디지털 성범죄 피해유형 분석과 영향요인 : 잠재계층분석을 활용하여 = A Study on the Classification of Digital Sex Crime Victimization and Its Influencing Factors : An Application of Latent Class Analysi

다국어 초록 (Multilingual Abstract)

This study applies Latent Class Analysis （LCA） to classify digital sex crime victimization and identify the factors influencing each type. Enabled by anonymity, rapid dissemination, persistence, and the absence of spatial and temporal limits, digital sex crime produces diverse, overlapping patterns of harm that frequently co-occur. Yet prior research has relied on binary or frequency-based measures, or on researcher-defined typologies, thus failing to capture the heterogeneity of victim experiences. Drawing on the cyber lifestyle-routine activity theory and online disinhibition theory, this study classifies victimization into empirically distinct types based on respondents' actual experiences and analyzes the determinants of each.
Data were collected through a nationwide online survey of 500 South Korean adults aged 20 and older, with equal allocation of male and female respondents. The survey measured experiences ranging from unwanted sexual content, online sexual harassment, and coerced exposure and solicitation to cyberstalking, illegal filming and non-consensual distribution of intimate images, and deepfake victimization. Independent variables formed two blocks, namely cyber lifestyle-routine activity factors （online exposure, proximity to motivated offenders, online guardianship, and target attractiveness） and a disinhibition factor （online deviant lifestyle）, with age, education, and income as controls, and the analysis proceeded from LCA to multinomial logistic regression.
The LCA identified three latent classes. The low-risk group showed minimal victimization overall but a relatively high probability of receiving unwanted sexual content, reflecting the indiscriminate nature of digital sex crime. The medium-risk group was characterized by interaction-based victimization such as online sexual harassment, coerced exposure, and solicitation, while the high-risk group showed a high likelihood of severe victimization such as illegal filming, non-consensual image distribution, and deepfake abuse that entails irreversible harm. Digital sex crime victimization is thus qualitatively heterogeneous rather than homogeneous.
Multinomial logistic regression showed that online exposure （time spent online） predicted medium-risk membership and SNS update frequency predicted the high-risk group, partially supporting the exposure hypothesis. Proximity to motivated offenders, such as interacting with strangers and being tagged on others' accounts, significantly increased both medium- and high-risk membership, partially supporting the proximity hypothesis. Online deviant lifestyle was among the strongest predictors across both higher-risk groups, supporting the disinhibition hypothesis, whereas online guardianship and target attractiveness were not significant and those hypotheses were rejected. Age was a strong predictor, with younger individuals more likely to belong to the higher-risk groups. These results indicate that guardianship and attractiveness must be reconceptualized for online environments, with guardianship operating at the structural and platform level and attractiveness through accessibility and exposure, and that the environmental-structural mechanism of routine activity theory and the behavioral-psychological mechanism of disinhibition operate complementarily. The non-significance of gender does not imply that digital sex crime is gender-neutral; its nature as gender-based violence remains.
Theoretically, the study adopts a person-centered approach, integrates the two theories with latent class modeling, and provides adult-population evidence beyond prior research's focus on children and adolescents. For policy, it underscores differentiated, type-specific prevention; reduced online exposure and proximity; stronger platform responsibility through structural safeguards such as default safety settings, automated filtering, and restricted anonymous access; and digital citizenship education targeting younger adults, all within a gender-based-violence framework. Limitations include the cross-sectional design, self-report bias, incomplete measurement of emerging forms of technology-facilitated sexual violence, and limited attention to platform-level factors, suggesting longitudinal, offender-side, qualitative, and cross-national designs for future research. In conclusion, this study demonstrates that digital sex crime victimization is a heterogeneous phenomenon that can be effectively classified through Latent Class Analysis, offering important implications for theory, empirical research, and the design of differentiated prevention policies.

목차 (Table of Contents)

• 목차
• 제1장 서 론 1
• 제1절 연구의 목적 1
• 제2절 연구의 범위 및 방법 4
• 1. 연구의 범위 4
• 2. 연구의 방법 5
• 제2장 이론적 배경 8
• 제1절 디지털 성범죄의 이해 8
• 1. 디지털 성범죄의 정의 8
• 2. 디지털 성범죄의 2차 피해 12
• 3. 디지털 성범죄의 실태 16
• 제2절 디지털 성범죄 관련 이론 18
• 1. 사이버 생활양식-일상활동이론 18
• 2. 탈억제이론 25
• 제3절 선행연구 검토 32
• 1. 디지털 성범죄 관련 선행연구 32
• 2. 사이버 생활양식-일상활동이론과 디지털 성범죄의 관계 35
• 3. 탈억제이론과 디지털 성범죄의 관계 40
• 제3장 연구의 설계 43
• 제1절 연구모형 및 연구가설 43
• 1. 연구모형 43
• 2. 연구가설 44
• 제2절 변수의 조작적 정의 및 측정 48
• 1. 종속변수 48
• 2. 독립변수 50
• 제3절 자료수집 및 분석방법 53
• 1. 자료수집 53
• 2. 분석방법 54
• 제4장 연구의 결과 및 논의 59
• 제1절 주요 변수의 기술통계 분석 결과 59
• 1. 연구대상자의 인구사회학적 특성 분석 결과 59
• 2. 측정도구의 타당도 및 신뢰도 분석결과 60
• 3. 주요 변수의 기술통계 및 상관관계 분석 62
• 제2절 디지털 성범죄 피해의 유형화 69
• 1. 디지털 성범죄 피해유형의 특성 69
• 2. 잠재계층 모델 선정 70
• 3. 잠재계층 명명 73
• 제3절 디지털 성범죄 피해유형의 영향요인 분석 결과 76
• 제4절 가설검증 및 논의 80
• 1. 가설 검증 80
• 2. 논 의 90
• 제5장 결론 105
• 참 고 문 헌 113
• ABSTRACT 125
• 부 록 128