How The Internet Skills Of European 11- To 16-Year-Olds Mediate Between Online Risk And Harm

Abstract

Not all children who use the internet will experience harm from the online risks they encounter. One of the factors that might moderate the relationship between risk and harm is children's internet skills. As there has been little research on this topic, this article examines the influence of internet skills on the prevalence of online risks and the degree to which 11- to 16-year-olds experience being harmed by these risks, using data from the EU Kids Online project. The findings suggest that, whilst older children (aged 13–16) are exposed to more online risks, younger children (11–12) report more often being harmed by these risks. After controlling for differences between children due to demographics and internet experience, as well as country differences (using multilevel analysis), the findings reveal that children with more self-reported internet skills experience more risks online. Such skills do not seem to contribute much to differences in being harmed by online risks.

Keywords:

• digital skills
• online risks
• harm
• children
• internet experience
• multilevel analysis

Notes

^1. In the questionnaire, the measure of harm was conditional upon whether a child had encountered the respective online risk. For this reason, the harm variable was constructed based on the number of risks encountered and the amount of harm experienced. The values assigned were as follows: 0 = encountered at least one of the four online risks studied, but did not report harm from any of these experiences; 1 = encountered one risk and reported harm from this one risk; 2 = encountered two risks and reported harm from only one of these risks; 3 = encountered three or four risks and reported harm from only one of these risks; 4 = encountered two risks and reported harm from both risks; 5 = encountered three or four risks and reported harm from two or more of these risks.

^2. This measure is the Kuder-Richardson Formula 20 (KR-20) that is analogous to Cronbach's alpha, but computes the reliability coefficient of a set of dichotomous items. It ranges between 0 and 1; the higher the scores, the more reliable the index variable.