Burgan, Barry JohnMules, Trevor2010-09-012010-09-012000Events beyond 2000 : setting the agenda : proceedings of conference on event evaluation, research and education, July 2000 / John Allen, Robert Harris, Leo K Jago and A J Veal (eds.), pp.52-57186365562Xhttp://hdl.handle.net/2440/60188When undertaking an economic evaluation of a major event there are two key issues that need to be addressed to provide appropriate estimates of economic impact. Firstly, one must provide an estimate of how many people came to the event, including why they came, and where they came from. This area of questioning provides the basis for estimating how much tourism has increased as a consequence of the event. Secondly, information is needed on how much the attendees spend during their visit – leading to estimates of what they spend because of the event. The major methodological approach, particularly for ex-post studies, has been to collect information by survey. The next key issue is the process of factoring the information for the survey up to represent the population as a whole. This might seem a simple issue to resolve but it often proves harder than might be expected. Even where an organiser has an accurate measure of the attendances this does not represent the actual number of people who might attend – due to multiple or repeat attendances etc. It is made even more complicated when we recall that in the economic impact measure we are more interested in attendance by visitors attracted to the destination because of the event, than in overall attendance. This paper considers the implications of these issues within a random sampling approach to surveying. In the first instance we need an estimate of aggregate attendance. In some cases the event organisers have very accurate estimates (tickets issued, turnstile counters). However this paper considers some of the issues in taking the survey results and applying them to aggregate attendances, and suggests that there are some sampling frame issues involved. The second issue that arises is that in many cases, events encourage multiple attendances – and different groups (eg visitors versus local attendances) will have different patterns of attendance. In this case the paper demonstrates that in factoring survey information up to the population level, the model must include an adjustment to the proportions of attendance to allow for 'bias'. The paper provides the 'rules' by which this adjustment must be performed. Thirdly some more general issues that need consideration in developing the sampling frame and translating to the economic modelling are discussed. These include difficulties introduced by using techniques such as proportional sampling in order to minimise standard error estimates of overall expenditure, obtaining and using information from ticketing, and issues in the selection of interview sites.en(c) Copyright Australian Centre for Event Management, 2000Conference paper0001001290