Verification of the Accuracy of Sample-Size Equation Calculations for Visual Sample Plan Version 0.9C [report]

J R Davidson
2001 unpublished
Visual Sample Plan (VSP) is a software tool being developed to facilitate the design of environmental sampling plans using a site-map visual interface, standard equations for computing the number of samples (sample size) required for conducting statistical tests of hypotheses, a variety of sampling grids and random sampling plans, and graphs to visually depict the results to the user. The development of VSP has been ongoing for a number of years and several beta versions of the VSP code have
more » ... n available for download from the web by individuals who wished to use a preliminary and partially tested code. The VSP code has now been developed to a point where increased attention is being given to documenting that the calculations conducted by the code are accurate and contain no errors. This report documents the comparisons made in 1999 between the number of samples calculated by VSP Version 0.9C and the number of samples calculated by a computer code written in the S-PLUS  language. The results of this report indicate that the sample-size equations used by version 0.9C of VSP to determine the recommended minimum number of sampling locations are properly implemented for the particular VSP inputs used. However, evaluations of version 0.9H of the VSP code, which were conducted after the evaluations of version 0.9C reported here, indicated that a revision of the sample-size equations was needed for three of the nine statistical tests in VSP; namely, the two-sample t test, the onesample test of proportions, and the two-sample test of proportion. These evaluations of version 0.9H, which are reported in Gilbert et al (2001), indicated that the sample sizes computed for these three tests were in error for certain VSP input options not considered in the evaluations reported here for version 0.9C. These errors in VSP were corrected and will carry over to later versions. This report also documents that the VSP implementation of the ELIPGRID-PC algorithm for hot spot probabilities matches previous results for 100 standard test cases. The Conclusions and Limitations section of this document lists some aspects of VSP that were not tested by this suite of tests and recommends simulation-based enhancements for future versions of VSP.  S-PLUS is a registered trademark of MathSoft, Inc., Cambridge, Massachusetts. iv
doi:10.2172/786780 fatcat:igbjdngworba7dnn4bh5l7h2f4