|
<< Click to Display Table of Contents >> Navigation: Reference Manual > Files > CATI data files |
CATI (Computer Assisted Telephone Interview) projects have some special requirements for the Sample file and the appointments (Partial) data file.
See also the Quota file for targets.
All records remain on the sample file. Records marked as failed ($ZZZSST/2) never leave the sample file.
Records marked as contacted ($ZZZSST/1) remain on the sample file, but also appear in the partial or main data file.
If a partial record is completed successfully, or is failed, then it moves to the main data file. If another appointment is made then it stays in the partial file.
The sample data file contains a list of contact telephone numbers along with any other sample data that is known about the contact.
The sample data file should be created using the main window [CATI] menu. The original file will usually be a an excel spreadsheet.
The sample data file is actually two files:
The data file that contains the sample information. This is a normal ASC data file.
The sample index file (.qsp), which is created whenever a sample data file is saved.
IMPORTANT: there is a further requirement that the names must end in the word "Sample" with a space or underscore before, for example:
My_Numbers Sample.asc and My_Numbers Sample.qsp
My_Numbers_Sample.asc and My_Numbers_Sample.qsp
This is a QSP file that contains an index to all the records in the sample data file. It used for rapid access to relevant records during CATI data entry.
Both files must have the same name with a different extension (.asc or .qsp).
Whenever the sample data file is altered (other than interviewing) the sample index file must be rebuilt. The program detects the " Sample.asc" ending to the file name and prompts for the index rebuild whenever such a data file is saved.
TIP: the quickest way to tidy a sample file is to use [Data] [Refresh data file]. This must be done when interviewing is not taking place.
You can get a summary report of the sample index file using the main window [CATI] menu.
The QPSMR Telin program uses the sample index file to choose random records to present to the interviewers.
When using a Winsock dialler, records can be retrieved by the telephone number sent in the "Called" command. If the project contains more than one telephone number (TELNUM or TELNUM_1, TELNUM_2, and TELNUM_3) then any of the numbers can be used to retrieve the record for interview. See netmr-download.scot for details of the protocol used.
Records still "live" will be $ZZZSST/B. Records failed will be $ZZZSST/2. Other records will have left the sample and will be $ZZZSST/1.
The partial file contains partially completed interviews (and interviews not yet started) along with appointment details for when contact should be tried again.
This is an ASC data file but every record must be exactly the same length. The length of each record is set in project global, with the extension ASC.
IMPORTANT: The name of the partial file must be the same as the project file (.qdf) with the word " Partial" at the end with a space before, for example:
My_Project.qdf
My_Project Partial.asc
In this way appointment records are associated with the project file, so switching to a new sample or main data file will affect appointments if the project file name remains the same.
Whenever the partial data file is altered (other than interviewing) it must be rebuilt with fixed length records. The program detects the "Partial.asc" ending to the file name and prompts for the fixed length rebuild whenever such a data file is saved.
IMPORTANT: if the record length in project global is changed the partial file must be rebuilt before interviewing restarts.
As interviewing progresses on a project the Partial file can become very large with a lot of wasted space where records have been completed an removed from the file. For this reason the Partial (appointments) data file should be rebuilt periodically on large projects.
Records on the partial file will have $ZZZEND/5.
TIP: the quickest way to tidy a Partial file is to use [Data] [Refresh data file]. This must be done when interviewing is not taking place.
This is where records are stored when they have finished. They may be:
Completed successfully ($ZZZEND/1)
Out of quota ($ZZZEND/2)
Not eligible ($ZZZEND/3)
Aborted before completion ($ZZZEND/4)
Not that only records contacted successfully will be on the main data file. Records still live ($ZZZSST/B) or failed contacts will not have left the sample file.
This is name for the current record for each serial number record.
All records start and remain on the Sample data file. When the interview is started they are marked as ZZZSST/1 on the sample file which means they have left the sample file. After that they will be found in the partial (appointments) file or the main data file when they are no longer to be used, see ZZZEND in ZZZ questions.
So "Current data" is all the records from the partial and main data files plus all records which have not left the sample file (ZZZSST is blank/not set).
NOTE: if interviewing is taking place then some records may be locked and not read from the partial file. These may be listed as "lost" records when the files are scanned.
To get the "Current data" the user is asked for the QIF file which Companion uses to determine the three data files to be scanned.
The current data is requested using [CATI] from the Main window menu