ABS uses operations research to improve household survey operational processes
The Operations Research Unit (ORU) was established within the Statistical Services Branch in July 2005 and evaluates operational processes to answer questions such as: Which strategies result in the highest rate of response for the lowest amount of cost? Where are we expending a lot of effort without receiving a lot in return? Where should we be targeting our effort for the greatest gain?
Household survey related investigations to date, have centred around the Monthly Population Survey (the main component of which is the Labour Force Survey) and selected Special Social Surveys. Household surveys use trained interviewers to collect the information needed via both face to face and telephone interviewing. Interviewers are paid for conducting interviews, the time they spend driving around to conduct face to face interviews, and an allowance for the use of their own car.
The enumeration strategies for these collections vary considerably:
- The Monthly Population survey is conducted every month for a period of eight months for each selected dwelling. A face to face visit is conducted during the first month and in subsequent months interviews are generally conducted over the telephone. The interview length is approximately 12 minutes; and
- Special Social Surveys (SSS) are conducted on a particular topic (for example health, time use, household income and expenditure) and vary from survey to survey in the enumeration period (many are conducted over 3-4 months but some over a whole year to pick up seasonal variations), and interview length (which generally range between 30 - 120 minutes). They are conducted via a face-to-face interview as the length of the interview generally makes telephone interviewing impractical.
Using Operations Research, investigations into selected SSS's have identified a number of patterns that can be used by the Population Survey Operations (PSO) area to improve the effectiveness of the enumeration processes. For a recent SSS for example, it was found that:
- Some households were visited up to 44 times without contact being made. In particular, visiting a household more than 10 times takes 6 per cent of the overall effort in terms of calls, and results in an increase of the response rate by approximately 3 percentage points. Being able to identify a point in the follow up process, after which further gains are expected to be small, will enable PSO to develop more effective strategies for following up outstanding households that do not respond to simple intensification of effort;
- Contact rates increased after 4 p.m., but the majority of first call attempts (77%) were prior to this time. Analysis of second call attempts revealed that the majority of interviewers did vary the time segment in which they attempted to make contact with a household. The first and second call attempt time of day sequence with the lowest contact rate (29%) was pre-noon followed by noon-4 p.m.. This information can be used to schedule household follow-up in a way that accommodates operational requirements (for example, it is mandatory that the first visit is made during daylight hours), but maximises the chances of making contact with the household, particularly on subsequent visits; and
- There was a high proportion of trips involving only one attempted household visit (30%), which increased to 55 per cent in the follow-up stages of enumeration due to workloads having few households assigned to them. This information can be used to develop more effective strategies for follow-up, for example possibly assigning interviewers to concurrent workloads.
Future work will extend the above analysis to the MPS survey, with the aim being to assist PSO in reviewing their overall strategy for follow- up procedures.
For further information, please contact Julie Cole on (03) 9615 7562.