SAS datasets, with their data correctly formatted, can be specified directly to Facets.
Data= SASdata.sas7bdat
Dvalues= is useful for constant values in the data
Facets analyzes integer data in the range 0-255.
Follow the format of SASdata.sas7bdat:
This data set has seven facets: Id, Female, Race, SES, Schtyp, Prog, Items
The five Items are: Read, Write, Math, Science, Socst
There are 11 SAS variables: id, female, race, ses, schtyp, prog, read, write, math, science, socst
The data are dichotomous: are on a rating scale 0-100
Here is SASspecs.txt, the Facets specification file which builds the list of element labels from SASdata.sas7bdat:
Title = "Facets analysis with SAS file: Build list of element labels"
Facets = 7 ; Id, Female, Race, SES, Schtyp, Prog, Items
Models = ?, ?, ?, ?, ?, ?, ?, R100
Positive = 1, 2, 3, 4, 5, 6, 7 ; all positive
Noncentered = 5 ; we want Schtyp to float
Umean = 100, 50 ; the user mean = 100, user scale = 50 units per logit
Yardstick = 0, 5 ; 5 rows per user unit in Table 6: rulers.
Labels=Build
1, Id, D ; a dummy facet for this analysis
*
2, Female
*
3, Race
*
4, SES
*
5, Schtype
*
6, Prog
*
7, Item
1 = Read
2 = Write
3 = Math
4 = Science
5 = Socst
*
Dvalues=*
7, 1-5 ; the data reference for facet 5 is always items 1 to 5
*
Data = SASdata.sas7bdat ; Example dataset from www.ats.ucla.edu
After a Facets analysis, the output file SASspecs.out.txt contains:
Table 2. Data Summary Report.
.....
This list is generated by Labels=Build. ....
Labels=
1, Id, D ; facet 1
1 =
2 =
.....
199 =
200 =
*
2, Female, ; facet 2
0 = ; this is ignored. It is Keepasnull=
1 =
*
3, Race, ; facet 3
1 =
2 =
3 =
4 =
*
4, SES, ; facet 4
1 =
2 =
3 =
*
5, Schtype, ; facet 5
1 =
2 =
*
6, Prog, ; facet 6
1 =
2 =
3 =
*
If this fails to produce the correct list of labels, check that Facets has input the SAS file correctly:
Click on: "Edit" menu, "Edit Data=".
The contents of the SAS file are displayed in tab-separated format:
Copy-and-paste the Labels= into SASspecs.txt and save it with a new name: SASgood.txt
We also want to change the Keepasnull= value to an unused element number, 999.
Title = "Facets analysis with SAS file: Build list of element labels"
Facets = 7 ; Id, Female, Race, SES, Schtyp, Prog, Items
Models = ?, ?, ?, ?, ?, ?, ?, R100
Positive = 1, 2, 3, 4, 5, 6, 7 ; all positive
Noncentered = 5 ; we want Schtyp to float
Umean = 100, 50 ; the user mean = 100, user scale = 50 units per logit
Yardstick = 0, 5 ; 5 rows per user unit in Table 6: rulers.
Keepasnull = 999 ; we want element 0 to be an active element
Labels=
1, Id, D ; facet 1
1 =
2 =
.....
199 =
200 =
*
2, Female, ; facet 2
0 =
1 =
*
3, Race, ; facet 3
1 =
2 =
3 =
4 =
*
4, SES, ; facet 4
1 =
2 =
3 =
*
5, Schtype, ; facet 5
1 =
2 =
*
6, Prog, ; facet 6
1 =
2 =
3 =
*
Dvalues=*
7, 1-5 ; the data reference for facet 5 is always items 1 to 5
*
Data = SASdata.sas7bdat ; Example dataset from www.ats.ucla.edu/stat/
Analysis of SASgood.txt is successful:
Table 6.0 All Facet Vertical "Rulers".
Vertical = (1*,2A,3A,4A,5A,6A,7A,S) Yardstick (columns lines low high extreme)= 0,5,97,103,End
+-----------------------------------------------------------------------------+
|Measr|+Id |+Female|+Race|+SES|+Schtype|+Prog|+Item |Scale|
|-----+------------+-------+-----+----+--------+-----+------------------+-----|
| 103 + + + + + + + +(76) |
| | | | | | | 2 | | --- |
| | | | | | | | | |
| | | | | | | | | 57 |
| | | | | | | | | --- |
| 102 + + + 4 + + + + + |
| | | | | | | | | 56 |
| | | | 2 | | | | | |
| | | | | 3 | | | | --- |
| | | | | | | | | 55 |
| 101 + + + + + + + + |
| | | | | | | | | --- |
| | | | | | | | | |
| | | | | | | | | 54 |
| | | 1 | | | | | Math Write | --- |
| 100 + ********** + + + 2 + + 1 + Socst + |
| | | 0 | | | | | Read | 53 |
| | | | | | | | Science | --- |
| | | | | | | | | |
| | | | | | | | | 52 |
| 99 + + + + + + + + |
| | | | | | | | | --- |
| | | | | 1 | | | | 51 |
| | | | 3 | | | | | |
| | | | 1 | | | | | --- |
| 98 + + + + + + + + 50 |
| | | | | | | | | |
| | | | | | 1 | | | --- |
| | | | | | | | | 49 |
| | | | | | | 3 | | |
| 97 + + + + + 2 + + +(26) |
|-----+------------+-------+-----+----+--------+-----+------------------+-----|
|Measr| * = 20 |+Female|+Race|+SES|+Schtype|+Prog|+Item |Scale|
+-----------------------------------------------------------------------------+
