### 1. Highlights

● produced in 2005

● 6 new directives, 28 new procedures and 16 new functions

● “formatted” output in html (for use e.g. in web pages), rtf (for wordprocessors) or LaTeX (for scientific publications)

● Cokriging (see `COKRIGE`

)

● Lin’s concordance coefficient (see `LCONCORDANCE`

)

● McNemar’s test (see `MCNEMAR`

)

● power calculations for analysis of variance, regression and generalized linear models (see `APOWER`

and `RPOWER`

)

● sample-size calculations for analysis of variance (`ASAMPLESIZE`

), t-tests (`STTEST`

), correlation coefficients (`SCORRELATION`

), sign tests (`SSIGNTEST`

), Mann-Whitney tests (`SMANNWHITNEY`

), binomial tests (`SBNTEST`

), Lin’s concordance coefficient (`SLCONCORDANCE`

), and McNemar’s test (`SMCNEMAR`

)

● sample-size calculations for tests for equivalence and tests for non-inferiority made by an analysis of variance (`ASAMPLESIZE`

) or a t-test (`STTEST`

)

● enhanced facilities for fitting proportional-hazards models, allowing sequences of models to be studied by adding or dropping terms (see `RPHFIT`

)

● generation of complete and quasi-complete Latin squares – used for experiments where treatment effects may carry over to adjacent plots (see `AGQLATIN`

)

● Nelder-Mead simplex algorithm for optimization – the target function may be calculated using expressions or by a GenStat procedure thus allowing any GenStat command to be involved (see `SIMPLEX`

)

● calculation of effective standard errors, for tables of means from an unbalanced analysis of variance, that give good approximations to the standard errors of differences (see `SED2ESE`

)

● algorithms for interactive identification e.g. of plants or animals (see `IDENTIFY`

)

● irredundant test sets, used e.g. for constructing efficient identification schemes (see `IRREDUNDANT`

)

### 2. What’s new

**2.1 Directives**

`CAPTION`

prints captions in standardized formats.

`COKRIGE`

calculates kriged estimates using a model fitted to the sample variograms and cross variograms of a set of variates.

`FCOVARIOGRAM`

forms a covariogram structure containing auto-variograms of individual variates and cross-variograms for pairs from a list of variates.

`IRREDUNDANT`

forms irredundant test sets for the efficient identification of a set of objects.

`MCOVARIOGRAM`

fits models to sets of variograms and cross variograms.

`REFORMULATE`

modifies a formula or an expression to operate on a different set of data structures.

**2.2 Procedures**

`AGQLATIN`

generates complete and quasi-complete Latin squares.

`APOWER`

calculates the power (probability of detection) for terms in an analysis of variance.

`ASAMPLESIZE`

finds the replication to detect a treatment effect or contrast.

`FDRMIXTURE`

estimates false discovery rates using mixture distributions.

`FVCOVARIANCE`

forms the variance-covariance matrix for a list of variates.

`IDENTIFY`

identifies an unknown specimen from a defined set of objects.

`LCONCORDANCE`

calculates Lin’s concordance correlation coefficient.

`MCNEMAR`

performs McNemar’s test for the significance of changes.

`NCONVERT`

converts integers between base 10 and other bases.

`PRCORRELATION`

calculates probabilities for product moment correlations.

`RMGLM`

fits a model where different units follow different generalized linear models.

`RPHCHANGE`

modifies a proportional hazards model fitted by `RPHFIT`

.

`RPHDISPLAY`

prints output for a proportional hazards model fitted by `RPHFIT`

.

`RPHFIT`

fits the proportional hazards model to survival data as a generalized linear model.

`RPHKEEP`

saves information from a proportional hazards model fitted by `RPHFIT`

.

`RPHVECTORS`

forms vectors for fitting proportional hazards data as a generalized linear model.

`RPOWER`

calculates the power (probability of detection) for regression models.

`SBNTEST`

calculates the sample size for binomial tests.

`SCORRELATION`

calculates the sample size to detect specified correlations.

`SED2ESE`

calculates effective standard errors that give good approximate standard errors of differences.

`SIMPLEX`

searches for the minimum of a function using the Nelder-Mead simplex algorithm.

`SLCONCORDANCE`

calculates the sample size for Lin’s concordance coefficient.

`SMANNWHITNEY`

calculates sample sizes for the Mann-Whitney test.

`SMCNEMAR`

calculates sample sizes for McNemar’s test.

`SPRECISION`

calculates the sample size to obtain a specified precision.

`SSIGNTEST`

calculates the sample size for a sign test.

`STTEST`

calculates the sample size for t-tests (including equivalence tests).

`VMATRIX`

copies values and row/column labels from a matrix to variates or texts.

**2.3 Functions**

`BJ0` |
Bessel function J_{0}. |
---|---|

`BJ1` |
Bessel function J_{1}. |

`BI0` |
modified Bessel function I_{0}. |

`BI1` |
modified Bessel function I_{1}. |

`MCENTRE` |
doubly centres a matrix so that its rows and columns have mean zero. |

`MEXP` |
matrix exponential. |

`MPOWER ` |
matrix power. |

`MSQRT` |
matrix square root. |

`RANGE` |
range of values. |

`STANDARDIZE ` |
standardizes the values. |

`TCOLUMN` |
converts a one-way table into a column matrix. |

`TDIAGONAL` |
converts a one-way table into a diagonal matrix. |

`TMATRIX` |
converts a two-way table into a matrix. |

`TPROJECT` |
converts a table into a variate. |

`TROW` |
converts a one-way table into a row matrix. |

`VRANGE` |
range of values across a pointer of variates. |

### 3. What’s changed

Most of the changes are completely compatible with Release 7, 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 8 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.

**3.1 Directives**

^{†}`AKEEP`

can now save effective degrees of freedom for combined means and effects, and codes indicating the properties of the design and the model terms in the analysis.

`ANOVA`

can now save a codes indicating the properties of the design.

`CALCULATE`

has a new `INDEX`

option, which is useful if the calculation has a list of structures before the assignment operator (`=`

). The option is set to a scalar which will indicate the position within the list of the structure currently being evaluated.

`ENQUIRE`

can save details of the output styles currently in use.

`FPSEUDOFACTORS`

can save the resolution number of the design.

`KRIGE`

has a new `SAVE`

parameter too allow model information to be supplied more conveniently from `MVARIOGRAM`

.

`OPEN`

allows you to set the style of an output channel (as either plain-text, html, rtf or LaTeX).

`OUTPUT`

allows you to set the style (plain-text or formatted) for future output to an output channel.

^{†}`PEN`

has a new option setting `METHOD=spline`

to plot a smoothing spline fitted through the points; the new `DFSPLINE`

sets the number of degrees of freedom.

`PRINT`

has a new parameter `HEADING`

which allows you to specify column headings to be used for vectors instead of their identifiers, and a new option `STYLE`

to request use of plain-text when printing to output files opened in html, rtf or LaTeX styles.

`PROCEDURE`

has an extra setting, `outstyle`

, for its `RESTORE`

option to allow you to restore the style of output.

^{†}`RKEEP`

can now save standard errors for the linear predictor.

`SKIP`

allows you to specify whether to use plain-text or formatted styles when skipping lines in output.

`TERMS`

has a new option `MVINCLUDE`

to specify that units with missing values in an explanatory variable should not be omitted; instead the variable should make no contribution to the fitted values for these units.

**3.2 Procedures**

`AUNBALANCED`

and `AUDISPLAY`

can now print approximate effective standard errors of means, and `AUKEEP`

can save them.

^{†}`DISCRIMINATE`

– new version with enhanced output including plots.

`FSTRING`

has an extra parameter to specify characters to go at the end of the string.

`FTEXT`

can now generate a text from any numerical structure, and has additional parameters to control the display of factor levels and missing values.

`KAPLANMEIER`

can now produce means and quantiles.

`MVARIOGRAM`

has a new `SAVE`

parameter too allow model information to be supplied more conveniently to `KRIGE`

.

^{†}`PRMANNWHITNEYU`

can calculate the exact probabilities when the data include tied values in many GenStat implementations (including MS Windows).

`QLIST`

has a new option `HELP`

to provide help information for the question.

^{†}`RSURVIVAL`

now also covers the log-logistic and lognormal distributions, and can save their shape parameter sigma as well as the estimates of the survivor function.

`SPCAPABILITY`

can produce histograms.

^{†}`SVSTRATIFIED`

can produce ratio estimates when `Y`

contains only the response data.

`TALLY`

can use weights in its tabulations, and it can include quantiles in the graphs.

`VTABLE`

can generate labels indicating the factor values corresponding to each unit of the generated variate.

In addition, the Library procedures have been modified where possible to use the new `CAPTION`

directive and the new `HEADING`

parameter and `STYLE`

option of `PRINT`

, so that their output is displayed correctly in GenStat’s new formatted output styles (see the `OUTPUT`

directive). The procedures that cannot generate formatted output correctly have been modified to switch the output style to plain text by the command

`OUTPUT [STYLE=plaintext]`

and then restore the original style by including the setting `outstyle`

in the settings of the `RESTORE`

option in the `PROCEDURE`

command.

**3.3 Incompatibilities**

`AKEEP` directive |
option `EXIT` inserted before `SAVE` ; parameter `DFCMEANS` inserted before `CBEFFECTS` , and parameter `DFCEFFECTS` inserted before `CBCEFFICIENCY` . |
---|---|

`AUKEEP` procedure |
parameter `ESEMEANS` inserted before `LSD` . |

`DISCRIMINATE` procedure |
new parameter `ALLOCATION` inserted before `MEANS` . |

`PEN` directive |
new parameter `DFSPLINE` inserted before `SAVE` . |

`PRMANNWHITNEYU` procedure |
new parameter `TIES` inserted before `CLPROBABILITY` . |

`REPLICATION` procedure |
option `MINREPLICATION` inserted before `MAXREPLICATION` , and `TMETHOD` inserted before `SAVE` . Note, however, that this procedure is superceded by `ASAMPLESIZE` ; it will be phased out in a future release. |

`RKEEP` directive |
parameter `SELINEARPREDICTOR` inserted before `INFLATION` . |

`RSURVIVAL` procedure |
options `SIGMA` , `SURVIVOR` and `PARAMETERIZATION` inserted before `MAXCYCLE` . |

`SVSTRATIFIED` procedure |
parameter `XTOTALS` inserted before `TOTALS` . |