Estimating the date of bone remains: a multivariate study

In previous works we have studied the time of death of bone residuals through the following parameters: total lipids, triglicerides , cholesterol, free fatty acids, total proteins, zinc, iron, manganese, and phosphorus. These elements were quantified in groups of recent bones of 1 and 2 years and of 10, 15, 18, and 20 years postmortem. In this present work we are putting these results under statistical analysis consisting of a stepwise regression. This program selects and introduces in the regression the element that shows the highest correlation with the time of death. In successive steps the partial correlations between the date and the elements not already included in the regression are studied, while keeping the effects of the elements already included fixed. As a result we put forward three formulas in which the time of death appears linked with the parameters that define it best. In the first the time of death of the bones Y is estimated according to the protein X1 Y = 40.0014 - 7. 4275X1 In the second formula the time of death Y, is estimated according to proteins X1 and triglicerides X2. Y = 45.5970 - 10. 8096X1 + 0. 4104X2 And in the third formula the time of death Y is estimated according to proteins X1, triglicerides X2, and cholesterol X3. Y = 52.2032 - 7. 8213X1 + 0. 6355X2 - 3.4930 In the three formulas the coefficients of the correlation between the time of death and the variables are improved when the logarithms of the variables are taken, instead of the original measurements.