STAT-PACK(ID:6790/sta037)Goddard biostats packageBiostatistical programming package developed by B. S. Shannon at the Goddard Computer Science Institute, Dallas, Texas written in a subset of FORTRAN II References: Contains 56 statistical programs written m a subset of FORTRAN II; it was designed to run on almost any machine with 40K characters or 6K words of storage with a minimum of modification. The programs run without modification on an IBM 1620, 1130, and an XDS Sigma 7, but they can be modified easily for other systems in the same class or larger. The programs are designed for card input and printer output, but flexible input can be achieved by means of a user'supplied read subroutine. The section on regression is quite complete, including programs for linear and polynomial regression, with confidence intervals and analysis of individual terms in polynomial regression, as well as the usual multiple-linear and stepwise regression calculations. Other regression programs include pooling of groups and multiple dependent- variable linear regression, as well as a general nonlinear regression routine that accepts the nonlinear function to be fitted by means of a subroutine. The section on analysis of variance is fairly limited due to the core size restriction, but it does include some standard designs The section on tests of hypotheses includes routines for Duncan's multiple range test, the paired test, Bartlett's test, and the Kohnogorov- Smirnov and chi-square goodness of fit tests The other sections are fairly standard. The outputs of all programs are extensively labeled for the user. The programs can be used by a nonprogrammer if his data fit a standard form; otherwise, a programmer must write a read subroutine. in [ACM] ACM Computing Surveys (CSUR) 4(2) June 1972 view details |