With the spread of Information Technology, the importance of media education has increased. The development of Critical Viewing Skills (CVS), the ability to make judgments about the reliability, bias and accuracy of information, has been considered to be a crucial factor in media education (e.g., D. Buckingham, 2004; DCMS, 2001). It is especially important to evaluate the effectiveness of media education. On the other hand, the evaluation of CVS is a difficult problem (H. Ruminski, & W. Hanks, 1997). To respond to the above, the authors developed a measurement of CVS using category analysis (Y. Gotoh, 2007). To assess their CVS, children were asked to write comments about how to identify the reliability of web pages, by considering a commercial site related to dieting. We developed a category system for analysis. For example, ‘reliability of source of information’, ‘meta-information from the web’, ‘support information from media’ , ‘support information from persons’, ‘importance of multi-point’ and so on. Using this category system we could evaluate children’s CVS into four levels; inappropriate description or no answer, list of either media or information, list of both media and information, and list of suitable media, adequate sources and useful information. Our measurement system is useful, but it has only four categories. To clarify the effectiveness of media education, we need more sensitive and practical measurements. In this study, the authors try to solve this problem using multiple correspondence analysis. Multiple correspondence analysis is one of the multivariate analysis techniques. Using response data, we can analyze the following two points. Firstly, the distance of each category can be shown in two dimensions. It looks as though the distances between each category are not equally spaced. For example, ‘importance of multi-point’, ‘support information from multi-point media’ and ‘support information from multi-point persons’ seem to be similar. Using multiple correspondence analysis, we find that similar categories plot close together and dissimilar categories plot further apart, which allows us to understand intuitively the structure of the CVS. Secondly, distances between children can be shown in two dimensions. Differences in CVS levels would be shown as distances between children. Compared with the four-category system, this system is more sensitive to individual differences in CVS levels. In future, we also aim to develop a measurement for pre-post differentials using this technique. By comparing the pre-position and post-position of individuals, the effectiveness of media education could be shown clearly and objectively.

Questionnaires were given to 557 elementary school children, 377 junior high school students, 385 senior high school students and 387 university students. Homeroom teachers gave the questionnaires to their classes. Students were asked to comment on how to identify the reliability of information from commercial web pages. A ”comic” style was used in order to encourage students to write. Using a category system for analysis, children's and students' comments were converted to categorical data. Multiple correspondence analysis was applied. To identify the structure of CVS categories, we used category plotting based on multiple correspondence analysis. Relationships and distances between each category could be shown. To identify individual differences in CVS, we also used sample plotting. Individual differences in CVS were shown by distances between plotting data. SPSS 15.0J and SPSS categories were used as a statistical analysis package.

We divided CVS categories into four quadrants. X-axis is ‘Objective – Subjective’. Y-axis is ‘multiple source – single source’. Therefore, the 1st quadrant is objective and multiple source, the 2nd quadrant is subjective and multiple source, the 3rd quadrant is subjective and single source and the 4th quadrant is objective and single source. Categories of ‘importance of multi-point’ and ‘support information from multi-point persons’ are plotted in the 1st quadrant. 'Inappropriate description’ is plotted in the 3rd quadrant. Using multiple correspondence analysis, we can plot category and sample simultaneously. Individual differences in CVS can then be shown clearly. Children with high level CVS plotted in the 1st quadrant. The plotting position provides useful information to media educators. If many children plot in the 4th quadrant, media educators should teach the importance of multiple sources of information. The presentation will provide statistical analysis and some case studies.