MOVING ON-LINE? AN ANALYSIS OF PATTERNS OF ADULT INTERNET USE IN THE UK, 2002–2010

Abstract

This paper presents the results of multi-variate analyses of the social, economic and educational characteristics associated with reporting both access to the Internet and using the Internet for four key purposes: banking and finance, purchasing goods or services, accessing government or official services and looking for work or employment. The research was conducted using nationally representative, individual-level, repeated cross-sectional data (n = 47,001) collected in annual surveys in the UK between 2002 and 2010. The results of the analyses show that although Internet access and use have increased over the period studied, both continue to be structured according to occupational class, educational background and, to some extent, age. The sex and ethnicity of respondents had little impact on the probability of reporting Internet access and were only strongly related to using the Internet for purchasing goods or services. Additionally, the presence of children in a household was unimportant in relation to both Internet access or use. While the findings differ slightly from previous studies they confirm that both Internet access and use remain structured along socio-economic and educational lines that work against already disadvantaged groups. This has remained the case in the UK throughout the 2000s despite considerable technological change and policy interventions specifically targeting marginalized sections of society. The paper concludes that policy interventions aimed at both increasing and widening Internet access and use will be ineffective unless the social, rather than technological, basis of inequalities in access and use are recognized.

Keywords:

• ICTs
• digital divide
• e-commerce
• e-government
• social policy

Acknowledgements

The authors thank Fiona Aldridge and NIACE for their permission to use the data sets for the secondary analyses reported in this paper. The authors also thank the anonymous referees for their comments on earlier drafts of the paper.

Notes

The terms ‘predictive power’ and ‘explanatory power’ are used in this paper to refer to how well individual cases within the sample could be predicted by the model. They are not intended to refer to the more general ability of the model to predict any of the dependent variable outcomes in a wider population, which cannot be confidently established from these analyses.

It should be noted that, even if the correct assumptions were met, the authors would be reluctant to use inferential statistical testing. As Cohen (1994) and Gigerenzer (2004) remind us, such tests provide the probability of observing our data given a null hypothesis p(D|H ₀) when we actually want to know the probability of the null hypothesis given our data p(H ₀|D). As Lindley (1957) demonstrated more than 50 years ago and Jeffreys (1939) noted nearly two decades earlier, the probabilities are not interchangeable and can in fact be radically different. Such tests, therefore, have the potential to be very misleading.