Example details
The following example illustrates chaining projects for an omnibus style study where an initial demographics project is run and then a least full quota is used to decide which of the other available projects are appropriate and least full.
The example files and scripts mentioned in this topic are suitable for web interviewing. They are in:
[INSTALL_FOLDER]\IBM\SPSS\DataCollection\7\DDL\Scripts\Interview\CATI\Chaining
The phone_chaining_sample.txt participants file contains demographics (Age and Gender) information that will be used for quota. This demographics information is initialized in the participants file but may also be updated by the demographics interview script. The participants file contains columns for the primary project and each candidate secondary project that are used to track the serial numbers. The fields CurrentProject and QuotaCells are used when a record is restarted or returned. The additional required fields are as follows:
▪Project specific fields: One field for each project including the primary project. The field is named to match the project. This field is initialized to 0 to indicate the project is active and has not been used. Once the project has been run, the specific field will be updated with the serial number. A negative number makes the project unavailable and can be used when the project is closed.
▪CurrentProject: Always contains the name of the project that is in progress and can be used to restart.
▪QuotaCells: Semicolon separated list of the pended quota cells. This list determines the projects to run and their order. When the current project completes, the next project is retrieved from the QuotaCells list and CurrentProject is updated. The QuotaCells list also holds the review order.
Quotas are setup based on the required demographics for the project and <project> >= 0. The example uses a least full quota check to decide which projects are appropriate for the respondent. This quota check is done when the initial project completes. If the chain is exited before all surveys complete then, on restart, the surveys that have not completed are checked for available quota. New quota cells are not calculated if selected cells are no longer open.
The first part of the quota name is set to match the project that should be run if the quota cell is selected. The sample management script uses the quota names to decide which projects are appropriate.
The UNICOM Intelligence Interviewer - Server Admin Quotas activity can be used to setup the quotas as least full quotas.
The phone_chaining_sample.mrs script is a special sample management script that implements the chaining. The script skips the usual quota check that occurs when records are retrieved and instead does a quota check when the sample is returned. The sample script sets up the CurrentProject and QuotaCells sample fields (as described above), and sets the NextProject InterviewProperty to the next project to be run. When the second project returns, the next project in the QuotaCells field is run.
QuotaAutoCommit must be set to False for the project. This ensures that all quotas are not completed immediately when the first project completes. The sample management script will manually rollback or complete quotas for each project (as they complete). If interview quotas are used, then the sample management script will need to be updated to also rollback or manually complete the interview quotas.
The phone_chaining_sample.mrs sample management script does not force each project to use the same serial number. This means that the project specific field is also used to store the project specific serial number. The Serial field contains the serial number for the primary project.
See also