Example 14: Multiple rescorings, response structures and widths

Introductory example: A test has 30 multiple choice question (keyed A, B, C, D) and 5 essay questions, each has its own rating scale definition (scored 0, 1, or 2), i.e., they accord with the partial credit model. ISGROUPS= is used to identify the partial credit items. IREFER= and IVALUE= are used for scoring the questions

; This control file is EXAM14.TXT

TITLE = "AN EXAMPLE OF MULTIPLE ITEM TYPES"

; THIS PART DEFINES THE ANALYTICAL MODEL

; MCQ ARE ONE GROUPING, WHICH WE WILL CALL "M" - PARTIAL CREDIT ITEMS ARE GROUPED AS "0"

; "0" MEANS EACH ITEM IS IN A GROUPING BY ITSELF, i.e., partial credit

ISGROUPS = MMMMMMMMMMMMMMMMMMMMMMMMMMMMMM00000

; THIS PART DEFINES THE RESPONSE SCORING

; The MCQ key is: ABCAAACBAAABABABBAAABABBAAABAB

IREFER = ABCAAACBAAABABABBAAABABBAAABAB00000 ; REFERENCE THE MCQ ITEMS BY THEIR SCORING KEY

CODES = ABCD012; ALL VALID CODES IN THE DATA FILE

IVALUE0= 0000012; THESE ARE THE ESSAY ITEMS

IVALUEA= 1000000 ; MCQ ITEMS WITH A AS THE CORRECT ANSWER

IVALUEB= 0100000 ; B IS CORRECT

IVALUEC= 0010000 ; C IS CORRECT

IVALUED= 0001000 ; D IS CORRECT

MISSING-VALUES-SCORED = 0 ; SO ALL 9'S (OR ANYTHING ELSE NOT IN CODES=) ARE SCORED 0

ITEM1 = 1 ; START OF ITEM STRING

NI = 35 ; NUMBER OF ITEM RESPONSES

NAME1 = 37 ; START OF PERSON IDENTIFICATION

&END

; ITEM LABELS HERE

END LABELS

ACAACBAA9DC9AB9C9DCC9ABBA9AACB00102 1 ; DATA ARE HERE

A9D9AAAB9DA999AA9BBD999999B9AA12100 2

A9C9AACD9AB9AB9B99D9DB9ACBD9AB10120 3

A9C9AACB9AD99999CA9ABD9999999900120 4

A9C9AAAB9AA99999CD9ABB9999999910120 5

A9C9ADCBBAA9CDABD9DACCABBA9ABC21212 6

B9D9AAAB9CB99999DB9ABC9999999900100 7

A9D9BAAB9AA99999BA9ABB9999999900101 8

A9D9BACB9DC99999CA9DBB9999999921201 9

A9A9AACD9BA99999AD9ABB9999999910120 10

A9C9AACB9AA99999DC9ABD9999999900120 11

A9C9AAAB9CA99999BA9CBB9999999921201 12

A9C9CACB9AC99999CB9ABD9999999910120 13

A9D9AACB9AA99999AA9ABD999999D900000 14

A9C9DDCB9AA99999CB9ABD99999C9D21201 15

A9C9AABB9AD9999CD9ABC9999ABDAB11110 16

A9CAACB9ABBC9ADBB9ABDABBA9ADCB00120 17

CBBAB9CAAC9BBBC9BCBACDD9ADDCAB10120 18

A9C9BAAB9AD99CB9BBBA9ABDACDD9A00120 19

C9D9BDDB9BBBB9ACBADBC9AADBBCBC21201 20

A9C9AABB9DABA9ABDDABCABBA9AACB00120 21

A9D9BDCB9DCAC9DBBADBBACBA9ABAC00001 22

A9C9BACB9DADA9DBBDABBACBA9ABBB21201 23

A9D9BACC9AADC9DBBAABBADBC9ABCB10012 24

A9D9BAAB9ABCC9ABBDABBACBB9ABBB21201 25

Example 14: A test comprises multiple-choice questions (MCQ), true-false, and two different response structure formats. These are combined in one data file. The MCQ scoring key, and also the True-False scoring key are used as item cluster references in the IREFER= and IVALUE= specifications to simplify key checking.

EXAM14DT.TXT data format is:
Cols 1-5	Person id.
Cols 6-10	5 MCQ items (A,B,C,D, # = missing, wrong. Some items may be miskeyed
Cols 11-15	5 True-False items, responses (S, N). For some of these S="True" is correct, for others "N"=False is correct
Cols 16-25	10 Rating scale items (N=0,P=1,M=2,S=3). Some items may be reversed.
Cols 26-29	2 Evaluation items (0-12). - (See exam14b.txt) - second analysis only.

First analysis: all responses one column wide

The control file, EXAM14.TXT, is:

; This file is EXAM14.TXT

TITLE="Multiple Response Formats, Same Response Width"

DATA=exam14dt.txt

;EXAM14DT.TXT data format is

;Cols 1-5 Person id.

;Cols 6-10 5 MCQ items (A,B,C,D, # = missing, wrong)

; Some items may be miskeyed

;Cols 11-15 5 True-False items, responses (S, N)

; For some of these S="True" is correct, for others "N"=False is correct

;Cols 16-25 10 Rating scale items (N=0,P=1,M=2,S=3)

; Some items may be reversed.

;Cols 26-29 2 Evaluation items (0-12). - (See exam14b.con)

NAME1=1

ITEM1=6

NI=20

; THESE CODES LINE UP WITH THE ITEM COLUMNS

; 0 1 2

; 12345678901234567890

; TO DEFINE RESPONSE STRUCTURE CLUSTERS

ISGROUPS = 11111222223333333333

; TO DEFINE RESCORING CLUSTERS

IREFER = BACDCSNSNSRRRRRRRRRR

; IREFER = X MATCHES IVALUEX=

; IVALUE?= MATCHES WITH CODES=

CODES = ABCD#SNPM ; CODES IN ORIGINAL DATA FILE

IVALUEA = 10000**** ; MCQ RESPONSE A IS CORRECT

IVALUEB = 01000**** ; MCQ RESPONSE B IS CORRECT

IVALUEC = 00100**** ; MCQ RESPONSE C IS CORRECT

IVALUED = 00010**** ; MCQ RESPONSE D IS CORRECT

IVALUES = *****10** ; "S" IS THE CORRECT ANSWER

IVALUEN = *****01** ; "N" IS THE CORRECT ANSWER

IVALUER = *****3012 ; "NPMS" RATING SCALE

STKEEP=YES ; KEEP UNUSED INTERMEDIATE CATEGORIES IN RATING SCALES

INUMBER=YES ; NO ITEM INFORMATION AVAILABLE

&END

Second analysis: responses one and two columns wide

Including the last two items with long numeric response structures and 2-column format, the control file becomes EXAM14B.TXT. Since some responses are in 2 column format, XWIDE=2. FORMAT= is used to transform all responses into XWIDE=2 format.

EXAM14B.TXT is:

&INST

TITLE="Multiple Response Formats, Same Response Width"

DATA=EXAM14DT.TXT

;EXAM14DT.TXT data format is

;Cols 1-5 Person id.

;Cols 6-10 5 MCQ items (A,B,C,D, # = missing, wrong)

; Some items may be miskeyed

;Cols 11-15 5 True-False items, responses (S, N)

; For some of these S="True" is correct, for others "N"=False is correct

;Cols 16-25 10 Rating scale items (N=0,P=1,M=2,S=3)

; Some items may be reversed.

;Cols 26-29 2 Evaluation items (0-12)

NAME1=1

ITEM1=6

NI=22 ;20-1 COLUMN + 2 2-COLUMN ITEMS

XWIDE=2 ;XWIDE FOR WIDEST FIELD

FORMAT=(5A1, 20A1,2A2) ; PERSON LABEL & FIRST 20 ITEMS 1 CHARACTER COLUMNS

; ; LAST 2 ITEMS ARE 2 CHARACTER COLUMNS

; THESE ARE SET UP FOR XWIDE=2

; FOR RESPONSE STRUCTURE DEFINITIONS

; TO DEFINE RATING SCALE CLUSTERS

ISGROUPS = 1111122222333333333344

IREFER = BACDCSNSNSRRRRRRRRRREE

; IREFER = X MATCHES IVALUEX=

; IVALUE?= MATCHES WITH CODES=

; CODES ARE 2 CHARACTERS WIDE

CODES = "A B C D # S N P M 121110 9 8 7 6 5 4 3 2 1 0"; CODES IN DATA

IVALUEA = "1 0 0 0 0 * * * * * * * * * * * * * * * * *"; MCQ A IS CORRECT

IVALUEB = "0 1 0 0 0 * * * * * * * * * * * * * * * * *"; MCQ B IS CORRECT

IVALUEC = "0 0 1 0 0 * * * * * * * * * * * * * * * * *"; MCQ C IS CORRECT

IVALUED = "0 0 0 1 0 * * * * * * * * * * * * * * * * *"; MCQ D IS CORRECT

IVALUES = "* * * * * 1 0 * * * * * * * * * * * * * * *"; "S" IS CORRECT

IVALUEN = "* * * * * 0 1 * * * * * * * * * * * * * * *"; "N" IS CORRECT

IVALUER = "* * * * * 3 0 1 2 * * * * * * * * * * * * *"; "NPMS" RATING SCALE

IVALUEE = "* * * * * * * * * 121110 9 8 7 6 5 4 3 2 1 0"; 0-12 RATING SCALE

STKEEP=YES ; KEEP UNUSED INTERMEDIATE CATEGORIES IN RATING SCALES

INUMBER=YES ; NO ITEM INFORMATION AVAILABLE

&END

This can also be done with MFORMS=

EXAM14C.TXT is:

; This file is EXAM14C.TXT

&INST

TITLE="Multiple Response Formats, Same Response Width"

;EXAM14DT.TXT data format is

;Cols 1-5 Person id.

;Cols 6-10 5 MCQ items (A,B,C,D, # = missing, wrong)

; Some items may be miskeyed

;Cols 11-15 5 True-False items, responses (S, N)

; For some of these S="True" is correct, for others "N"=False is correct

;Cols 16-25 10 Rating scale items (N=0,P=1,M=2,S=3)

; Some items may be reversed.

;Cols 26-29 2 Evaluation items (0-12)

; Reformatted data record is:

;Cols 1-5 Person id

;Cols 6-7 Item 1 = original Col. 6

;Cols 8-9 Item 2 = original Col. 7

;.....

;Cols 48-49 Item 22

NAME1=1

ITEM1=6

NI=22 ;20-1 COLUMN + 2 2-COLUMN ITEMS

XWIDE=2 ;XWIDE FOR WIDEST FIELD

mforms=*

data=exam14dt.txt ; the name of an input data file

L = 1 ; the are 2 lines in input data file for each data record

P1-5 = 1 ; person label characters 1 through 5 start in column 1

; in the following "C" is used because "I" uses XWIDE=2

C6 = 6 ; original item 1 in column 6 goes in column 6

C8 = 7 ; original item 2 in column 7 goes in column 8

C10 = 8

C12 = 9

C14 = 10

C16 = 11

C18 = 12

C20 = 13

C22 = 14

C24 = 15

C26 = 16

C28 = 17

C30 = 18

C32 = 19

C34 = 20

C36 = 21

C38 = 22

C40 = 23

C42 = 24

C44 = 25 ; original item 20 in column 25 goes in column 44-45

I21-22 = 26 ; two-character items 21 and 22 start in column 26

* ; end of mforms= command

; THESE ARE SET UP FOR XWIDE=2 FOR RESPONSE STRUCTURE DEFINITIONS