Abstract
In order to carry out new discovery in any study, a new tool and technology are effective. The purpose of this paper is to propose the method of a new data analysis about achievement tests. If the correlation between the problems used for an investigation is metrically analyzed from various angles, a new understanding about the correlation of the investigation domain will be obtained. By conducting such analysis, the useful fundamental data for improving a curriculum and lessons based on results of the investigation can be obtained. By considering based on these data, it can reach to a concrete proposal. The OECD is performing international achievement tests (PISA) from 2000 every three years. The results of investigations of PISA raised various problems about school education of Japan, especially science and mathematics education. In order to profit this investigation in a curriculum or a lesson improvement, it is required to analyze the investigation data from a new viewpoint and to connect it to a concrete proposal. The method of the data analysis proposed in this paper is called "the network type data analysis" in order to put the foundation on the consideration about a network. The vertex set of this network is the whole investigation problems. The two arbitrary vertices are connected by an edge with the weight given by the mutual information between corresponding problems. The mutual information expresses the interdependence between two random variables metrically.
