Lars Roar Frøyland et al – NSfK ResearchSeminar 2021

Typologies of sexual assaults among young people: A latent class analysis approach

Lars Roar Frøyland, Senior Researcher*
Patrick Lie Andersen, Researcher*
Kari Stefansen, Research Professor*
*Norwegian Social Research (NOVA) – Oslo Metropolitan University


Sexual assault is a serious social problem among young people and the long-term outcomes of such experiences may be dire. However, the scientific evidence on the incidentspecific characteristics of sexual assault situations among young people is scarce. In this paper we critically engage with existing conceptualizations and aim to develop a typology that can inform theoretical development as well as prevention policies. The typology is developed by using Latent Class Analysis (LCA) to delineate distinctly different classes of sexual assault
among young people. LCA is a statistical method for identifying membership in unmeasured latent classes based on response patterns in observed study variables. Sexual assault is defined here as experiences ranging from indecent exposure to rape.

In this paper our focus is what latent categories of sexual assaults young people report and along which dimensions they can be meaningfully distinguished. We also plan to investigate how the latent categories of sexual assault differ in their association to individual
and situational characteristics and the outcomes of the experience.

Labels and categorizations of sexual assault

Labels and categorizations of sexual assault are important to address as they offer different inroads to understanding the phenomenon. Our point of departure here is that differentcategorizations may be useful for different purposes. This does not preclude a need for critical
assessment of the implications of relying on different types of categorization of sexual assault. An example is how penal categories and ensuing hierarchizations often are taken for granted as valid and justified through prevalence research.

The literature offers a range of different categorizations of sexual assault, most often based on external or theoretical criteria. In survey research, and following the lead of Mary P. Koss, much attention has been directed at developing measures that encompass all acts that is
considered rape in the legal sense (Krebs, 2014; Stefansen, Løvgren, & Frøyland, 2019). There are also studies on specific forms of rape defined in the penal code, such as incapacitated rape (Stefansen, Frøyland, & Overlien, 2020) and statutory rape (Small, 2020) as well as ‘minor’ sexual offences such as unwanted touching (Stefansen, 2019) and sexual harassment (Fileborn, 2019). Sexual assaults can also be categorized according to the relationship between the victim and offender, with labels such as stranger rape and acquaintance rape (Friis-Rødel, Leth & Astrup, 2021; Persson & Dhingra, 2020), date rape (Ogunwale & Oshiname, 2017), or marriage / partner rape (Tarzia, 2020). The latter type of differentiation has been important in highlighting the prevalence of sexual assault and rape beyond the stereotypical stranger rape script. Other types of labels and categorizations highlight the social or personal context of sexual assaults. Examples include the campus rape concept (Abu-Odeh, Khan, & Nathanson, 2020) and the idea of “vulnerability rapes.” The latter is used for instance in statistics from the Norwegian Police with reference to rapes against victims who “live a life in loneliness, and are often isolated from society. They are in a quite vulnerable situation, which makes them at risk for repeat assaults. Prostitutes, drug addicts, and persons in institutional care, are particularly at risk” (Kripos, 2020: p. 8, authors’ translation).

The LCA-approach to sexual assault categorization: Previous studies

LCA is a more inductive approach compared to the theoretically driven categorizations described above, as it allows for patterns unknown to the researcher to form the basis of labelling and typologization. It also allows for the inclusion of a range of situational characteristics in the construcution of categories as such. LCA has been used in studies on different issues to identify empirically based categorizations. To date there are only a few studies using this approach in relation to sexual offending and sexual victimization. Among Canadian convicted serial sex offenders, LCA has been used to identify target selection scripts (Deslauriers-Varin & Beauregard, 2010). Similar methods have been used among Danish adults (Heinskou, Schierff, Ejbye-Ernst, Bank Friis, & Liebst, 2017) and US college students (Khan, Hirsch, Wamboldt, & Mellins, 2018) to identify classes of sexual assault victimization. The main intake from previous studies is how classes of sexual assault vary according to individual, relational, and situational characteristics of the assault.

Our analyses build on this body of work and aim to contribute to the discussion on how sexual assaults can be meaningfully distinguished so as to capture real life experiences and their relationship to background factors and adverse outcomes. In particular, we include a wide
range sexual assault experiences, and victim descriptions and understandings of such experiences to delineate a typology of sexual assault using nationally representative samples of Norwegian young people.

Data and methods

The data are obtained from the UngVold (Youth Violence) study, conducted by Norwegian Social Research. The study comprises two large-scale surveys among students (18–20-years old) in the final year of high school in Norway, in 2007 (n = 7,033) and 2015 (n =4,530). The data material from the surveys are pooled together in order to obtain the necessarry statistical power for the proposed analyses. Both surveys included questions on sexual assaults during childhood and adolescence, and an “incident report” relating to the first and last
incident, with questions about the perpetrator and the situation. Sexual assault was measured using a behaviorally specific questionnaire, including experiences such as indecent exposure, unwanted sexual touching, and penetration. Only respondents that answered in the affirmative to at least one of the questions on sexual assault completed the incident report. The preliminary analyses are based on the first reported incidence of sexual assault by the victims.

Latent class analysis (LCA) was utilized for identifying classes of sexual assault among Norwegian adolescents. Classes were identified based on information on the age of the perpetrator, the victims’ relationship to the perpetrator (friend, boyfriend/girlfriend, acquaintance, stranger, trainer/teacher, and family member), whether alcohol was involved in the situation around the assault (victim, assailant, or both drunk when it happened), and nine statements on how the victims would describe what happened in the situation, with response options ranging from Very true (1) to Not true at all (4).


The included surveys contain data on 1,863 sexual assaults experienced by Norwegian 18–20-year olds. Sexual assaults predominantly involve a female victim and a male assailant, with 85 percent of the victims being female and 90 percent of the assailants being male. The
overall mean age of the perpetrators were 24 years old, while the mean age of the victims were 14 years old. Strangers were the most common assailants, responsible for one out of three sexual assaults, while acquaintances were the assailant in 27 percent of the situations. Friends were the perpetrator in 18 percent of the assaults, while boyfriends/girlfriends and family members each were the perpetrator in around one out of ten assaults.

In the latent class analyses we tried models with different number of classes, from two to 13. The 4-class LCA model displayed the best fit based on BIC and AIC values and theoretical reasoning. The first class (17.9%) comprises assaults where a young (typically) girl is tricked into a situation that ends up with a sexual assault by an older person. The mean age of the assailants were 46 years old, while the mean age of the victims were 11 years old. About half of the assailants were strangers to the victims, while a quarter were family members. The victims commonly interpreted the assault as caused by getting involved in something they were too young to understand or being tricked or persuaded. Alcohol was seldomly involved in the incidents. This class captures what is often described as childhood sexual abuse.

The second class (35.9%) comprises assaults involving alcohol use and strangers or acquaintances as assailants. The mean age of the assailants were 20 years old, while the mean age of the victims were 15 years old. 37 percent of the assaults were conducted by a stranger,
while an acquaintance was the perpetrator in 30 percent of the instances. The victims typically do not report any forms of coercion or being restrained. This class captures youth sexual assault by strangers and aquaintances.

The third class (27.0%) also comprises sexual assaults that involve the use of alcohol, but unlike the second class, the perpetrators are typically either friends or a boyfriend/ girlfriend. The perpetrator and the victim are also closer in age, with mean ages of respectively 17 and 14. The victims commonly report being exposed to mild coercion, being persuaded or tricked, or participating voluntary, but regretting later. Many also reports being too young to understand the situation. This class captures youth sexual assault in close or trusting relationships.

The final class (18.8%) of sexual assaults among Norwegian adolescents comprises violent assaults from strangers or acquaintances. The mean age of the perpetrators were 20 years old, while the mean age of the victims were 14 years old. 96 percent of the victims report being restrained, while three out of four were the object of heavy coercion. 40 percent were threatened with violence, while 30 percent were violently victimized.

Summary and discussion

The latent class analyses return four distinctly different classes of sexual assault among
Norwegian adolescents. The next step of the analyses is to investigate how the classes differ in
their association to both individual and situational characteristics, as well as outcomes of the
experience. This understanding of typologies of sexual assault among adolescents, and their
association to related factors, may facilitate the creation of targeted preventive measures, taking
incident-specific characteristics into consideration.



Abu-Odeh, D., Khan, S., & Nathanson, C. A. (2020). Social constructions of rape at Columbia University and Barnard College, 1955–90. Social Science History, 44(2), 355–379.

Deslauriers-Varin, N., & Beauregard, E. (2010). Victims’ routine activities and sex offenders’ target selection scripts: A latent class analysis. Sexual Abuse, 22(3), 315–342.

Fileborn, B. (2019). Naming the unspeakable harm of street harassment: A survey-based examination of disclosure practices. Violence Against Women, 25(2), 223–248.

Friis-Rødel, A. M., Leth, P. M., & Astrup, B. S. (2021). Stranger rape; distinctions between the typical rape type and other types of rape. A study based on data from Center for Victims of Sexual Assault. Journal of Forensic and Legal Medicine, 102159.

Heinskou, M., Schierff, L. M., Ejbye-Ernst, P., Bank Friis, C., & Liebst, L. (2017). Seksuelle krænkelser i Danmark – Omfang og karakter. Det Kriminalpræventive Råd.

Khan, S., Hirsch, J., Wamboldt, A., & Mellins, C. (2018). “I didn’t want to be ‘that girl'”: The social risks of labeling, telling, and reporting sexual assault. Sociological Science, 5, 432–460.

Krebs, C. (2014). Measuring sexual victimization: On what fronts is the jury still out and do we need it to come in? Trauma, Violence, & Abuse, 15(3), 170–180.

Kripos. (2020). Voldtektssituasjonen i Norge 2019. Kripos.

Ogunwale, A. O., & Oshiname, F. O. (2017). A qualitative exploration of date rape survivors’ physical and psycho-social experiences in a Nigerian university. Journal of Interpersonal Violence, 32(2), 227–248.

Persson, S., & Dhingra, K. (2020). Attributions of blame in stranger and acquaintance rape: A multilevel meta-analysis and systematic review. Trauma, Violence, & Abuse, 1524838020977146.

Small, J. L. (2020). Conceptualizing consent: How prosecutors identify sexual victimization in statutory rape cases. Inquiry, 45(1), 111–131.

Stefansen, K. (2019). Understanding unwanted sexual touching. A situational approach. In M. Heinskou, M. L. Skilbrei, & K. Stefansen (Eds.), Rape in the Nordic Countries. Continuity and Change (pp. 49–65). Routledge.

Stefansen, K., Frøyland, L. R., & Overlien, C. (2020). Incapacitated sexual assault among youths: Beyond the perpetrator tactics framework. Journal of Youth Studies.

Stefansen, K., Løvgren, M., & Frøyland, L. R. (2019). Making the case for ‘good enough’ rapeprevalence estimates. Insights from a school-based survey experiment among Norwegian youths. In M. Heinskou, M. L. Skilbrei, & K. Stefansen (Eds.), Rape in the Nordic Countries. Continuity and Change (pp. 66–82). Routledge.

Tarzia, L. (2020). Women’s emotion work in the context of intimate partner sexual violence. Journal of Family Violence.