There are some methods to split an item into two or more columns based on person classification group.
1. From Winsteps, output your data to Excel using Output Files menu, RFILE= or IPMATRIX=
In Excel:
a.Sort the rows (cases) by the target demographic (or whatever) variable
b.Cut the responses to the target items by one demographic group into new columns
c.Label (row 1) the new columns with their item labels and demographic information
d.Save the Excel file under a new name
e.Use the Excel/RSSST menu to convert the new Excel file into a Winsteps control file
f.Analyze the new control file which now has the split items.
2. Use MFORMS=, then EDFILE=
3. Use FORMAT=, then EDFILE=
These examples use FORMAT=. Here are the changes to a control file to split an item:
Example 1: We want to split item 3 "Enjoy cooking", of our 12 item test into two items, based on the Gender (M,F) indicator in column 2 of the person label.
1. Increase number of items:
NI= 13 (add one NI=12)
NAME1= (add one if to the right of the items)
The new item will be the last item
2. Add coding, grouping for new item 13 in IREFER=, ISGROUPS=, KEY1=
3. Add a new item name before END LABELS for item 13:
Enjoy cooking (male)
Edit the current item name for item 3:
Enjoy cooking (female)
4. Use FORMAT= duplicate the old item into the new item's position, e.g.,
FORMAT = (12A,T3,1A,T13,99A)
adds the item in column 3 again in column 13. Then follows it with the old column 13 onwards, now starting in column 14..
5. Delete responses in the two versions of the item, based on the person demographics (Gender: M, F) in column 2 of the person label:
EDFILE=*
"?M" 3 x ; all Male responses to item 3 are changed to "x" (scored as missing - not administered)
"?{~M}" 13 x ; all not Male responses to item 13 are changed to "x"
*
6. The Control file is:
Title = "Survey of Personal Preferences"
NI = 13 ; was NI=12
ITEM1 = 1 ; responses start the line
NAME1 = 15 ; was NAME1=14
CODES=12345
FORMAT = (12A,T3,1A,T13,99A) ; repeat column 3 at column 13
EDFILE=*
"?M" 3 x ; all Male responses to item 3 are changed to "x" (scored as missing - not administered)
"?{~M}" 13 x ; all not Male responses to item 13 are changed to "x"
*
RFILE = rfile.txt ; File which shows the reformatted data
&END
Item 1
Item 2
Enjoy cooking (female) ; Item 3
Item 4
....
Item 12
Enjoy cooking (female) ; Item 12
END NAMES
321453324213 PM George ; M is Male in column 2 of person label
532134324522 RF Mary ; F is Female in column 2 of person label
......
The data become:
32x4533242131 PM George ; M is Male in column 2 of person label
532134324522x RF Mary ; F is Female in column 2 of person label
......
Example 2: Splitting item 19 in Liking for Science, example0.txt, data by gender, and moving gender to first column of person label:
Title= "Liking for Science"
ITEM1= 1 ; Starting column of item responses
NI= 26 ; Number of items ; one item added to the standard 25
XWIDE = 1 ; this matches the biggest data value observed
CODES= 012 ; matches the data
NAME1 = 28 ; Starting column for person label in formatted data record (gender)
; show column 19 again in column 26, and also column 48 (gender) again in column 28
FORMAT = "(25A1, T19, 1A1, T26, 1A1, T48, 2A1, T27, 50A1)"
EDITFILE=*
"{~F}" 26 m ; m = deliberately omitted response
"{~M}" 19 m
*
; reformatted file looks like ...
;123456789012345678901234567890123456789012345678901234
;1211102012222021122021020 M ROSSNER, MARC DANIEL M 1
;101010100111100112 1111102 F ROSSNER, REBECCA A. F 5
;101121101112101012 101210 ROSSNER, TR CAT 6
&END ; Item labels follow: columns in label
WATBIRDS ; Item 1
BOOKANIM ; Item 2
BOOKPLNT ; Item 3
WATGRASS ; Item 4
BOTLCANS ; Item 5
LOOKENCY ; Item 6
WATAMOVE ; Item 7
LOOKCRAK ; Item 8
NAMEWEED ; Item 9
LISTBIRD ; Item 10
FINDALIV ; Item 11
MUSEUM ; Item 12
GROGARDN ; Item 13
PIXPLNTS ; Item 14
ASTORIES ; Item 15
MAKEMAP ; Item 16
WATAEAT ; Item 17
PICNIC ; Item 18
GOTOZOOM ; Item 19 - Zoo - Male
WATBUGS ; Item 20
WATNEST ; Item 21
WHATEAT ; Item 22
WATCHRAT ; Item 23
FLWRSEAT ; Item 24
TALKPLNT ; Item 25
GOTOZOOF ; Item 26 - Zoo - Female
;23456789012345678901234567890123456789012345678901
END NAMES
1211102012222021122021020 ROSSNER, MARC DANIEL M 1
2222222222222222222222222 ROSSNER, LAWRENCE F. M 2
2211011012222122122121111 ROSSNER, TOBY G. M 3
1010010122122111122021111 ROSSNER, MICHAEL T. M 4
1010101001111001122111110 ROSSNER, REBECCA A. F 5
1011211011121010122101210 ROSSNER, TR CAT 6
2220022202222222222222222 WRIGHT, BENJAMIN M 7
2210021022222020122022022 LAMBERT, MD., ROSS W M 8
0110100112122001121120211 SCHULZ, MATTHEW M 9
2100010122221121122011021 HSIEH, DANIEL SEB M 10
2220011102222220222022022 HSIEH, PAUL FRED M 11
0100220101210011021000101 LIEBERMAN, DANIEL M 12
1211000102121121122111000 LIEBERMAN, BENJAMIN M 13
2110020212022100022000120 HWA, NANCY MARIE F 14
1111111111111111112211111 DYSON, STEPHIE NINA F 15
2221121122222222222122022 BUFF, MARGE BABY F 16
2222022222222222222222122 SCHATTNER, GAIL F 17
2222021022222221222122022 ERNST, RICHARD MAX M 18
2221011002222211122020022 FONTANILLA, HAMES M 19
1100010002122212122020022 ANGUIANO, ROB M 20
1111011112122111122021011 EISEN, NORM L. M 21
1211111112222121122121121 HOGAN, KATHLEEN F 22
2211112112222121222122122 VROOM, JEFF M 23
1221012102222110122021022 TOZER, AMY ELIZABETH F 24
2221120222222221222122022 SEILER, KAREN F 25
1111111111111111111111111 NEIMAN, RAYMOND M 26
1110011101122111022120221 DENNY, DON M 27
2211102002122121022012011 ALLEN, PETER M 28
1000101001110000022110200 LANDMAN, ALAN M 29
1000000002122000022012100 NORDGREN, JAN SWEDE M 30
2211011112221011122121111 SABILE, JACK M 31
1210010011222110121022021 ROSSNER, JACK M 32
2212021212222220222022021 ROSSNER, BESS F 33
2222122222222222222222122 PASTER, RUTH F 34
1110211011122111122011111 RINZLER, JAMES M 35
1110201001122111122011111 AMIRAULT, ZIPPY M 36
1210010012122111122012021 AIREHEAD, JOHN M 37
2211021112222121122222121 MOOSE, BULLWINKLE M 38
2221022112222122222122022 SQURREL, ROCKY J. M 39
2221011022222221222122022 BADENOV, BORIS M 40
2222222222222222222222122 FATALE, NATASHA F 41
2121011112122221122011021 LEADER, FEARLESS M 42
2220112022222121222022021 MAN, SPIDER M 43
1210021012212211022011021 CIANCI, BUDDY M 44
2221022222222222222222022 MCLOUGHLIN, BILLY M 45
2200020002222000222122010 MULLER, JEFF M 46
1110220112212111022010000 VAN DAM, ANDY M 47
2221022222222221222122022 CHAZELLE, BERNIE M 48
1100110101110112111111210 BAUDET, GERARD M 49
2222121222222222222122122 DOEPPNER, TOM M 50
1210021022222221122021022 REISS, STEVE M 51
1111111111111111122111111 STULTZ, NEWELL M 52
0100101010110011021010000 SABOL, ANDREW M 53
1111021121122112122021011 BABBOO, BABOO 54
1010010011122112222111110 DRISKEL, MICHAEL M 55
2221012012222222222122022 EASTWOOD, CLINT M 56
2220022022222222222022022 CLAPP, CHARLIE M 57
1221010122222221222021021 CLAPP, DOCENT 58
2221021122222221222022222 CLAPP, LB 59
1221111122222121122121021 SQUILLY, MICHAEL M 60
2111010022122111122011022 SQUILLY, BAY OF 61
2220000022222220222020000 BEISER, ED M 62
1211010122222122222122010 BECKER, SELWIN M 63
2210022112222121222122011 CORLEONE, VITO M 64
1211211002222012122121200 CORLEONE, MICHAEL M 65
1201010012222120222022021 PINHEAD, ZIPPY 66
1211001001212121222121012 MALAPROP, MRS. F 67
1200010102120120022001010 BOND, JAMES M 68
1210010122222122222021021 BLOFELD, VILLAIN M 69
2222022012222122222022021 KENT, CLARK M 70
2201222001220022220222201 STOLLER, DAVE M 71
1001000100011010021200201 JACKSON, SOLOMON M 72
2012010122210201102202022 SANDBERG, RYNE M 73
2220022002222222222022012 PATRIARCA, RAY M 74
1200110112122221022020010 PAULING, LINUS M 75
Example 3: We have non-uniform DIF on an item and want to split the persons high-low.
We need to split the persons on the item by raw score or Rasch measure. To do this conveniently, use RFILE= to output data to Excel. Also output the PFILE= to Excel. Then you can copy the score/measure column from the PFILE to the RFILE. Sort the RFILE by the new score/measure column. then split the desired item column into two columns high/low or whatever.
