ABSTRACT
Cybercrime has become a major societal concern, and a better understanding OF cybercrime is needed to target and prevent it more effectively, minimize its consequences, and provide support for victims. Research on cybercrime victimization has exploded in the past few years, but much of it relies on convenience samples and is largely descriptive in nature. The research presented here involves the collection of data from a large sample of Virginia households in 2022 (n = 1,206). The data are analyzed to provide a partial test of routine activity theory to better understand fraud and theft via the Internet. The data provide a solid baseline for describing the extent of cyber victimization across the state. Bivariate and multivariate analyses (logistic regressions) show support for routine activity theory and provide important insights for future research. In particular, we find that certain routine Internet activities may better predict unique forms of cybervictimization than others and that length of time on the Internet is not a good indicator of exposure to motivated offenders. Further, protective guardianship mediates the effects of exposure to motivated offenders; thus, efforts to educate the public on best practices are needed. We conclude that to better assess cybercrime, victimization and engagement, better measurement and longitudinal panel data will be needed.
Acknowledgments
The research was funded by the Coastal Virginia Center for Cyber Innovation (COVA CCI-21-02), 2021/2022. Special thanks to Hayley Jackie and Laquana Askew for research assistance during the project.
Disclosure statement
No potential conflict of interest was reported by the author(s).
Notes
1. There were only 20 and 33 cases in Eastern and Southwest regions and less than 80 in the Southside and Valley regions. Although we do not readily have population estimates for the regions, the smaller sample sizes seem to reflect the rural (less populated) nature of these regions and thus we are reasonably confident of the regional representation of the sample.
2. Income was statistically significant and positive at the bivariate level, and we first ran the model including income. Income was not statistically significant, and the other substantive parameters did not change. However, the sample size dropped by over an additional 85 cases (from 805 to 720) so this model is not presented in text. Results are available from the authors.
3. We note that the lack of statistical significance here seems to be the drop in statistical power with the significant drop in sample size. Univariate comparisons provided in the analyses also show only banking to be statistically significant.