RESCORE= response recoding (with NEWSCORE= or KEYn=)

The responses in your data file may not be coded as you desire. The responses to some or all of the items can be rescored or keyed using RESCORE=. RESCORE= and NEWSCORE= are ignored when KEYn= is specified, except as below. If rescoring implies that the items have different rating (or partial credit) scale structures, ISGROUPS= may also be required.

RESCORE= has three forms: RESCORE=1101110 and RESCORE=* list * and RESCORE=*filename

RESCORE=" " or 2 or is omitted

All items are recoded using NEWSCORE=. RESCORE=2 is the standard when NEWSCORE= is specified.

RESCORE= some combination of 1's and 0's

Only items corresponding to 1's are recoded with NEWSCORE= or scored with KEYn=. When KEYn is specified, NEWSCORE= is ignored.

If some, but not all, items are to be recoded or keyed, assign a character string to RESCORE= in which "1" means "recode (key) the item", and "0" (or blank) means "do not recode (key) the item". The position of the "0" or "1" in the RESCORE= string must match the position of the item-response in the item-string.

When XWIDE=2 or more, then

either        (a) Use one character per XWIDE and blanks,

               NI=8

               XWIDE=2

               RESCORE=' 1 0 1 0 1 0 1 1'

or        (b) Use one character per item with no blanks

               NI=8

               XWIDE=2

               RESCORE='10101011'

Example 1:        The original codes are "0" and "1". You want to reverse these codes, i.e., 1 0 and 0 1, for all items.

                       XWIDE=1                one character wide responses (the standard)

                       CODES =01                valid response codes are 0 and 1 (the standard)

                       NEWSCORE=10                desired response scoring

                       RESCORE=2                rescore all items - this line can be omitted

Example 2:        Your data is coded "0" and "1". This is correct for all 10 items except for items 1 and 7 which have the reverse meaning, i.e. 1 0 and 0 1.

                       NI=10                ten items

                       CODES =01                standard, shown here for clarity

(a) old method - which still works:

                       NEWSCORE=10                revised scoring

                       RESCORE=1000001000        only for items 1 and 7

or

                       NEWSCORE=20

                       RESCORE=*

                       1        1                item 1 is to be rescored

                       7        1                item 7 is to be rescored

                       *

(b) new method - recommended:

                       IVALUE1 =10                revised scoring

                       IVALUE0 =01                scoring unchanged, so this line can be omitted.

                       IREFER   =1000001000        only for items 1 and 7

If XWIDE=2, use one or two columns per RESCORE= code, e.g., " 1" or "1 " mean recode (key). " 0" or "0 " mean do not recode (key).

Example 3:        The original codes are " 0" and " 1". You want to reverse these codes, i.e., 1 0 and 0 1, for items 1 and 7 of a ten item test.

                       NI    =10                        ; ten items

                       XWIDE =2                        ; two characters wide

                       CODES   =" 0 1"                ; original codes

                       NEWSCORE=" 1 0"                ; new values

                       RESCORE =" 1 0 0 0 0 0 1 0 0 0"        ; rescore items 1 & 7

Example 4:        The original codes are "0", "1", and "2". You want 0 0, 1 1, and 2 1 for all items

                       XWIDE=1                one character wide (standard)

                       CODES =012                valid codes

                       NEWSCORE=011        desired scoring

Example 5:        The original codes are "0", "1", and "2". You want to make 0 2, 1 1, and 2 0, for even-numbered items in a twenty item test.

                       NI=20                                twenty items

                       CODES =012                        three valid codes

                       NEWSCORE=210                desired scoring

                       RESCORE=01010101010101010101        rescore "even" items

Example 6:        The original codes are "0", "1", "2", "3" and some others. You want to make all non-specified codes into "0", but to treat codes of "2" as missing.

                       CODES = 0123                four valid codes

                       NEWSCORE= 01X3        response code 2 will be ignored

                       MISSCORE=0                treat all invalid codes as 0

Example 7:        The original codes are "0", "1", "2", "3". You want to rescore some items selectively using KEY1= and KEY2= and to leave the others unchanged - their data codes will be their rating values. For items 5 and 6, 0 0, 1 0, 2 1, 3 2; for item 7, 0 0, 1 0, 2 0, 3 1. Responses to other items are already entered correctly as 0, 1, 2, or 3.

                       CODES =0123        valid codes

                       RESCORE=0000111000        rescore items 5,6,7

                       KEY1  =****223***        keyed for selected items

                       KEY2  =****33X***        the X will be ignored

                                       ^ read these columns vertically

Example 8:        Multiple score key for items 1 to 10. Items 11 to 15 are on a rating scale of 1 to 5

                       CODES  = abcd12345

                       KEY1   = bacdbaddcd*****

                       RESCORE= 111111111100000        ; RESCORE= signals when to apply KEY1=