Analysis/plot generation code with significance levels computed using Kolmogorov-Smirnov statistics valid for both large and small samples [electronic resource].
- Oak Ridge, Tenn. : Oak Ridge National Laboratory, 1983.
Oak Ridge, Tenn. : Distributed by the Office of Scientific and Technical Information, U.S. Dept. of Energy.
- Additional Creators:
- Oak Ridge National Laboratory
United States. Department of Energy. Office of Scientific and Technical Information
- This report describes a version of the TERPED/P computer code that is very useful for small data sets. A new algorithm for determining the Kolmogorov-Smirnov (KS) statistics is used to extend program applicability. The TERPED/P code facilitates the analysis of experimental data and assists the user in determining its probability distribution function. Graphical and numerical tests are performed interactively in accordance with the user's assumption of normally or log-normally distributed data. Statistical analysis options include computation of the chi-square statistic and the KS one-sample test statistic and the corresponding significance levels. Cumulative probability plots of the user's data are generated either via a local graphics terminal, a local line printer or character-oriented terminal, or a remote high-resolution graphics device such as the FR80 film plotter or the Calcomp paper plotter. Several useful computer methodologies suffer from limitations of their implementations of the KS nonparametric test. This test is one of the more powerful analysis tools for examining the validity of an assumption about the probability distribution of a set of data. KS algorithms are found in other analysis codes, including the Statistical Analysis Subroutine (SAS) package and earlier versions of TERPED. The inability of these algorithms to generate significance levels for sample sizes less than 50 has limited their usefulness. The release of the TERPED code described herein contains algorithms to allow computation of the KS statistic and significance level for data sets of, if the user wishes, as few as three points. Values computed for the KS statistic are within 3% of the correct value for all data set sizes.
- Published through SciTech Connect.
Fields, D.E.; Kurtz, S.E.
- Funding Information:
View MARC record | catkey: 14359939