Joined: Jun 01 2005
|Posted: Wed Jun 08, 2005 5:15 pm Post subject: Record Layout in Test Result Data File (.DAT)
|Record Layout for .DAT file matches the basic Raw Data Export column list:
Rep, Block, Judge, Sample, Order, Attrib 1, Attrib 2,...,Attrib N
where Attrib 1 to Attrib N represent the presentation order of questionnaire attributes excluding comments. Calculated Attributes, if any, appear at the end of the list.
There are never any column labels in the .DAT file.
For the .DAT files created through the Statistical Analysis module, each record element is surrounded by quotation marks (") and separated by commas. There are no blank lines in the file.
Here is an example:
Each row represents the combination of a Rep, Panelist, and Sample Number. In this case there are two attributes associated with each Sample and Rep.
Each of two panelists was presented with two samples. Remembering that the Absolute Sample Number is in the fourth column, and Sample Presentation Order Number is in the fifth column, we see that:
- Panelist 0000000003 saw Sample 1 first, and Sample 2 second.
- Panelist 0000090001 saw Sample 2 first, and Sample 1 second.
Look at Row 1:
Rep = 1
Block = 1
Panelist = "0000000003" (Real Panelist ID)
Sample = 1
X1 = 5
X2 = "." (Missing)
where X1 and X2 represent the 1st and 2nd attributes in the questionnaire that have response values recorded. If an attribute is not seen with a particular sample, its value will be recorded as Missing (".").
Note on Anonymous Panelists:
For Anonymous Panelists only, the Panelist Number is recorded as 11 digits.
The First Digit n, usually a "1", represents the nth test selected to be included in the Statistical Analysis.
When you Combine multiple tests together for analysis, SIMS 2000 remembers the sequence by which each test was selected. It assigns that sequence number to the First Digit of the Anonymous panelist.
The remaining 10 digits represent the Ballot Number assigned to the Anonymous Panelist.
So, for example, a Panelist Number of "10000000012" represents the Anonymous Panelist who completed Ballot 12 in the first selected test.
Since the Statistical Analysis Model often uses both Judge (Panelist Number) and Sample as independent variables, we must insure that Anonymous Panelists always have different ID's regardless of how many tests are combined together.
Using Excel to Edit a Data File
If you need to edit a large .DAT file, it can be difficult to make your changes accurately when you are using a text editor, because each line looks nearly the same. As an alternative you can:
- Export Raw Data to Excel from SIMS 2000.
- Use Excel features such as Sort on the Data menu and Replace on the Edit menu to manipulate the data faster and with improved accuracy.
- On the Excel File menu click Save As.. to save the Excel worksheet to a .CSV (comma delimited) file.
- Delete the original mytest.DAT file that you want to modify.
- Rename the mytest.CSV file as mytest.DAT.
- Edit the mytest.DAT file to eliminate the row of column labels.
- Resubmit the .SAS script or .EXL Excel Export File for analysis.
For further information please refer to the post
Using Excel to Modify a Test Result Data File