Reasoning With Evidence – Development of a Scale Professor Jim Ridgway, Dr. Sean McCusker & James Nicholson University of Durham, UK Email jim.ridgway@dur.ac.uk; sean.mccusker@dur.ac.uk, j.r.nicholson@dur.ac.uk Introduction Much of the data presented by politicians and the media is multivariate in its nature. However, in the UK at least, the general public has little training to deal with such information. It is reasonable to explore the school curriculum to determine the nature and extent of students’ preparation for dealing with multivariate data. In the UK, high-stakes examinations have a profound effect on the curriculum; it follows that one can best understand the curriculum by examining the content of high-stakes tests. We analysed all of the specimen assessment materials produced for the year 2004 by all the examination bodies in the UK for statistics courses taken as part of mathematics, by students aged 17 and 18 years. Not one question required students to work with 3 or more variables. Furthermore, no examples were found where the relationship between two variables was anything other than linear. (Ridgway, Nicholson & McCusker, 2006, in press). The result of this disjuncture is that we are not equipping citizens to be able to take part in social debates or make important decisions concerning life choices or well-being. It may be the case that reasoning with multivariate data is actually very difficult; certainly, students struggle to master relatively simple statistical concepts (Batanero, Godino, Vallecillos, Green, and Holmes, 1994). Here, we present evidence that, when supported with appropriate technology, students from even a young age are well able to work with multivariate data, and to draw informed conclusions from complex data. This has profound implications for the curriculum, and for assessment. It suggests that education may well be able to improve the quality of political debate, and the quality of the life choices made by individuals. Our research poses as many questions as it answers, and there is an urgent need to understand in detail the nature of reasoning from data, and to map its stages of development. Research Design A study was carried out with 102 students from a selective school in Northern Ireland and 92 students from a non-selective comprehensive school in the North-East of England. Students were aged between 12 and 15 years. A test of Reasoning with Data was constructed using paper-based and computer-based tasks across a broad range of difficulties. Paper based items comprised of a number of tasks from the Watson & Callingham (2003) studies of statistical literacy together with a new paper based task involving reasoning with multivariate data. The computer-based tasks were selected from a range of items from the World Class Arena (http://www.worldclassarena.org, Richardson, Baird, Ridgway, Ripley, Shorrocks- Taylor, & Swan, 2002).). which specifically focussed on reasoning from data. These items were designed originally to assess the problem solving skills of very able students. Test were administered by the students’ usual teachers in the presence of one of the authors and lasted approximately 70 minutes.