Professor Terence Hull is the John C. Caldwell Professor of Population, Health & Development at the Australian Demographic & Social Research Institute

This seminar explores some of the surprises and pitfalls that occur between the interviewers' visits to the homes of census respondents, and the delivery of demographic reports and calculations to the policy makers in the capital. In particular we will review the scanning, processing and tabulation of information using innovative technologies and the way they have given rise to some very puzzling results. First there is the case of direct estimation of maternal mortality with direct questions about deaths in the household. Second is the variety of estimates of fertility that are produced from three distinct sets of questions, and particularly the assumptions implanted in regression formulae from potentially unreliable information about marriage in a multicultural country. Finally, the debates that arose concerning the appropriate tabulations to use to display coded information on over 1,300 ethnic groups and 1,300 languages. Ultimately the Indonesian statisticians' problem is that the census collected the information from 237 million unique stories, but the calculations and tables create pictures of large entities -- nation and provinces, urban and rural, male and female -- with few of the nuances and shades that are inherent in the data. Without an understanding of the weaknesses and errors of the census data the true strength of the results and interpretations cannot be appreciated.