Consider the following scenario: Suppose that the base has been adjusted by the base . Occasionally we will be concerned with partial adjustments, and to this purpose for the partial adjustment results we suppose that we adjust initially by the base baseE, and then partially by base ; and that represents the complete adjustment . Suppose that we might observe , , and so too (otherwise, data-related results - and thus these options - are not available).
Usage
BD>option a
switches on standard observed adjusted
expectation output.
BD>option a+
switches on extended observed adjusted
expectation output.
BD>option -a
switches off observed adjusted
expectation output.
BD>show (a )
displays standard observed adjusted
expectation output for the current adjustment.
BD>show (a+ )
displays extended observed adjusted
expectation output for the current adjustment.
The a and a+ options are used for the output of observed
adjustments, delivering standard and extended output respectively. The
extended output shown depends upon whether or not results are (or will
be) available for both an adjustment and a partial adjustment.
and the standardised value of
the change: . (See [48][sections 6.2.3,
13.2.3].)
The b option is used to display output relating to the size and
bearing for the adjustment, and for the partial adjustment if any. (See
[31][sections 4.4, 5.4].) Similar output is available using
the b* option.
Both the bearing and the partial bearing may be
retained interactively as assignments by using respectively the
b and pb arguments to the KEEP: command.
The output shown consists of one or
more columns, each representing a linear combination of the
elements of :
Beneath each column are summary quantities for the observed adjustment:
the size of the adjustment, the expected size of the adjustment, and the
size ratio respectively. Hence the values reported are (assuming that
the partial results are available, otherwise only the first column is
reported):
The b* option is used to display output relating to the size and
standardised bearing for the adjustment, and for the partial adjustment
if any. (See [31][sections 4.4, 5.4].) Similar output is
available using the b option.
Both the bearing and the partial bearing may be
retained interactively as assignments by using respectively the
b and pb arguments to the KEEP: command, but
they cannot be retained in standardised form.
The output shown consists of one
or more columns, each representing a linear combination of the
standardised elements of , where each element is
standardised to have expectation zero and prior variance unity:
Beneath each column are summary quantities for the observed adjustment:
the size of the adjustment, the expected size of the adjustment, and the
size ratio respectively. Hence the values reported are (assuming that
the partial results are available, otherwise only the first column is
reported):
BD>option becorr
switches on bearing-element
correlation output.
BD>option -becorr
switches off bearing-element
correlation output.
BD>show (becorr )
displays
bearing-element correlation output for the current adjustment.
The becorr option yields the prior correlation between the
bearing for the current adjustment, , and every element
. If a partial adjustment has been made, then the prior
correlations for the bearings for the previous and partial adjustments,
and , with every element are
given.
BD>option bcorr
switches on bearing
inter-correlation output.
BD>option -bcorr
switches off bearing
inter-correlation output.
BD>show (bcorr )
displays
bearing inter-correlation output for the current adjustment.
When the results of a partial adjustment are available, the
bcorr option yields the prior correlations between the three
kinds of bearing: the bearing for the current adjustment, ;
the previous adjustment, ; and the partial adjustment,
. One of these three correlations (depending upon the type
of fit) is the path correlation, also available using the pc
option. In the case of model fitting, the path correlation is the prior
correlation between the bearings for the previous and partial
adjustments, and in the case of model reduction, it is the prior
correlation between the bearings for the current and partial
adjustments.
The dv option is used to display output relating to the
adjustment
discrepancy and
discrepancy vector for the adjustment, and for the partial adjustment if
any. Similar output is
available using the dv* option.
Both the discrepancy vector and the partial discrepancy vector may be
retained interactively as assignments by using respectively the
dv and pdv arguments to the KEEP: command.
The output shown consists of one
or more columns, each representing a linear combination of the
elements of :
Beneath each column are summary quantities for the observed adjustment:
the discrepancy for the adjustment, the prior expected size of the
discrepancy for the adjustment (the expected size of this discrepancy is
equal to the rank of the corresponding resolution matrix), and the
discrepancy ratio respectively.
This summary is also given with the
dv* option. Hence the values reported are (assuming
that the partial results are available, otherwise only the first column
is reported):
BD>option dv*
switches on standardised
discrepancy vector output.
BD>option -dv*
switches off standardised
discrepancy vector output.
BD>show (dv* )
displays standardised
discrepancy vector output for the current adjustment.
The dv* option is used to display output relating to the size and
standardised discrepancy vector for the adjustment, and for the partial adjustment
if any. Similar output is
available using the dv option.
Both the discrepancy vector and the partial discrepancy vector may be
retained interactively as assignments by using respectively the
dv and pdv arguments to the KEEP: command, but
they cannot be retained in standardised form.
The output shown consists of one
or more columns, each representing a linear combination of the
standardised elements of , where each element is
standardised to have expectation zero and prior variance unity:
Beneath each column are summary quantities for the observed adjustment:
the discrepancy for the adjustment, the prior expected size of the
discrepancy for the adjustment (the expected size of this discrepancy is
equal to the rank of the corresponding resolution matrix), and the
discrepancy ratio respectively. This summary is also given with the
dv option. Hence the values reported are (assuming
that the partial results are available, otherwise only the first column
is reported):
BD>option dvecorr
switches on discrepancy vector-element
correlation output.
BD>option -dvecorr
switches off discrepancy vector-element
correlation output.
BD>show (dvecorr )
displays
discrepancy vector-element correlation output for the current adjustment.
The dvecorr option yields the prior correlation between the
discrepancy vector for the current adjustment, B, and every
element . If a partial adjustment has been made, then
the prior correlations for the discrepancy vectors for the previous and
partial adjustments, B and B, with every
element are given.
BD>option dvcorr
switches on discrepancy vector
inter-correlation output.
BD>option -dvcorr
switches off discrepancy vector
inter-correlation output.
BD>show (dvcorr )
displays
discrepancy vector inter-correlation output for the current adjustment.
When the results of a partial adjustment are available, the
dvcorr option yields the prior correlations between the three
kinds of discrepancy vector: the discrepancy vector for the current
adjustment, B; the previous adjustment, B; and the
partial adjustment, B.
BD>option cda
switches on standard observed
adjusted expectation output for the canonical directions.
BD>option cda+
switches on extended observed
adjusted expectation output for the canonical directions.
BD>option -cda
switches off observed adjusted
expectation output for the canonical directions.
BD>show (cda )
displays standard observed
adjusted expectation output for the current adjustment for the canonical
directions.
BD>show (cda+ )
displays extended observed adjusted
expectation output for the current adjustment for the canonical
directions.
The cda and cda+ options are used to display standard
and extended results for the observed adjustments of canonical
directions. Suppose that is a canonical direction. Then the
standard output consists of its name and its observed adjusted
expectation, . The extended output consists additionally of its
initial expectation (always by convention); and its
standardised adjustment . Note that the adjusted
expectations for the canonical quantities can be obtained using the
cde or cde* options.
BD>option mcda
switches on standard observed
adjusted expectation output for the maximal canonical directions.
BD>option mcda+
switches on extended observed
adjusted expectation output for the maximal canonical directions.
BD>option -mcda
switches off observed adjusted
expectation output for the maximal canonical directions.
BD>show (mcda )
displays standard observed adjusted
expectation output for the current adjustment for the maximal canonical
directions.
BD>show (mcda+ )
displays extended observed adjusted
expectation output for the current adjustment for the maximal canonical
directions.
The mcda and mcda+ options are used to display standard
and extended results for the observed adjustments of maximal canonical
directions. Suppose that is a maximal canonical direction. Then
the standard output consists of its name and its observed adjusted
expectation, . The extended output consists additionally of its
initial expectation (always by convention); and its
standardised adjustment . The option is ignored when there is
no adjustment, and also when lack of exchangeability in implies
that there are no maximal results available.
BD>option msd
switches on a summary of the effects
of adjustment on individual elements.
BD>option -msd
switches off a summary of the effects
of adjustment on individual elements.
BD>show (msd )
displays a summary of the
effects of the current adjustment on individual elements.
The msd option summarises the output given by the a+ and
v+ options for individual elements .
For each such element the output consists of the initial and adjusted
expectation; the initial and adjusted standard deviations; and the
standardised value of the change in expectation.
BD>option pathsum
enables path summaries and
disables all other adjustment output options.
BD>option -pathsum
disables output of path
summaries.
The pathsum option is rather special because its being set
disables all other adjustment output options, and the setting of any
other adjustment output option likewise disables the pathsum
option. This is because the two sorts of option are incompatible, giving
messy output.
When switched on, the option gives a summary of sequential observed
adjustments, and in particular of the path correlations between
successive adjustments. Suppose that the base is adjusted
initially by the base , and then partially by the base\
, and so forth; and suppose that is the union
of the bases . Suppose that we
observe to be for each , and that we write
for the observed union .
Suppose that we have adjusted already by . For the
stepwise partial adjustment, when the base is
added, the information displayed is
(See [31][section 5.6].) The path correlation is undefined for the
first such adjustment.
When the results of a partial adjustment are available, the pc
option yields , the path correlation.
In the case of model fitting, this is the prior correlation between the
bearings for the previous and partial adjustments, and in the case of
model reduction, this is the prior correlation between the bearings for
the current and partial adjustments.
BD>option pcda
switches on standard observed
adjusted expectation output for the partial canonical directions.
BD>option pcda+
switches on extended observed
adjusted expectation output for the partial canonical directions.
BD>option -pcda
switches off observed adjusted
expectation output for the partial canonical directions.
BD>show (pcda )
displays standard observed adjusted
expectation output for the current adjustment for the partial canonical
directions.
BD>show (pcda+ )
displays extended observed adjusted
expectation output for the current adjustment for the partial canonical
directions.
The pcda and pcda+ options are used to display standard
and extended results for the observed adjustments of partial canonical
directions. Suppose that is a partial canonical direction. Then
the standard output consists of its name and its observed adjusted
expectation, . The extended output consists additionally of its
initial expectation (always by convention); and its
standardised adjustment . The option is ignored when there is
no partial adjustment.
Usage
Usage
Usage
Usage
Usage
Usage
Usage
Usage
Usage
Usage
Usage
Usage
Usage
Usage