Abstract
Given increasing calls for children and young people to participate via the Internet in civic and political activities), this article examines how far, and with what success, such participation is occurring among UK teenagers. Findings from a national survey conducted by the UK Children Go Online project show that young people are using the Internet for a wide range of activities that could be considered ‘participation’, including communicating, peer-to-peer connection, seeking information, interactivity, webpage/content creation and visiting civic/political websites. The findings are closely examined using path analysis techniques to identify the direct and indirect relations among different factors that may explain how and why some young people participate more than others. The results suggest that interactive and creative uses of the Internet are encouraged by the very experience of using the Internet (gaining in interest, skills, confidence, etc.) but that visiting civic websites depends primarily on demographic factors (with older, middle-class girls being most likely to visit these sites). Finally, cluster analysis is used to identify three groups of young people – interactors, the civic-minded and the disengaged – each of which is distinctive in its social context and approach to the Internet.
Acknowledgements
This paper reports on research funded by an Economic and Social Research Council grant (RES-335-25-0008) as part of the ‘e-Society’ Programme, with co-funding from AOL, BSC, Childnet International, Citizens Online, ITC and Ofcom (see www.children-go-online.net). The authors would like to thank the Advisory Panel (Camille de Stempel, Karin Sieger, Simon Kinnersley, Andrea Millwood Hargrave, Nigel Williams, Stephen Carrick-Davies, Mary Louise Morris, John Fisher and John Carr), and BMRB for conducting the survey.
Notes
1 See the United Nations Convention on the Rights of the Child, www.unicef.org/crc/crc.htm
2 Examples of such sites include youngGov (www.young.gov.uk), the UK government's new web portal for 11- to 16-year-olds, Children's BBC (www.bbc.co.uk/cbbc), which encourages creating and disseminating material online and on television, sending feedback to television programmes, using story circles to engage disadvantaged and disengaged groups (e.g. disabled children), and the Childnet Academy (www.childnet- academy.org, part of Childnet International), which runs an annual award competition for websites made by children and young people that benefit other children worldwide.
3 ACORN is a standard classification of residential neighbourhoods used by UK market research organizations as a segmentation system that enables consumers to be classified according to the type of area they live in (specifically, by postcode).
4 Although broadly representative of the UK population, the sample sizes for specific minority groups were too small to include a breakdown in the analysis (of the 1511 respondents, 91 respondents were Asian, 35 black, 4 Chinese and 39 of mixed ethnicity).
5 The model in shows, for example, that sociodemographic variables are understood to directly influence interaction and civic participation on the Internet (visualized by an uninterrupted arrow from demographics to interaction and civic website visiting). However, gender, for example, is also assumed to have indirect influences on participation through the mediating effect of other variables (Internet use). Another way of saying this is that the effects of sociodemographic characteristics on participation are understood to be mediated by Internet use variables.
6 We caution here that using dichotomous and ordinal variables (such as gender and social grade) in path analysis is problematic. However, this is mainly an issue when endogenous (dependent) variables are dichotomous or ordinal (Bollen Citation1989; Kline Citation1998). In the present study, the only dichotomous and ordinal variables used were gender and social grade, and both are used only as exogenous (independent) variables.
7 In the survey, we asked about eight interactive activities. As the factor analysis showed that these did not form a single scale (with doing a quiz and sending pictures/stories to a website being separate from the other six), these two items are omitted from the present analysis.
8 Kline (Citation1998) notes that a non-normal distribution of the variables inflates the chi-squared values and would thus lead the researcher to be conservative in the acceptance of a model (i.e. if the chi-squared value is inflated, the model would not be accepted as a good fit even if in reality it is a good fit). As shown in , years online, self-efficacy, average time online per day all have a relatively neat normal distribution. However, both interaction and civic website visiting are skewed towards the lower end of the scale and have relatively high kurtosis. The decision was made to use these variables and to accept the more conservative estimate of which models fit the data best.
9 In path analysis, the desired chi-squared statistic for the whole model should ideally be as low as possible. The hypothesis is that the model fits the data, and a p-value higher than 0.05 is required if the hypothesis is not to be rejected. In other words, a significant chi-squared statistic (p < 0.05) means that the model does not fit. However, for the individual paths to be included in the model, the relationships between the variables must be significant because this indicates that the observed relationships are not due to chance.
10 The cluster analysis used the furthest neighbour technique with standardized values in the statistical program SPSS 11.5.