CATI Allocation (Theme)


To be able to perform CATI interviews sample elements have to be linked to CATI interviewers at some point, so that the interviewers can call and interview them by telephone. In order to be able to do this, several steps have to be taken. An important one is that interviewers have to be scheduled in a timetable, taking various conditions into account, such as days and/or parts of the day where interviewers are not able to work, the surveys for which the interviewers are trained and for which they can be deployed, who has to be called (in our case: businesses) and the time of day when this preferably should be done in order to increase the response rate. If the interviews are conducted from a call room, say at the premises of a national statistical office, its maximum capacity has to be taken into account (if the number of interviewers that can be employed simultaneously exceeds this capacity). The interviewers present in the call room (at a certain day, part of day (DPoD) combination) have to be fed with telephone numbers of sample elements to be contacted, as they were drawn into respective samples, so the interviewers can call them for interviews. The most difficult problem is the allocation of interviewers to DPoDs and surveys. This can be done by hand, but (preferably) by using optimisation models. It is possible to take deadlines for surveys into account in these models, or (salary) costs of the interviewer corps that should be adequate for its task, but not bigger than necessary. When these allocations have been made (and the interviewers know when they should come to work and for what surveys), they have to be ‘fed’ telephone numbers. We assume that the interviewers receive such telephone numbers ‘on their demand’, whenever they have indicated to be ready for a new interview.

A special issue is about the updating of an allocation that has been determined for a planning period. This is necessary as things change. Time flows and plan and reality tend to diverge more and more from each other as time passes. Interviewers get ill, go on vacation, have to take leave, quit their job, new ones get hired and trained. Also as time passes, new surveys come into sight that also have to be carried out. All these changes require a schedule to be updated, very regularly.

It should be noted that instead of a single method, in reality there is actually a complex of models, that are rather strongly related and that differ by using certain alternative constraints (or discarding such constraints), or certain objective functions that are different (but with the same set of constraints).


To read the entire document, please access the pdf file (link under "Related Documents" on the right-hand-side of this page).


Your feedback is appreciated. Please send your remarks, suggestions for improvement, etc. to memobust@cbs.nl.