19 February 2009
RE: COMMENT ON CHILD SUN PROTECTION: SUN-RELATED ATTITUDES MEDIATE THE ASSOCIATION BETWEEN CHILDREN'S KNOWLEDGE AND BEHAVIOURS
We thank de Cock and colleagues1 for their perspective on our results. We were guided by the relationships in our data, of which some could have been due to chance. We also acknowledge that the children of different societies will vary in the degree that parents influence their behaviour at different ages. We will now address the specific points raised.
First, when we investigated the associations between child sun protection-related knowledge, attitudes and usual behaviours, combining the age groups (age 8 and age 12 years) our model results indicated that the relationship between knowledge and behaviour was fully mediated by attitudes.
When age groups were considered in isolation, this was not the case. In the model for the older children, the path from knowledge to behaviour was not statistically significant (path coefficient = 0.08 (0.06), t= 1.32). However, for the younger children, the relationship between sun-protective knowledge and behaviour existed independently of the effect of attitude (path coefficient = 0.19 (0.06), t= 2.95). Attitudes partially mediated the relationship between knowledge and behaviour, where some of the effect of knowledge was explained by attitudes, and some was independent of attitudes. This finding was contrary to what de Cock et al. predicted.
With respect to parental influence, as children progress towards adolescence their behaviours are likely not only to become less influenced by their parents, as de Cock et al. suggest, but also more influenced by the pro-tan models that tend to be provided by their ‘aspirational’ peer groups, older adolescents and young adults. This helps to explain why their attitudes rather than knowledge are related to behaviour, and is consistent with the findings by O'Riordan et al. (2003) of less protective behaviours among the older age group, as we found in our study.
Second, we agree with de Cock et al. that causality cannot be inferred simply from a statistical association in a cross-sectional study. Testing our findings among a sufficiently large sample of children using a longitudinal study design would be the ideal. However, as the resources necessary for such a study were not available to us, the assumption was made that knowledge was a contributing factor for attitudes.
Apart from parental influence, skin colour and family history of skin cancer are plausible factors influencing child sun protection, as de Cock et al. suggest. However, we were unable to obtain the substantial additional funding that would have been required to conduct family interviews, take objective measures of child skin colour and access reliable skin cancer records, if available. However, it would be valuable that these factors be considered and budgeted for in future child sun protection research.
Finally, we confirm that sun-protective knowledge alone may not be sufficient to influence sun-related behaviours, but may be a necessary component or moderator, together with positive attitudes, for behavioural change. We agree that further research would increase understanding of these relationships.