This is only needed for complicated rescoring.
Specifies the score values assigned to response choices which match KEY1= etc. To assign responses matching key to the "missing" value of -1, make the corresponding KEYSCR= entry blank or some other non-numeric character.
When XWIDE=1, each value can only take one position, so that only KEY1= through KEY9= can be reassigned. KEY10= through KEY99= can also be used but keep their standard values of 10 through 99.
When XWIDE=2, each value takes two positions, and the values corresponding to all keys, KEY1= through KEY99=, can be reassigned.
Example 1: Three keys are used, and XWIDE=1.
Response categories in KEY1= will be coded "1"
Response categories in KEY2= will be coded "2"
Response categories in KEY3= will be coded "3"
KEYSCR=123 (standard)
Example 2: Three keys are used, and XWIDE=1.
Response categories in KEY1= will be coded "2"
Response categories in KEY2= will be coded "1"
Response categories in KEY3= will be coded "1"
KEYSCR=211
Example 3: Three keys are used, and XWIDE=2
Response categories in KEY1= will be coded "3"
Response categories in KEY2= will be coded "2"
Response categories in KEY3= will be coded "1"
KEYSCR=030201
or
KEYSCR=" 3 2 1"
Example 4: Three keys are used, and XWIDE=1
Response categories in KEY3= will be coded "1"
Response categories in KEY6= will be coded "missing"
Response categories in KEY9= will be coded "3"
KEY3=BACDCACDBA response keys
KEY6=ABDADCDCAB
KEY9=CCBCBBBBCC
KEYSCR=xx1xxXxx3 scores for keys
The "x"s correspond to unused keys, and so will be ignored.
The "X" corresponds to specified KEY6=, but is non-numeric and so will cause responses matching KEY6= to be ignored, i.e. treated as missing.
Example 5: Some items in a test have two correct answers, so two keys are used. Since both answers are equally good, KEY1= and KEY2= have the same value, specified by KEYSCR=. But some items have only one correct answer so in one key "*", a character not in CODES=, is used to prevent a match.
CODES=1234
KEY1 =23313141324134242113
KEY2 =31*324321*3142314*** ; * is not in CODES=
KEYSCR=11 both KEYs scored 1
Example 6: More than 9 KEYn= lines, together with KEYSCR=, are required for a complex scoring model for 20 items, but the original data are only one character wide.
Original data: Person name: columns 1-10
20 Item responses: columns 21-40
Looks like: M. Stewart..........1321233212321232134
Solution: reformat from XWIDE=1 to XWIDE=2
TITLE="FORMAT= from XWIDE=1 to =2"
FORMAT=(10A1,10X,20A1) ; 10 of Name, skip 10, 20 of responses
NI=20
NAME1=1
ITEM1=11 Responses in column 11 of reformatted record
XWIDE=2
CODES="1 2 3 4 " Original response now "response blank"
KEY1 ="1 2 1 3 2 1 2 3 1 4 3 2 1 1 1 1 1 2 1 1 " Keying 20 items
KEY2 ="2 1 2 1 1 2 1 1 2 1 1 1 2 3 3 2 2 * 2 1 "
....
KEY10="3 3 3 2 3 3 4 2 3 * * * 4 4 4 4 4 4 4 4 "
KEYSCR="1 2 3 2 2 2 3 4 1 4 " ; Renumbering 10 KEYn=
&END
