Multivariate statistics applied to the evaluation of environmental and chemical data sources

Constantly expanding chemical and environmental information sources increas e the need for descriptive statistical analysis. This paper gives a compara tive evaluation of data sources, i.e. online databases, databases on CD-ROM and Internet resources in the field of environmental chemicals. The evalua tion is based on information in three metadatabases for environmental chemi cals: DADB-Metadatabase of Online Databases, DACD-Metadatabase of CD-ROMs, DAIN-Metadatabase of Internet Resources. A data matrix of 50 environmental and chemical descriptors found in DADB, DACD and DAIN is analysed and a tec hnique is applied to transform the data set into a data matrix of a more ho mogeneous structure. This method is based on algorithms for solving the so- called travelling salesman problem. Two different ways of analysing the dat a set are applied and the results are compared. Also, media combination pat terns are identified and discussed. For most descriptors the information de pth is higher in commercial online databases and databases on CD-ROM than i n free Internet resources. Exceptions, e.g. some health-related parameters which have a higher percentage in Internet resources, are identified and ex plained.