Facets requires specifications and data. The data are qualitative observations. These are numbered in ascending sequence along the latent variable. For instance:
Qualitatively-ordered observations
must be exclusive, extensive, exhaustive
|
Numerical (scored, coded) observations:
higher number = more of the latent variable
must be 0 or positive integers
|
Yes-No
|
1-0
|
Correct-Incorrect
|
1-0
|
None-Some-All
|
0-1-2
|
Likert scale: Strongly Disagree-Disagree-Neutral- Agree-Strongly Agree
|
1-2-3-4-5
|
Essay rating
|
1-2-3-4-5-6-7-8-9
|
Attitude survey
|
1-2-3-4-5-6-7-8-9-10
|
Visual-analog scale or Percent
|
0-1-2-3-4-....-96-97-98-99-100
|
The specifications define how these data are modeled to be the outcome of a linear combination of elements. Each element belongs to a facet. All facets and elements are numbered. For instance:
Facet (number)
|
Elements (numbers)
|
Examinees (1)
|
Albert (1), Beryl (2), Claudia (3), ...
|
Raters (2)
|
Prof. Cilic (1), Dr. Nadal (2), Ing. Laver (3), ...
|
Items (3)
|
Item 1 (1), Item 2 (2), Item 3 (3), ...
|
Time-points (4)
|
Admission (1), Discharge (2)
|
The measurement model becomes:
Examinee ability + Rater leniency - Item difficulty - Time-point challenge → Observation
a) Specifications are instructions on how to analyze the data. They are in the Facets specification file. They look like:
Specification = value
Most specifications have standard default values, suitable for most analyses.
Example 1: an excerpt from a list of specifications:
Title = Paired-judge Reading Test ; the report heading line
Facets = 4 ; facets are 1=examinees, 2=raters, 3=items, 4=time-point
Positive = 1, 2 ; examinees are able, raters are lenient
Noncenter = 1 ; examinee measures float relative to raters, items, time-points
b) The data are the qualitative observations. Their location is specified by Data=. These can be in the same file as the specifications or in separate files. Each data point, response, observation or rating is analyzed as the interaction of one (or more) elements from each of the specified facets.
Example 2: an excerpt of Facets data:
Examinee 23 is rated by rater 4 on item 6 at time-point 2 with a rating of 7
Data=
23, 4, 6, 2, 7
Example 3: Here is a typical data file, after data collection:
123 A 25467 ; Examinee 123 is rated by judge A on items 1-5 with ratings 2, 5, 4, 6, 7
452 C 34245 ; Examinee 452, judge C, ratings 3, 4, 2, 4, 5
312 B 34452 ; Examinee 312, judge B, ratings 3, 4, 4, 5, 2
123 B 24443 ; Examinee 123, judge B, ratings 2, 4, 4, 4, 3
The Facets specification and data file:
Specification
|
Explanation
|
Title = Sample Analysis
|
The title of the analysis appears on every output report.
|
Facets = 3
|
There are 3 facets: examinees, judges, items.
|
Model = ?,?,?,R
|
The facets measurement model is: any examinee (?,) can combine with any rater (?,) and with any item (?,) to produce a rating (R)
|
Noncenter = 1
|
Center the elements all the facets except facet 1. This established the zero-point (local origin) of the measurement frame-of-reference.
|
Positive = 1, 2
|
Facets with "higher score = higher measure" (ability, leniency, easiness) are positive.
Facets with "higher score = lower measure" (disability, severity, difficulty) are negative.
|
Umean = 0, 1
|
The user-defined mean for the measures is zero. The user-defined scaling is 1. The reported measures will be in logits.
|
Labels =
|
The list of labels (identifiers) follows:
|
1,Examinees
123 =
452 =
312 =
*
|
Facet 1 is the Examinees.
123 is the first examinee element.
453 is the second examinee
312 is another examinee
* is the end of Facet 1
|
2,Judges
1 = A
2 = B
3 = C
*
|
Facets 2 is the Judges
A is the label for the first judge, with element number 1.
|
3,Items
1-5
*
|
Facet 3 is the Items.
The item element numbers are 1, 2, 3, 4, 5
|
Data=
|
The data follows, or the data can be in a separate file: Data = filename.txt
|
123, 1, 1_5, 2, 5, 4, 6, 7
|
(comma-separated) Examinee 123 is rated by judge 1 (A) on items 1-5 with ratings 2, 5, 4, 6, 7
|
452 3 1-5 3 4 2 4 5
|
(blank separated) Examinee 452, judge 3 (C), items 1_5, ratings 3, 4, 2, 4, 5
|
312→2→1-5→3→4→4→5→2
|
(→tab separated) Examinee 312, judge2 (B), items 1_5, ratings 3, 4, 4, 5, 2
|
123, B, 1_5, 2, 4, 4, 4, 3
|
Element labels are also allowed in the data. They will be automatically replaced by their element numbers
|
... more data
|
|
Look for a problem like yours in our examples. Then copy-and-edit it!
You can also use the template, Excel, SAS, SPSS, STATA and Facform.
End of Introductory Overview: Now start on your own! Run one of the Example files: how about the Diving Competition?
