The vast majority of surveys analysed by the UK social research community employ complex sample designs and weighting adjustments, yet are often treated as un-weighted simple random samples by analysts. This is unfortunate because failing to take these factors into account is likely to result in biased point and variance estimation. In this short article, I use data from the 2000 UK Time Use Survey (UKTUS) to show how these factors should be incorporated into the estimates produced from complex surveys and to illustrate the threats to accurate inference if they are ignored. This is not intended as a detailed statistical treatment of these issues but as a general discussion aimed at substantive and policy-oriented users of large scale social surveys. Readers in search of more detailed treatments are directed toward Kish (1962) and Groves et al (2004).