● produced in 2008
● 7 new directives, 5 new functions and 26 new procedures
● factor analysis (
● easier specification of a wider range of graphics colours (
● ability to put a box around the graphics frame and control the background colours inside title bars and around graphics windows (
● plots of bitmaps (
● Cochran-Armitage test for trend (
● Dunnett’s test (
● Steel’s test (
● plots of estimates from regression and generalized linear models (
● Wald tests for terms in regression and generalized linear models (
● efficient fitting of HGLMs with large numbers of random effects (
● nonparametric estimates of species richness (
● printing of tables of textual strings (
● tables of standard errors of skewness and kurtosis (
● rejection sampling (
● QR decomposition (
● links to NAG Library algorithms (
2. What’s new
AFMINABERRATION forms minimum aberration factorial or fractional-factorial designs.
DBITMAP plots a bit map of RGB colours.
FARGUMENTS forms lists of arguments involved in an expression.
FCA performs factor analysis.
NAG calls an algorithm from the NAG Library.
QRD calculates QR decompositions of matrices.
RKESTIMATES saves estimates and other information about individual terms in a regression analysis.
ADETECTION calculates the minimum size of effect or contrast detectable in an analysis of variance.
AGFACTORIAL generates minimum aberration block or fractional factorial designs.
AMDUNNETT forms Dunnett’s simultaneous confidence interval around a control.
CATRENDTEST calculates the Cochran-Armitage chi-square test for trend.
CCA performs canonical correspondence analysis.
ECNPESTIMATE calculates nonparametric estimates of species richness.
FITMULTINOMIAL fits generalized linear models with multinomial distribution.
GETRGB gets the RGB values of the standard graphics colours.
GREJECTIONSAMPLE generates random samples using rejection sampling.
HGGRAPH draws a graph to display the fit of an HGLM or DHGLM analysis.
HGWALD prints or saves Wald tests for fixed terms in an HGLM.
PRSPEARMAN calculates probabilities for Spearman’s rank correlation statistic.
RDESTIMATES plots one- or two-way tables of regression estimates.
RTCOMPARISONS calculates comparison contrasts within a multi-way table of means.
RWALD calculates Wald and F tests for dropping terms from a regression.
SOM declares a self-organizing map.
SOMADJUST performs adjustments to the weights of a self-organizing map.
SOMDESCRIBE summarizes values of variables at nodes of a self-organizing map.
SOMESTIMATE estimates the weights for self-organizing maps.
SOMIDENTIFY allocates samples to nodes of a self-organizing map.
SOMPREDICT makes predictions using a self-organizing map.
STEEL performs Steel’s many-one rank test.
BLUE calculates the blue components of RGB colour values.
GREEN calculates the green components of RGB colour values.
RED calculates the red components of RGB colour values.
RGB calculates RGB colour values from red, green and blue components.
3. What’s changed
The representation of colours in graphics has been changed in Release 11 to allow you to use standard colour names. So virtually all options and parameters of the directives and library procedures that define colours will now take strings or texts as their settings. At the same time the use of colour numbers has been modified to allow you to define a colour directly by its RGB value, instead of requiring you first to redefine one of the 256 standard GenStat colours. Now instead of putting e.g.
COLOUR 3; RED=50; GREEN=205; BLUE=50
PEN 3; COLOUR=3
to define pen 3 as lime green, you can put either
PEN 3; COLOUR='limegreen'
or, if you still want to set individual RGB values, you can use the
PEN3; COLOUR=RGB(50; 205; 50)
For details of the new colours; see the description of the
This may cause incompatibilities in existing programs as, if you supply a number as the setting of a colour option or parameter, GenStat will now interpret it as an RGB value instead of as the number of a standard colour.
COLOUR directive thus need no longer be used, although it will be retained for use in old programs, together with the old standard colours numbers. If you have programs and procedures that rely on the old way of specifying the colours, you can use the
SET directive to request that number settings for colour options and parameters be interpreted in the old way rather than as RGB values. Also, the
GET directive can find out the current colour method (as, for example, inside a procedure), and
the RESTORE option of the
PROCEDURE directive can be set to restore the colour method automatically on exit from a procedure.
To convert programs, the colour names corresponding to the initial default settings of the old colours in Release 10 were as follows:
These colours were then recycled over colours 33-64, 65-96, 97-128, 129-160, 161-192, 193-224 and 225-256.
Most of the other changes are compatible with Release 10, the previous release. There are a few commands, however, where new options or parameters have been inserted into the existing lists. These may cause problems in statements where option or parameter names have been omitted or abbreviated (see Section 1.7.1 of Part 1 of the Guide to the GenStat Command Language for details). To avoid any difficulty, the name of the option/parameter after the new option/parameter should be given explicitly, and not abbreviated to fewer than four characters.
Any command, where changes in Release 11 may cause incompatibilities in existing programs, is marked in Sections 3.1 and 3.2 by the symbol †. The full details are given in Section 3.3.
ASSIGN can control the number that a dummy setting of the
STRUCTURE parameter is substituted before it is assigned (useful for procedure writers).
BARCHART now allows you to set the lower heights of the error bars to be different from the upper heights, and you can control the scale used to label the y-axis.
DGRAPH allows you to control the appearance of vertical or horizontal error bars, and to specify explicitly the order in which items are plotted.
FACROTATE can now save communalities.
FCLASSIFICATION now allows you to include only terms that can be dropped individually from the formula.
FRAME can now put a box around the whole frame, and control the background colours for the windows, their frames and title bars.
PEN now allows you to control whether missing x- or y-values cause a break when drawing splines through the points. You can also control the positioning of labels by the points, and set the thickness of error bars.
TABULATE can form tables of standard errors of skewness and kurtosis.
TXCONSTRUCT allows you to control how to represent missing values in numerical structures. Also the
LASTSEPARATOR option has been modified so that it takes its default from the setting of the
SEPARATOR option (the previous default was not what most users seemed to expect).
AMCOMPARISON can now produce mean-mean plots.
DBARCHART can now plot error bars.
DISCRIMINATE can produce canonical correlation coefficients.
DSCATTER can produce matrix plots of one set of variables against another set.
FRESTRICTEDSET can save a variate with each unit set to 1 or 0 according to whether or not that unit is restricted in any of the input vectors.
HGANALYSE allows you to specify a random term whose factor combinations should be used as a groups factor during the fitting of the augmented mean model; this allows models with large numbers of random effects to be fitted much more efficiently.
HGDISPLAY allows you to print information from the analysis of the dispersion parameter of a random term.
HGKEEP allows you to specify which method to use for the adjusted profile likelihood when calculating the likelihood statistics (as in
HGDISPLAY), and to save the numbers of fixed and random parameters in the mean and dispersion models.
MANNWHITNEY now compares each group against a control group (specified by a new
CONTROL option) when there are more than 2 groups.
RDA can now perform partial redundancy analysis, and also allows you to control the scaling of species and site scores.
SVTABULATE can now provide bootstrap quantiles and confidence.
NEWLEVELS gives ordinals (1, 2 …) if the second argument omitted (previously this was undefined).