Perceived safety in digital information environments: objective violence, online disinformation, internet use, and...

Perceived safety is a psychologically consequential judgment shaped by objective threats, prior victimization, and mediated information. However, few studies jointly examine longitudinal crime data, national digital-information environments, and individual Internet use.

We integrated European Social Survey Rounds 1-11 with homicide data from the United Nations Office on Drugs and Crime, Digital Society Project indicators, and World Development Indicators. Study 1 examined 308 country-round-year cells from 38 European countries, representing 442,100 respondents, using weighted country and survey-round fixed-effects models. Study 2 analyzed 408,149 respondents from 37 countries and 345 country-round-year cells using cell fixed-effects models.

In Study 1, conventional country-clustered models indicated that within-country increases in lagged homicide were associated with lower perceived safety (-1.42 percentage points per within-country SD, 95% CI [-2.47, -0.38], p = 0.008), whereas increases in expert-coded government online disinformation were associated with higher perceived safety (+0.87 points, 95% CI [0.34, 1.40], p = 0.001), particularly in countries with higher long-run homicide rates (+0.52-point interaction, 95% CI [0.24, 0.80], p < 0.001). However, the corresponding restricted wild-cluster bootstrap p-values were 0.087, 0.052, and 0.091, and country-specific trends attenuated the two main effects. In Study 2, more frequent Internet use was associated with higher perceived safety (+1.31 points per SD, 95% CI [0.83, 1.78], p < 0.001). Victimization was associated with substantially lower perceived safety (-13.65 points), but the positive association between Internet use and perceived safety was stronger among victims (+2.67-point interaction, 95% CI [2.00, 3.34], p < 0.001). These three individual-level terms remained significant at p < 0.001 under 9,999-replication wild-cluster inference, and their directions remained unchanged when each cell received equal total weight.

The national-level findings should be considered provisional. The individual-level findings are consistent with-but do not establish-mechanisms involving access to information, social support, or perceived efficacy. Given the observational design, all results represent conditional associations rather than causal effects.