The PLM Procedure

PROC PLM Statement

PROC PLM RESTORE=item-store-specification <options> ;

The PROC PLM statement invokes the PLM procedure. The RESTORE= option with item-store-specification is required. Table 73.2 summarizes the options available in the PROC PLM statement.

Table 73.2: PROC PLM Statement Options

Option

Description

Basic Options

RESTORE=

Specifies the source item store for processing

SEED=

Specifies the random number seed

STMTORDER=

Affects the order in which statements are grouped during processing

FORMAT=

Specifies how the PLM procedure handles user-defined formats

WHEREFORMAT

Specifies the constants (literals) in terms of the formatted values of the BY variables

Computational Options

ALPHA=

Specifies the nominal significance level

DDFMETHOD=

Specifies the method for determining denominator degrees of freedom

PERCENTILES=

Supplies a list of percentiles for the construction of HPD intervals

Displayed Output

MAXLEN=

Determines the maximum length of informational strings

NOCLPRINT

Suppresses the display of the Class Level Information table

NOINFO

Suppresses the display of the Store Information table

NOPRINT

Suppresses tabular and graphical output

PLOT

Controls the plots produced through ODS Graphics

Singularity Tolerances

ESTEPS=

Specifies the tolerance value used in determining the estimability of linear functions

SINGCHOL=

Tunes the singularity criterion in Cholesky decompositions

SINGRES=

Sets the tolerance for which the residual variance or scale parameter is considered to be zero

SINGULAR=

Tunes the general singularity criterion

ZETA=

Tunes the sensitivity in forming Type III functions


You can specify the following options:

ALPHA=$\alpha $

specifies the nominal significance level for multiplicity corrections and for the construction of confidence intervals. The value of $\alpha $ must be between 0 and 1. The default is the value specified in the source item store, or 0.05 if the item store does not provide a value. The confidence level based on $\alpha $ is $1-\alpha $.

DDFMETHOD=RESIDUAL | RES | ERROR
DDFMETHOD=NONE
DDFMETHOD=KENROG | KR | KENWARDROGER
DDFMETHOD=SATTERTH | SAT | SATTERTHWAITE

specifies the method for determining denominator degrees of freedom for tests and confidence intervals. The default degree-of-freedom method is determined by the contents of the item store. You can override the default to some extent with the DDFMETHOD= option.

If you choose DDFMETHOD=NONE, then infinite denominator degrees of freedom are assumed for tests and confidence intervals. This essentially produces z tests and intervals instead of t tests and intervals and chi-square tests instead of F tests.

The KENWARDROGER and SATTERTHWAITE methods require that the source item store contain information about these methods. This information is currently available for item stores that were created with the MIXED or GLIMMIX procedures when the appropriate DDFM= option was in effect.

ESTEPS=$\epsilon $

specifies the tolerance value used in determining the estimability of linear functions. The default value is determined by the contents of the source item store; it is usually 1E–4.

FORMAT=NOLOAD | RELOAD

specifies how the PLM procedure handles user-defined formats, which are not permanent. When the item store is created, user-defined formats are stored. When the PLM procedure opens an item store, it uses this option as follows. If FORMAT=RELOAD (the default), the stored formats are loaded again from the item store and formats that already exist in your SAS session are replaced by the reloaded formats. If FORMAT=NOLOAD, stored formats are not loaded from the item store and existing formats are not replaced.

With FORMAT=NOLOAD, you prevent the PLM procedure from reloading the format from the item store. As a consequence, PLM statements might fail if a format was present at the item store creation and is not available in your SAS session. Also, if you modify the format that was used in the item store creation and use FORMAT=NOLOAD, you might obtain unexpected results because levels of classification variables are remapped.

The Class Level Information table always displays the formatted values of classification variables that were used in fitting the model, regardless of the FORMAT= option. For more details about using formats with the PLM procedure, see User-Defined Formats and the PLM Procedure.

MAXLEN=n

determines the maximum length of informational strings in the Store Information table. This table displays, for example, lists of classification or BY variables and lists of model effects. The value of n determines the truncation length for these strings. The minimum and maximum values for n are 20 and 256, respectively. The default is n = 100.

NOCLPRINT<=number>

suppresses the display of the Class Level Information table if you do not specify number. If you specify number, only levels with totals that are less than number are listed in the table. The PLM procedure produces the Class Level Information table by default when the model contains effects that depend on classification variables.

NOINFO

suppresses the display of the Store Information table.

NOPRINT

suppresses the generation of tabular and graphical output. When the NOPRINT option is in effect, ODS tables are also not produced.

PERCENTILES=value-list
PERCENTILE=value-list

supplies a list of percentiles for the construction of highest posterior density (HPD) intervals when the PLM procedure performs a sampling-based analysis (for example, when processing an item store that contains posterior parameter estimates from a Bayesian analysis). The default set of percentiles depends on the contents of the source item store; it is typically PERCENTILES=25, 50, 75. The entries in value-list must be strictly between 0 and 100.

PLOTS <(global-plot-option)> <=specific-plot-options>

controls the plots produced through ODS Graphics. ODS Graphics must be enabled before plots can be requested. For example:

ods graphics on;

proc plm plots=all;
   lsmeans a/diff;
run;

ods graphics off;

For more information about enabling and disabling ODS Graphics, see the section Enabling and Disabling ODS Graphics in Chapter 21: Statistical Graphics Using ODS.

Global Plot Option

The following global-plot-option applies to all plots produced by PROC PLM.

UNPACKPANEL
UNPACK

suppresses paneling. (By default, multiple plots can appear in some output panels.) Specify UNPACK to display each plot separately.

Specific Plot Options

You can specify the following specific-plot-options:

ALL

requests that all the appropriate plots be produced.

NONE

suppresses all plots.

SEED=number

specifies the random number seed for analyses that depend on a random number stream. You can also specify the random number seed through some PLM statements (for example, through the SEED= options in the ESTIMATE, LSMEANS, and LSMESTIMATE statements). However, note that there is only a single random number stream per procedure run. Specifying the SEED= option in the PROC PLM statement initializes the stream for all subsequent statements. If you do not specify a random number seed, the source item store might supply one for you. If a seed is in effect when the PLM procedure opens the source store, the Store Information table displays its value.

If the random number seed is less than or equal to zero, the seed is generated from reading the time of day from the computer clock and a log message indicates the chosen seed value.

SINGCHOL=number

tunes the singularity criterion in Cholesky decompositions. The default value depends on the contents of the source item store. The default value is typically 1E4 times the machine epsilon; this product is approximately 1E–12 on most computers.

SINGRES=number

sets the tolerance for which the residual variance or scale parameter is considered to be zero. The default value depends on the contents of the source item store. The default value is typically 1E4 times the machine epsilon; this product is approximately 1E–12 on most computers.

SINGULAR=number

tunes the general singularity criterion applied by the PLM procedure in divisions and inversions. The default value used by the PLM procedure depends on the contents of the item store. The default value is typically 1E4 times the machine epsilon; this product is approximately 1E–12 on most computers.

RESTORE=item-store-specification

specifies the source item store for processing. This option is required because, in contrast to SAS data sets, there is no default item store. An item-store-specification consists of a one- or two-level name as with SAS data sets. As with data sets, the default library association of an item store is with the WORK library, and any stores created in this library are deleted when the SAS session concludes.

STMTORDER=SYNTAX | GROUP
STMT=SYNTAX | GROUP

affects the order in which statements are grouped during processing. The default behavior depends on the contents of the source item store and can be modified with the STMTORDER= option. If STMTORDER=SYNTAX is in effect, the statements are processed in the order in which they appear. Note that this precludes the hierarchical grouping of ODS objects. If STMTORDER=GROUP is in effect, the statements are processed in groups and in the following order: SHOW, TEST, LSMEANS, SLICE, LSMESTIMATE, ESTIMATE, SCORE, EFFECTPLOT, and CODE.

WHEREFORMAT

specifies that the constants (literals) specified in WHERE expressions for group selection are in terms of the formatted values of the BY variables. By default, WHERE expressions are specified in terms of the unformatted (raw) values of the BY variables, as in the SAS DATA step.

ZETA=number

tunes the sensitivity in forming Type III functions. Any element in the estimable function basis with an absolute value less than number is set to 0. The default depends on the contents of the source item store; it usually is 1E–8.