Persons to be selected may be specified by using the PSELECT= instruction to match characters within the person name. Persons deleted by PDFILE= etc. are never selected by PSELECT=.
This can be done before analysis in the control file or with "Extra specifications". It can also be done after the analysis using the "Specification" pull-down menu.
Persons to be selected may be specified by using the PSELECT= instruction to match characters within the person name. Persons deleted by PDFILE= etc. are never selected by PSELECT=.
This can be done before analysis in the control file or with "Extra specifications". It can also be done after the analysis using the "Specification" pull-down menu.
Selection rules. Person and item selections must be in quotation marks " " if they contain blanks or tabs, and follow the selection rules:
|
Control characters match label or name. They start at the first column of the label or name.
|
?
|
matches any character
|
*
|
matches any string of characters - must be last selection character.
If * is in the first column, then every available person or item is selected.
|
A
|
matches A in the person label, and similarly all other characters except these: { }"
|
{..}
|
braces characters which can match a single character: {ABC} matches A or B or C.
|
{.. - ..}
|
matches single characters in a range. {0-9} matches digits in the range 0 to 9.
|
{.. --..}
|
matches a single "-" {AB--} matches A or B or "-".
|
{~ABX}
|
omits persons or items which match A or B or X
|
{" "}
|
matches a blank in the label
|
@fieldname=
|
positions the next selection character at the start of the specified field
|
Each PSELECT= performed using the "Specification" pull-down menu selects from all those analyzed. For incremental selections from the Specification menu, i.e., selecting from those already selected, specify +PSELECT=
Example 0: In Example 0, select only Males:
PSELECT=@GENDER=M
In Example 0, select only Males with last names starting with R
PSELECT=@GENDER=M?R
In Example 0, select only Males with last names not starting with R
PSELECT=@GENDER=M?{~R}
To select only Males and Females, not missing data codes or data entry errors:
PSELECT=@GENDER={MF}
Example 1: Select for analysis only persons with M in the 5th column of person name. Person name starts in column 6 of the data record:
NAME1=6 Person name field starts in col. 6
NAMLEN=8 Person name field is 8 characters long
PSELECT=????M* Column 5 of person name is sex
|
END NAMES
xxxxxBPL M J 01101000101001 selected
xxxxxMEL F S 01001000111100 omitted
-----1234 selection column
Example 2: Select for analysis all persons with code "A 4" in columns 2-4 of their names. Person name starts in column 23, so target characters starts in column 24:
NAME1=23 person name starts in column 23
PSELECT="?A 4*" quotes because a blank is included. A is in col. 2 etc.
selects ZA 4PQRS
Example 3: Select all Male (M in column 2) persons who are Asian or Hispanic (A or H in column 4):
PSELECT=?M?{AH}*
selects 1M3A456 MPQRS
selects 1M5H689 ABCDE
omits 1X2A123 QWERT
Example 4: Select Males (M in column 8) in School 23 (023 in column 14-16):
PSELECT= ???????M?????023*
Selects: 1234567MABCDE023XYZ
Example 5: Select codes 1,2,3,4,5,6,7, in column 2:
PSELECT=?{1-7}*
Example 6: Analyze only males (column 4 or person-id). Then report only School C (column 1). Then only report Grades 4 and 5 (column 2) in School C.
PSELECT=???M* in the Control file or at the Extra Specifications prompt.
PSELECT=C* using the Specification pull-down menu, after the analysis
+PSELECT=?{45}* using the Specification pull-down menu.
Example 7: There are 100000 examinees from 17 school districts in one data file. I want to process only 16 districts.
If the district code is one character, then you can select the 16 codes you want, for instance:
If the codes are letters in column 4 of the person label
PSELECT=???{ABDEFGHIJKLMNOPQR}
or
PSELECT=???{~C} ; omits unwanted school district C
Example 8: If the field to be selected is already identified, then this simplifies counting the column:
@Dx = 16E17 ; $C44W2
16E17 means that the field is in columns 16 and 17 of the person label. Look at a data record to see what is actually there.
PSELECT=???????????????23 ; selects "23" in columns 16 and 17
Example 9: We want person group X (in column 4 of the person label) except for subgroups 11, 24 (in columns 6, 7 of the person label):
PSELECT = "???X?{12}{14}" ; this selects X and 11, 14, 21, 24
EDFILE=*
"?????14" "?" . ; convert 14 to missing data
"?????21" "?" . ; convert 21 to missing data
*
Example 10: Item statistics are needed for each person classification.
If you want PSELECT= to alter the item statistics, then please do separate Winsteps analyses for each PSELECT= value. Place the PSELECT= in your control file, or enter it at the "Extra Specifications (if any)" prompt. This PSELECT= will change what is shown for the persons and the items.
If you want to keep the item difficulties, person abilities and rating scale structure from the complete analysis, rather than re-estimating them. Then
Complete analysis -> PFILE=pf.txt, IFILE=if.txt, SFILE=sf.txt,
PAFILE=pf.txt, IAFILE=if.txt, SAFILE=sf.txt, PSELECT= -> selected analysis.
The "Specification" pull-down menu alter what is included in directly-relevant output tables.
So, entering PSELECT= at the "Specification" pull-down menu will change what is shown in the person tables, but not what is shown in the item tables.
Example 11: Select persons by ability level:
We need a code in the person label for the person's performance level:
Suggestion: analyze the data with Winsteps
Output the original data codes with IPMATRIX=, the person labels and the person measures to Excel
Sort the Excel worksheet by person measure: put the level code in a new Excel columns
Winsteps Excel/RSSST menu to go from the Excel worksheet back to the Winsteps control and data file. Include the level code column with the person label
Add in the scoring key etc.
Analyze the revised Winsteps control and data file
Then we can PSELECT= each performance level
