Factors affecting accuracy and precision of thermal summation models of insect development used to estimate post-mortem intervals

This paper investigates the effects that different summary statistics (minimum, median, mean, or maximum), temporal sampling resolutions (duration between sampling events), and sample sizes (number of individuals sampled per sampling event) had on the accuracy and precision of the regression coefficients of a typical thermal summation model used to calculate minimum post-mortem interval (PMI). No significant differences were found in the values of the developmental constants calculated from different summary statistics of the duration of development. Sample size was found to affect the precision of measurement of the duration of development but had little overall influence on thermal summation constant (K) and developmental threshold (D (0)) calculations (and therefore, subsequent PMI estimates), but temporal sampling resolution had a direct influence on the accuracy of K and D (0) calculations. These data suggest that when numbers of experimental maggots are limited, it is more important to sample more frequently using smaller sample sizes than to sample less frequently with large sample sizes. Furthermore, we suggest that the median is the most representative summary measure of the duration of development and should be used preferentially.