The internal consistency of the five scales (active mediation, safety mediation, restrictions, monitoring, and use of technical controls) developed by EU Kids Online is supported by the present survey data, with Cronbach's alpha³ levels ranging from 0.79 to 0.94 (see Table 1). The scale for child-initiated support, developed for the present survey, achieved an alpha of 0.67; this was considered just sufficient, though not strong, given that no established measure exists for a concept of growing theoretical value.⁴

From the mean scores, it can be seen that parents most favor active safety mediation, along with responsiveness to child-initiated support. Active mediation is favored next, with use of monitoring, restrictions, and technical controls all less popular. The correlation matrix in Table 2 shows these to be positively intercorrelated, suggesting that the more parents do one strategy, the more they do the others, with the exception of the restrictions scale, which is little correlated with most strategies except it is moderately positively correlated with active mediation of Internet use and negatively correlated with child-initiated support.

The correlation matrix suggests a possible factorization based on positive correlations among most variables but a markedly different pattern for parental restrictions.⁵ A factor analysis using principal components analysis and varimax rotation found a solution in which the first two factors accounted for 65% of the variance in the data⁶ (see Table 3). We labeled factor 1 "enabling mediation" (with active mediation of Internet use, active mediation of Internet safety, technical controls, and parental monitoring all loading highly) and factor 2 "restrictive mediation" (characterized by a positive loading for restrictive mediation). Child-initiated support loaded positively on the "enabling mediation" and negatively on "restrictive mediation" factors. Two composite scales were constructed by summing the scores for the items on the component scales. On the basis of its positive loading and the relevant literature on family dynamics, we decided to include child-initiated support in calculating the enabling mediation scale, along with the above-listed high-loading items. It should be borne in mind, however, that child-initiated support is negatively related to the restrictive mediation factor. For enabling mediation, the mean was 3.00 (range 0-5, SD = 0.96) and for restrictive mediation, it was 2.47 (range 0-5, SD = 1.35).

Table 3 Factor Analysis of Parental Mediation Strategies Enabling Mediation Restrictive Mediation Active mediation of Internet use 0.77 0.28 Child-initiated support 0.57 0.54 Active mediation of Internet safety 0.78 0.22 Technical controls 0.68 00? Parental monitoring 0.79 0.07 Parental restrictions 0.14 0.91