Allelic frequencies of 13 STR loci in autochthonous Basques from the province of Vizcaya (Spain).

Allelic frequencies of 13 STR loci (D3S1358, VWA, FGA, D8S1179, D21S11, D18S51, D5S818, D13S317, D16S539, TH01, TPOX, CSF1PO, and D7S820) were estimated from a sample of 73 unrelated healthy donors natives of the Spanish Basque province of Vizcaya. These STR loci constitute the core of polymerase chain reaction (PCR)-based DNA genetic markers in the US Combined DNA Index System (CODIS). All STR loci analysed met Hardy-Weinberg expectations. Based upon the allelic frequencies, forensically important parameters including gene diversity (GD), polymorphism information content (PIC) and power of discrimination (PD) were calculated.

Biases in summary statistics of slopes and intercepts in linear regression with errors in both variables.

Monte Carlo simulations are employed to investigate the bias in linear regression parameters for cases in which both variables are subject to normally distributed errors. Both homoscedastic and heteroscedastic errors are treated. The results show that, in general, the arithmetic mean, geometric mean, and angle mean (tangent of the mean of the arctangents) of the slopes are biased and non-normally distributed. The arithmetic and geometric means of the intercepts are generally biased and non-normally distributed as well. However, for all the cases considered, the medians of the slopes and of the intercepts are found to be unbiased. In view of the non-normal distributions of the slopes and intercepts, a nonparametric method is used to determine confidence intervals for the slopes and intercepts.

Limiting efficiencies of the non-parametric median method for A exp(-kt) data.

Weighted least squares analysis and the median method are used to extract A and k from simulated single exponential decays. The standard deviations of these parameters (sigma p) depend upon the method of analysis. The sigma p are compared as efficiencies (= sigma 2p[least squares]/sigma 2p[median]). These efficiencies depend upon the range of the decay and the underlying error structure. These include errors independent of the data, approximately equal to the data and approximately equal to the square root of the data. Some efficiencies fall well below 25% for the median method. Such instances reflect the large cost of not knowing or not using the weights; they amount to ignoring greater than 75% of the data we strive to collect.