Measuring trauma and violence exposure: comparing telephone and web survey modes across two population samples

Abstract

Abstract

Background: Monitoring the prevalence of violence is necessary to design adequate policies and mitigate adverse consequences. Methodological debate persists, particularly regarding survey modality for assessing violence and trauma exposure.Objective: This study compares interviewer-administered telephone surveys (CATI) with web-based self-administered surveys across feasibility, safety, sample representativeness, and measurement equivalence.Method: Participants aged 18–74 were randomly drawn from the Norwegian National Population Registry (telephone: n = 4,299; web: n = 1,074). The telephone survey achieved a 25.3% response rate among those reached, while the web survey yielded 10.7%. Using identical violence measures, data were analysed with multivariable logistic regression.Results: Web recruitment was considerably faster (86.2% completed in the first month vs. 2.4% by telephone) and had sixfold lower costs per interview. The CATI survey was quota-based, resulting in sample differences (men 51.1% vs. 37.0%; mean age 48.2 vs. 45.5; age 70–74: 8.9% vs. 3.3%). Several prevalence estimates were comparable, including lifetime forcible rape (7.9% vs. 8.9%), past year less severe physical violence (2.9% in both), and partner control (15.5% vs. 16.9%). Adjusted models showed lower odds in the web survey for severe adulthood physical violence (aOR = 0.71, 95% CI 0.61–0.82) and childhood physical violence (severe aOR = 0.72, 95% CI 0.54–0.96; less severe aOR = 0.81, 95% CI 0.69–0.96). Odds were higher for image-based sexual violations in the web survey (aOR = 1.40, 95% CI 1.04–1.88), particularly among women (aOR = 1.71, 95% CI 1.22–2.41; interaction p = .028). A slightly higher proportion of respondents reported distress and a need for follow-up after participation.Conclusions: Web-based surveys offer substantial efficiency advantages, yet differences in sample composition and variation in several prevalence estimates highlight the importance of considering representativeness, respondent safety, and reporting conditions. Mixed-mode strategies may help improve sample diversity and disclosure.