Dvalues for data facets =

Up  Previous  Next

If a facet entry in the data file has the same value for all data lines, then it can be omitted from the data, and replaced by a Dvalue= statement, e.g., a set of items administered to all persons.

 

If a facet entry in the data file has its value in the label of another element, then it can be omitted from the data, and replaced by a Dvalue= statement, e.g., a demographic indicator in the person element label.

 

I. Elements numbers the same for every data line. This occurs when, for instance, every examinee is rated on items 1-5, or when every patient is observed on items 1-13.

 

The format for one Dvalue= is:

Dvalue = facet entry number in data, constant value

 

The format for several Dvalue= is:

Dvalue=

facet entry number in data, constant value

facet entry number in data, constant value

....

*

 

Example 1: In the Guilford.txt example, the standard layout is:

Facets = 3

Data =

1, 1, 1-5, 5,5,3,5,3

1, 2, 1-5, 9,7,5,8,5

....

3, 7, 1-5, 7,7,7,5,7

 

This is the same as:

Facets = 3

Dvalue = 3, 1-5

Data =

1, 1, 5,5,3,5,3

1, 2, 9,7,5,8,5

....

3, 7, 7,7,7,5,7

 

II. An element identifier for one facet (the target) is in the element label of another facet (the reference) for the same observation. If the segment in the reference element label is a number, that becomes the element number for the target facet. If non-numeric, then the segment is matched to the start of the elements labels given for the target facets, up to :. If ":" is omitted, then a match is made to the left-end of the target element label.

The format is:

Dvalue=

facet entry number for target facet, reference facet location in data, start column in the label, length of field in the label

*

 

Example 1. The Knox Cube Test with interactions example data set contains replacements for catets 3 and 4:

Dvalues = ; The child gender and number of taps are in the labels, so can be omitted from the data

3, 1, 1, 1 ; Child gender element for Facet 3 is indicated in the label of Facet 1, column 1 length 1

4, 2, 1, 1 ; Item length element for Facet 4 is indicated in the label of Facet 2, column 1 length 1

*

 

Example 2. Ethnicity is indicated by a numeric value in column 1 of the student label. Gender is in column 2.

Facets=4  ; persons, item, ethnicity, gender

Models = ?,?B,?,?B,D ; we want to look for item-gender DIF

Labels=

1, Student

234 = 61 Jose  ; Jose is a Hispanic (6 in column 1), male (column 2 of label=1)

235 = 32 Mary  ; Mary is a Caucasian (3 in column 1), female (column 2 of label =2)

......

*

2, Items

1-4   ; 4 dichotomous items, scored 0-1.

*

3, Ethnicity, A  ; elements anchored at 0 to avoid distorting student measures

3, Caucasian, 0

6, Hispanic, 0

*

4, Gender, A  ; elements anchored at 0 to avoid distorting student measures

1, Male, 0

2, Female, 0

*

Dvalues=

2, 1-4   ; all students administered items 1-4

3, 1, 1, 1  ; element number for facet 3 (ethnicity) is in element label for facet 1, column 1 with a length of 1

4, 1, 2, 1  ; element number for facet 4 (gender) is in element label for facet 1, column 2 with a length of 1

*

Data=

234, 0,1,1,0  ; facet 1 element 234 then 4 observations. would be 234, 1-4, 6, 1, 0,1,1,0 without Dvalues =

235, 1,1,0,0  ; facet 1 element 235 then 4 observations. would be 235, 1-4, 3, 2, 1,1,0,0 without Dvalues =

 

Example 3. Gender is indicated by a non-numeric value in column 2 of the student label

Facets=3 ; persons, item, DIF indicator

Models = ?,?B,?B ; we want to look for gender-related item DIF

Labels=

1, Student

234 = 6M Jose  ; Jose is a male (column 2 of label=1)

235 = 3F Mary  ; Mary is a female (column 2 of label =2)

......

*

2, Items

1-4   ; 4 dichotomous items, scored 0-1.

*

3, Gender, A  ; elements anchored at 0 to avoid distorting student measures; only for DIF analysis

1, M:Male, 0  ; M: - colon, so M is used to match the code in the student label

2, Female, 0  ; no :, so match is made to left-end of label for length of dvalues= field (width 1 in this example).

*

Dvalues=

2, 1-4   ; all students administered items 1-4

3, 1, 2, 1  ; element number for facet 3 is in column 2 of element label for facet 1.

*

Data=

234, 0,1,1,0  ; facet 1 element 234 with 4 observations. woould be 234, 1-4, 1, 0,1,1,0 without Dvalues =

235, 1,1,0,0  ; facet 1 element 235 with 4 observations. would be 235, 1-4, 2, 1,1,0,0 without Dvalues =

 

Example 4: To test that Dvalues= is functioning correctly:

title= "dvalue test"

facets = 2

residualfile=dvalueresiduals.txt

models=?,?,R

labels=

1, Facet 1

1=Fred

2=Mary

*

2, Facet 2

1=M:supervisor

2=F:janitor

*

dvalues=

2,1,1,1 ; facet 2 is the first character of the element for facet 1

*

data=

1-2, 0,1

1-2, 1,0

 

; this analyzes as (see residualfile=)

; 1 1 2 2 Obs

; 1 Fred 2 F:janitor 0

; 2 Mary 1 M:supervi 1

; 1 Fred 2 F:janitor 1

; 2 Mary 1 M:supervi 0

 


Help for Facets Rasch Measurement Software: www.winsteps.com.