Anthropometric characteristics of Japanese aged 0-59 years by analysis of a unified correlation matrix.

The correlation coefficient of each measured item pair for each partial group of similar-age, same-gender subjects is used to produce correlation coefficients for the entire subject group. Manipulation of the resulting equation yields a unified correlation matrix. Principal component analysis using this equation is applied to the anthropometric data of about 40,000 Japanese men and women 0-59 years old. As the distribution of subject age is not uniform, the number of subjects in each group is corrected for in the analysis, and the influence of this non-uniformity on the analytical results is investigated. An analysis for each individual partial group is also performed, and the physical features represented by the obtained principal factors are examined. Finally, the items crucially affecting each principal factor are shown to vary with age and gender.