There is much research studying bullying and bystanding behaviors during adolescence. Recently, research has focused on social variables in an attempt to prevent or discourage bullying and encourage defending behaviors. This dissertation attempts to merge this research with a sociocognitive perspective. Moral disengagement, a sociocognitive variable, may help researchers have a better understanding of why adolescents bully their peers. This dissertation studies the role of moral disengagement in predicting bullying and bystander behaviors in adolescents from the US and Sweden. Results and implications of these findings are also discussed.

Criminologists have long since marginalized passivity as a variable of interest when studying the phenomenon of harm-doing. In this thesis, I explore the role of passivity in such instances and build a case for its centrality to deviance. I also undertake a number of other tasks. First, I review the extent to which research throughout the academy has connected passivity to violence. Second, I explore whether criminological theories have incorporated the variable of passivity and how they could. Lastly, I reflect on why more work on passivity has not been done given its manifest connection to harm, and I offer suggestions on how criminology can move forward in integrating bystander behavior in its theories of harm-doing.

Tidigare forskning har visat att barn kan delas in i olika roller vid mobbningssituationer i grundskolan,det är därför av intresse att hitta faktorerna som påverkar vilken roll som elever kan komma att ha. I denna uppsats är syftet att analysera samband mellan motivationer (Extrinsic, Amotivation,Identification) och beteenden (Defending, Passive och Pro-Bullying). Det är även intressant attundersöka om det finns eventuella skillnader i beteende beroende på ålder och kön. För attanalysera samband mellan psykologiska begrepp används latenta variabler i samröre medfaktoranalysmodeller (SEM), denna typ av modell tar inte hänsyn till skillnader mellan klasser,således används även multilevelmodeller. Den senare av de två modellerna tar hänsyn till skillnadermellan klasser men inte det faktum att variablerna är latenta. En ytterligare aspekt av uppsatsen är att jämföra metoderna med varandra och analysera eventuellaskillnader i slutsats. Det kommer att användas två typer av multilevelmodeller i uppsatsen, en medmedelvärden som skattning på den latenta variabeln och den andra multilevelmodellen använderfaktorpoäng som beräknas via en konfirmativ faktoranalys. Resultatet visade att det finns signifikanta samband mellan alla typer av beteende och motivationer iminst en av metoderna som användes. I alla metoder fanns det signifikant positiva relationer mellanPassive och Amotivation, Defending och Identification samt mellan Pro-bullying och Extrinsic.Samband som hade en signifikant negativ relation i alla metoder är Passive och Identification, Pro-bullying och Identification samt mellan Defending och Amotivation. Relationer som är signifikanta men endast i en metod är Pro-Bullying och Amotivation, Passive och Extrinsic samt Defending ochExtrinsic. Dessa relationer har positiva samband. Det finns ingen relation mellan kön respektive ålder och ett specifikt beteende som blev signifikant ialla metoder. Relationer mellan beteende och ålder respektive kön som blev signifikant i minst enmetod är Passive och kön, Defending och kön samt mellan Defending och ålder. Resultatet från dettaär att killar är mindre passiva och mer Defending jämfört med tjejer. Relationen mellan Defendingoch ålder är negativ. Intraclass-korrelationen och Likelihood ratio-test visar att det finns signifikant varians mellan klasser ide tre beteendena när en modell utan förklarande variabler testas. Testet visade att det inte fannssignifikant varians i beteende mellan skolor. Likelihood ratio-test på modeller med förklarandevariabler visar att det endast vid beteendetypen Defending fanns en signifikant skillnad mellan eleveri olika klasser. I övriga beteende finns det alltså inte signifikanta skillnader mellan klasser. Skillnaden mellan faktorpoäng och medelvärde blev väldig liten, det blev endast skillnader irelationen mellan Defending och ålder. Mellan SEM och multilevel är skillnaden större och där finnsdet skillnader i slutsats i fem stycken relationer. Dessa är Passive och Extrinsic, Pro-Bullying ochAmotivation, Passive och kön, Defending och Extrinsic samt mellan Defending och kön. / Previous research has shown that children can be divided into different roles during situations ofbullying in primary school. Therefore, it is of interest to find factors that affect what those roles couldbe. In this thesis, the purpose is to analyze the relation between motivation (Extrinsic, Amotivation,Identification) and behavior (Defending, Passive, Pro-Bullying). It is also of interest to analyzepossible differences in behavior depending on age and gender. To be able to analyze relationsbetween psychological concepts, latent variables will be used in conjunction with models for factoranalysis (SEM). This type of model does not take differences between classes into consideration,therefore multilevel models will also be used, where the latter takes differences between classes intoconsideration but not the fact that the variables are latent. Another aspect of this thesis will be to compare the methods to each other and analyze potentialdifferences in drawn conclusions. There will be two different types of multilevel models, one ofwhich uses mean as the estimator for the latent variable while the other model will be using factorscores that are calculated with a confirmative factor analysis. Results showed that there are significant relations between all types of behavior and motivations inat least one of the methods used. In all the methods, there were significant positive relationsbetween Passive and Amotivation, Defending and Identification, as well as Pro-Bullying and Extrinsic. Relations that had a significant negative association were Passive and Identification, Pro-Bullying andIdentification, as well as Defending and Amotivation. Relations that were significant, but only in one method, are the following: Pro-Bullying andAmotivation, Passive and Extrinsic, Defending and Extrinsic. These have a positive relation. There is no relation between gender, or age, and a specific behavior that was significant in all themethods used. Relation between behavior and age, or gender, that was significant in at least one, isPassive and gender, Defending and gender, Defending and age. The results from this show that boysare less passive and more defending compared to girls. The relation between Defending and age isnegative. The intraclass correlation, and Likelihood ratio test, shows that there is a significant variancebetween classes in the three behaviors when a model without explanatory variables is being used.The tests also indicated that there was no significant variance in behavior between schools. ALikelihood ratio test of models with explanatory variables show that the only type of behavior with asignificant difference between students and classes is Defending. The difference between factor scores and mean was very small, there were only differences in therelation between Defending and age. Between SEM and Multilevel, the difference is greater, wherethere are differences in conclusion in five of the relations; Passive and extrinsic, Pro-Bullying andAmotivation, Passive and gender, Defending and Extrinsic, as well as Defending and gender.