For this study, over 10 000 media texts on the 2001, 2004 and 2007 elections were collected from television, newspapers, radio and the internet. I used a combination of methods to analyse segments of this material. Quantitative content analysis is one of the few ways of systematically classifying and describing media content and I used this method on a sample of 965 election reports (Tables i and ii). I also used qualitative analysis for specific samples, paying particular attention to discourse - words, text and talk – as well as to the visual dimension of television clips, newspaper, magazine and website photographs and graphics. The other main method was a process I called ‘media mapping’ to trace the subject and sources of media reports across each day of the 2007 election campaign. This allowed me to analyse processes of agenda-setting, news cycles and news flows (mostly outlined in Chapter 8).

Space limits prevent me from describing the methodology in its entirety but, given the quantitative basis for some of the claims I make, it is important that I outline the sample used for the content analysis. To narrow down the sample to a reasonable size appropriate for content analysis, I used a systematic sampling method often known as the ‘nth’ method or ‘constructed week’. This approach involves selecting every nth unit from the total population available.