METHODS: In regression analysis, the following ten independent variables were taken from thirty-one Japanese males (40.0+/-11.9y.o.) living in rural areas. : Bone strength measured by ultra-sound method (SOS), %body fat (FAT), waist-hip ratio (WHR), trunk flexion (FLEX), grip strength (GRIP), sit-ups in 30sec. (SITUP), maximal oxygen intake (MAP), foot balance with eye closed (BALANCE), and two measures of blood pressure (SBP, DBP).
RESULTS: The following equation was selected as the optimal one according to Akaike's information criterion (AIC).: y=90.0 - 1.294*SITUP - 0.467*MAP - 0.094*BALANCE (R=0.837). When all ten variables were included in the initial equation, R was 0.864, and the highest R* (R adjusted for d.f.) was observed at the 5th step (squared R* is 0.682).
DISCUSSION: To obtain an optimal equation with as few independent variables as possible, but with a higher R, different statistical information criteria were applied in regression analysis. Adding to AIC or R*, the different criteria were proposed for variable selections in the regression model: Mallow's Cp, Schwarz Information Criterion (SIC).
The examination of the residuals of the fitted model is a standard
tool for checking whether or not the underlying model assumptions
hold and detecting the deficiencies of regressive model. In order to
ensure that the assumption of least squares theory is in order, a
careful examination of residuals is necessary. SD of residuals was
6.75, and kurtosis and skewness of its distribution were -0.37 and
-0.299, respectively. In plotting standardized residuals as ordinate
against the predicted values, the specific relationships were not
observed.
Key words:
Regression Analysis, Prediction of Age, Fitness Test, Japanese
Problems or questions regarding this site should be directed to
the organiser