017 led to acquittance.

The theoretical starting points are first presented, in order for the reader to comprehend the terminology used in the analysis and the significance of it. This entails an account of the legal framework for evidence evaluation by the court and the legal framework of the crime in question. It also includes an account of the basic knowledge of Bayesian probability theory as a method of evaluating evidence as well as fundamental knowledge if witness psychology. The critical, quantitative court case analysis has encompassed all cases between 2018 and 2025 where the defendant has been acquitted of the act.

The analysis has identified cases of cognitive bias that has been graded according to a scale of low, medium and high certainty depending on with which certainty the cognitive bias has been confirmed. Reasoning has been carried out regarding whether the cognitive bias in a case has had a causal effect on a prerequisite of the crime not being considered proven by the court. In the case a causal effect has been considered to exist, it has also been graded with a low, medium or high certainty.

30 cases in total where the defendant has been acquitted of the charge has been analyzed. In about 79% of the district court cases, and about 82% of the Court of Appeal cases, a cognitive bias was confirmed with varying degrees of certainty. In about 26% of the district court cases, and about 18% of the Court of Appeal cases, a causal effect was identified between a cognitive bias and the related prerequisite by the court not being considered proven.

The final assessment is that there are strong signs indicating an occurrence of flawed evidence evaluation within the courts which contributes to prerequisites of a crime not being considered proven. These flaws consists mainly of lackluster credibility assessments regarding oral statements and the person giving them, which carries a significant risk since this type of evidence is more often than not the core evidence of the case. (Less)
Please use this url to cite or link to this publication:
author
Mårtensson, Alexander LU
supervisor
organization
alternative title
Truthseeker - A bayesian evidence evaluation analysis of unproven prerequisites in court cases of human trafficking where the accused has been aquitted
course
JURM02 20251
year
type
H3 - Professional qualifications (4 Years - )
subject
keywords
allmän rättslära, straffrätt, bevisvärdering, bayesiansk sannolikhetsteori, bayesiansk bevisvärderingsmetod, människohandel, domstolens bevisvärdering, bevisproblematik
language
Swedish
id
9189029
date added to LUP
2025-06-03 11:17:10
date last changed
2025-06-03 11:17:10
@misc{9189029,
  abstract     = {{The purpose of this paper has been to critically analyze Swedish court cases of human trafficking where the accused has been acquitted of such a criminal act. The critical analysis has been executed with the aid of Bayesian probability theory as an evidence evaluation method. This has been done to determine whether there are signs of flawed evidence evaluation from the court itself which has contributed to necessary conditions not being fulfilled for the crimes in these cases, and in extension contributed to the problems regarding proving this type of crime. This purpose is achieved through a critical, quantitative court case analysis to identify potential signs of cognitive bias regarding the necessary conditions for the crime in these cases that are not considered proven by the court. The term cognitive bias is in this paper used to encompass not only typical cases of cognitive bias, but also by the court lackluster credibility assessments regarding oral statements and the person giving them. The research questions for the purpose of the paper have been regarding how in many cases the court’s evaluation of the evidence appears to be affected by cognitive biases. It also includes which types of different cognitive bias that are present as well as the significance of these potential cases of cognitive bias. This significance has been determined from whether a cognitive bias has had a causal effect on a prerequisite of the crime not being considered proven by the court.
 
There is a substantial difficulty regarding investigating and prosecuting suspects of human trafficking. Some of this difficulty stems from proving the criminal act before a court. I did a lesser statistical analysis based on all the 45 cases of human trafficking in the country from between 2018 and 2025 to determine how many charged acts of human trafficking were acquitted or convicted. 58,5% of all charged acts in the district courts, and 28,7% in the Courts of Appeal, led to an acquittance. The average percentage of acquitted acts across all courts of instance were 45,7%. In comparison a third of all judgements regarding cases of rape, a notoriously hard crime to prove, in 2017 led to acquittance. 

The theoretical starting points are first presented, in order for the reader to comprehend the terminology used in the analysis and the significance of it. This entails an account of the legal framework for evidence evaluation by the court and the legal framework of the crime in question. It also includes an account of the basic knowledge of Bayesian probability theory as a method of evaluating evidence as well as fundamental knowledge if witness psychology. The critical, quantitative court case analysis has encompassed all cases between 2018 and 2025 where the defendant has been acquitted of the act.

The analysis has identified cases of cognitive bias that has been graded according to a scale of low, medium and high certainty depending on with which certainty the cognitive bias has been confirmed. Reasoning has been carried out regarding whether the cognitive bias in a case has had a causal effect on a prerequisite of the crime not being considered proven by the court. In the case a causal effect has been considered to exist, it has also been graded with a low, medium or high certainty.

30 cases in total where the defendant has been acquitted of the charge has been analyzed. In about 79% of the district court cases, and about 82% of the Court of Appeal cases, a cognitive bias was confirmed with varying degrees of certainty. In about 26% of the district court cases, and about 18% of the Court of Appeal cases, a causal effect was identified between a cognitive bias and the related prerequisite by the court not being considered proven. 

The final assessment is that there are strong signs indicating an occurrence of flawed evidence evaluation within the courts which contributes to prerequisites of a crime not being considered proven. These flaws consists mainly of lackluster credibility assessments regarding oral statements and the person giving them, which carries a significant risk since this type of evidence is more often than not the core evidence of the case.}},
  author       = {{Mårtensson, Alexander}},
  language     = {{swe}},
  note         = {{Student Paper}},
  title        = {{Sanningssökare - En bayesiansk bevisvärderingsanalys av ej styrkta rekvisit i friande mål om människohandel}},
  year         = {{2025}},
}