By downloading this code you agree to the
Source Code Use License (PDF). |
{ version = 1.21; (* of multtest.p 2007 Mar 21} (* begin module describe.multtest *) (* name multtest: multiple Student's t-test synopsis multtest(distributions: in, multtestp: in, list: out, xyin: out, key: out, output: out) files distributions: parameters to control the program: A set of 3 numbers on one line defines a distribution: N (integer) mean (real) standard deviation (real) This file contains a series of such distributions, one per line. fudgefactor: If a line begins with 'f' then all standard deviations from that point on (until the next 'f') will be multiplied by the number following on the same line as f. This allows one to see the effects if the standard deviation does not account for all uncertainties. Lines that begin with '*' are comments. multtestp: a single character OR pairs of numbers, one per line, that represent the distributions to be compared by a ttest. The distributions listed are the only ones output to the list file. controlchar: If a line begins with a character then: l: compute the natural log of the mean and adjust the standard deviation accordingly (divide by the mean). c: set the color transform. The next character is the colortype: the colors produced in xyin are adjusted: n: normal colors Hue is set to the probability. i: interval colors. The 'i' must be followed by the number of intervals to truncate the colors to. Hue is set to the truncated probability. For example, if the number of intervals is 2, then the colors will be 0.0 to 0.5 and 0.5 to 1. s: standard interval colors. The colors will cut off at the points 0.50, 0.90 to 0.95 and 0.99. list: Input values, calculated Student's t value and probabilities that the numbers are the same, along with the Z score for the energy deviation from zero, and the probability that this deltadeltaG is zero. The pairs to be listed are defined in the multtestp file. xyin: input to xyplo program: column 1: (integer) first distribution number column 2: (integer) second distribution number column 3: (integer) hue fraction column 4: (integer) saturation fraction = 1 column 5: (integer) brightness fraction = 1 The hue is computed from the probability according to the colortype defined in multtestp. key: input to xyplo program, same colomuns but generates the key for the probabilities. output: messages to the user description This program performs T test computations for pairs of distributions and converts them to probabilities that the means are the same. This is a two-tailed Student's t test. examples For a multtestp file: 1 2 with a distributions file: 3 118 36 3 205 26 The resulting list file is: multtest 1.02 distribution 1 | distribution 2 number 3 | 3 mean 118.00000 | 205.00000 standard dev. 36.00000 | 26.00000 sigma-D = 25.63851 degrees of freedom = 4 t = 3.39333 p(1=2) = 0.02745 This is the probability that they are the same. documentation @book{Press1989, author = "W. H. Press and B. P. Flannery and S. A. Teukolsky and W. T. Vetterling", title = "Numerical Recipies in Pascal. The Art of Scientific Computing", publisher = "Cambridge University Press", address = "Cambridge", year = "1989"} Given a t value from a Student's t test, and the degrees of freedom, df, return the probability for a two tailed test. The code for doing this was originally in java script, from: Richard Lowry Department of Psychology Vassar College Poughkeepsie, NY 12604-0396 USA office: (914)437-7381 fax: (914)437-7538 lowry@vassar.edu http://faculty.vassar.edu/~lowry/VassarStats.html The original functional html containing this code is also given. It was translated to Pascal by Tom Schneider. see also The source program: ttest.p The xyin file can be graphed using: xyplo.p Results can be checked at: http://www.statsol.com/tools/stattools/ttestindependenttool.html http://faculty.vassar.edu/~lowry/VassarStats.html The method of conversion to a log scale comes: calc.p author Thomas Dana Schneider bugs technical notes *) (* end module describe.multtest *) {This manual page was created by makman 1.45}{created by htmlink 1.62}