1987 Volume 34 Issue 10 Pages 647-653
A statistical method for the yearly trend of seasonality of food consumption was developed here. Seasonal concentration and mean consumption time were employed for two indexes of seasonality. Seasonal concentration and mean consumption time in a year were defined by length and angle of the mean vector, respectively, where the mean vector was taken over 12 monthly consumption vectors in the year. The trend of seasonal concentration was analyzed by the linear regression and then by maximum marginal likelihood method for smoothing serial data. The trend of mean consumption time was analyzed by the linear regression of angular statistics. This method was applied to the data of processed food consumption, which were believed to have little seasonality of consumption. The linear regression of seasonal concentration in years showed that the 27 species among 66 of processed foods were under the trend of year-round consumption and 14 species were under the trend of season consumption. For 24 species, the use of the smoothed curve given by the maximum marginal likelihood method for smoothing serial data was recommended for obtaining the trend of seasonal consumption rather than that of the linear regression line. Mean consumption time were shown to shift significantly for 34 species, just more than half the number of the data sets in this study. For only 9 species, any changes could not be detected in both seasonal concentration and mean consumption time.