Perception of Statistical Presentations - Investigated by Means of Internet Experiments

A main responsibility of a national statistical office is to disseminate statistical information about the state and development of the country to its citizens. In statistical presentation, it has been assumed that statistical tables are particularly useful and required for precise analysis and research, charts are suitable for fast visual orientation of the main trends within the area on which information is requested, and that textual or vocal presentations are best suited for mass media presentation, such as in newspapers, television, or Internet news services. To date, there have been few studies of how the general public recipient perceives information in these various formats. The primary objective has been to investigate how these three presentation forms for statistical facts are perceived and interpreted by general public readers of statistical information. A second purpose has been to study the utilization of a Web based experiment built on statistical design principles. 1. Project justification A national statistical office fulfills its primary mission to disseminate statistical information about the state and development of its country through response to requests for statistical publications, tabulations, or through general information distribution in such mass media as newspapers, radio, television, or Internet publication. The information disseminated may be in the form of text, table, chart, vocal, and/or even animated presentations depending on the topic and the public to which the presentation is aimed. It has been assumed that statistical tables are particularly useful for precise analysis and research, charts are suitable for fast visual orientation of the trends and/or highlights within a topic area, while textual and vocal types are suited for newspapers and broadcasting, respectively [Tufte, 1990]. While it is generally assumed that graphical presentation improves comprehension, studies have shown that graphic displays, showing relationships between objects, are not easily understood. A study of the errors made by young students (12-14 years), asked to interpret Cartisean graphs, found that half of the error types were caused by graphic presentation characteristics, such as: domain specification, language wording, graph appearance, graph concepts, and/or graphic syntax [Preece, 1983]. Only 21% of young adults (21-25 years) were found to be proficient at locating and using information contained in maps, tables, or charts [Gillespie, 1993]. A comparison between textual and graphical presentations of three-level nested program structures concluded that graphic presentations were not necessarily more accessible, comprehensible, or memorable than textual presentations. Particularly, novice readers have problems locating and focusing on essential parts of the graph [Petre, 1995]. A comparison of graphic models of different styles found that graphic syntax affected model comprehension [Nordbotten, 1999]. These studies indicate that comprehension of graphic models is a skill that requires learning. Considerable resources are spent to disseminate statistics from the national statistical offices to inform the citizens about their environment and prepare them for more informed decisions. What are the best ways to present information on social facts for the public? Can the dissemination of statistical information to the public be improved and how? One objective of this project is to investigate how different types of presentations of statistical facts are perceived. The scenario is the general public receiving statistical news by different news media such as newspapers, radio, television, or Internet information providers. These passive ‘statistical users’ do not request any particular information and may have limited interest in the statistical 0-7695-0981-9/01 $10.00 (c) 2001 IEEE 1