Guilford-extended-rater: extended rater representation |
Elliott & Buttery (2022) propose Extended rater representations. According to Section 6.2.1 of the RaschPy manual, there are 4 extended rater representations: global, item, thresholds and matrix (two options). These can be implemented in several ways using Facets. Suggestions are made at Extended rater representations. Most are easy to implement in Facets.
Here is a Facets specification and data file for Items (suggestion 2b) and Matrix (suggestion 4b). These are more awkward to implement:
; Guilford-extended-rater.txt
Title = Ratings of Scientists: Extended rater representation: Items
Facets = 4 ; four facets: raters, examinees, items, raters+items
Inter-rater = 1, 2, 3 ; facet 1 is the rater facet, raters matched only on facets 2 and 3
Arrange = N,m ; arrange tables by measure-descending for all facets,
Positive = 2 ; the examinees have greater creativity with greater score
Non-centered = 2 ; examinees and items are centered on 0, judges are allowed to float
Unexpected = 2 ; report ratings if standardized residual >=|2|
;Usort = (1,2,3),(3,1,2),(Z,3) ; sort and report unexpected ratings several ways (1,2,3) is Senior, Junior, Trait
;Vertical = 2N,3A,2*,1L,1A ;define rulers to display and position facet elements
;Zscore = 1,2 ;report biases greater in size than 1 logit or with z>2
;Pt-biserial = measure ; point-measure correlation
; Extended rater representation: Items: 2b
Model = ?,?,?,?,R9 ; raters, examinees, items, raters+items produce ratings
; Extended rater representation: Matrix (Marginal=True): 4b
; Model = #,?,?,?,R9 ; raters (PCM), examinees, items, raters+items produce ratings
; Extended rater representation: Matrix (Marginal=False): 4b
; Model = ?,?,?,#,R9 ; raters, examinees, items, raters+items (PCM) produce ratings
Labels= ; to name the components
1,Raters, D ; dummy facet for rater fit and rater element number
1=Avogadro
2=Brahe
3=Cavendish
*
2,Examinees
2=Betty
5=Edward
7=George
1=Anne
3=Chris
4=David
6=Fred
*
3,Items ; these are the global item difficulties and for item element number
1=Attack
2=Basis
3=Clarity
4=Daring
5=Enthusiasm
*
4= Rater-item, G ; rater severity by item group-anchored by item
11 = 11, 0, 1 ; build this list of element numbers by starting with no element numbers here
12 = 12, 0, 2 ; - see below for instructions -
13 = 13, 0, 3
14 = 14, 0, 4
15 = 15, 0, 5
21 = 21, 0, 1
22 = 22, 0, 2
23 = 23, 0, 3
24 = 24, 0, 4
25 = 25, 0, 5
31 = 31, 0, 1
32 = 32, 0, 2
33 = 33, 0, 3
34 = 34, 0, 4
35 = 35, 0, 5
41 = 41, 0, 1
42 = 42, 0, 2
43 = 43, 0, 3
44 = 44, 0, 4
45 = 45, 0, 5
51 = 51, 0, 1
52 = 52, 0, 2
53 = 53, 0, 3
54 = 54, 0, 4
55 = 55, 0, 5
61 = 61, 0, 1
62 = 62, 0, 2
63 = 63, 0, 3
64 = 64, 0, 4
65 = 65, 0, 5
71 = 71, 0, 1
72 = 72, 0, 2
73 = 73, 0, 3
74 = 74, 0, 4
75 = 75, 0, 5
*
Dvalues=
4,1,$Element ; $Element - element number of facet 1 (raters) is used in element number of facet 4
4,3,$Element ; combined with element number of facet 3 (items) in element number of facet 4
*
Data=
; facet 1,2,3,(dvalues=4,) 5 ratings
1,1,1-5, 5,5,3,5,3 ; facet 4 element number = rater+item is inserted by Dvalues=
1,2,1-5, 9,7,5,8,5
1,3,1-5, 3,3,3,7,1
1,4,1-5, 7,3,1,3,3
1,5,1-5, 9,7,7,8,5
1,6,1-5, 3,5,3,5,1
1,7,1-5, 7,7,5,5,5
2,1,1-5, 6,5,4,6,3
2,2,1-5, 8,7,5,7,2
2,3,1-5, 4,5,3,6,6
2,4,1-5, 5,6,4,5,5
2,5,1-5, 2,4,3,2,3
2,6,1-5, 4,4,6,4,2
2,7,1-5, 3,3,5,5,4
3,1,1-5, 5,5,5,7,3
3,2,1-5, 7,7,5,7,5
3,3,1-5, 3,5,5,5,5
3,4,1-5, 5,3,3,3,1
3,5,1-5, 9,7,7,7,7
3,6,1-5, 3,3,3,5,3
3,7,1-5, 7,7,7,5,7 ; last line of data, and end of file
How to construct an Extended specification and data file from a standard specification and data file
1. Start with a standard 3-facet specification and data file, such as Guilford.txt
Facets = 3 ; three facets: judges, examinees, items
Model = ?B,?B,?,Creativity ; judges, examinees and items produce ratings on "Creativity".
Labels =
1, Senior scientists ; name of first (judge) facet
1 = Avogadro ; names of elements within senior scientist facet
2 = Brahe
3 = Cavendish
*
2,Junior Scientists ; name of second (examinee) facet
7=George ; element order does not matter
1=Anne
2=Betty
3=Chris
4=David
5=Edward
6=Fred
*
3,Traits ; name of third (item) facet
1=Attack
2=Basis
3=Clarity
4=Daring
5=Enthusiasm
*
Data=
1,1,1_5,5,5,3,5,3
....
2. Include rater-item interactions as a 4th facet:
Facets = 4 ; three facets: judges, examinees, items, judges+items
Model = ?,?,?,?, Creativity ; judges, examinees and judges+items produce ratings on "Creativity".
Labels=
1, Senior scientists, D ; Judges anchored at 0. Their measures are in Facet 4: judge-item
....
6=Fred
*
3,Traits ; name of third (item) facet - this will show overall item difficulty
......
5=Enthusiasm
*
4, Judge-Trait, G ; we will group-anchor by trait (item)
; nothing here yet
*
Dvalues=
4,1,$Element ; Facet 4 element number is "judge" (rater)
4,3,$Element ; "+trait" (item
*
Data=
1,1,1_5,5,5,3,5,3 ; no change to the data file
....
3. Analyze the modified specification and data file with Facets.
The analysis will fail, but will report the elements for Facet 4 in Table 2
4, Judge-Trait, G
11 = 11, 0,
12 = 12, 0, -
13 = 13, 0,
14 = 14, 0,
15 = 15, 0,
21 = 21, 0,
.....
4. Rectangular-copy (alt+mouse in NotePad++, etc.) the trait (item) numbers (2nd digit of element number) as group number
4, Judge-Trait, G
11 = 11, 0, 1 ; judge-Trait elements are group-anchored by Trait
12 = 12, 0, 2 -
13 = 13, 0, 3
14 = 14, 0, 4
15 = 15, 0, 5
21 = 21, 0, 1
....
5. Paste the Facet 4 element details into Labels= in the specification file
6. Do the Facets analysis of the modified speficiation file
Help for Facets (64-bit) Rasch Measurement and Rasch Analysis Software: www.winsteps.com Author: John Michael Linacre.