Page 276 - Special Topic Session (STS) - Volume 2
P. 276
STS493 Stéphane D. et al.
The remaining 20% of surveys are planned for migration to the new system
within the next 24 months.
The new e-questionnaire platform is achieving two goals. The first is to
provide respondents with their preferred response mode. The second is cost
savings, since the self-response mode is reducing the hours of interviewing
required. The estimated annual savings from offering an e-questionnaire
option (not including the census) are CAN 2.9 million so far.
Case prioritization and interviewer allocation
Case prioritization was developed to improve sample representativeness by
targeting high-priority surveys or cases that belong to domains with lower
response rates. In some circumstances, case prioritization might be used to
target specific cases for various operational reasons. The objective is to monitor
data collection while it is in progress to identify the cases to prioritize. It is one
of two “adaptive” approaches (the other being responsive collection design),
which use information available before and during collection to adjust the
collection strategy for the remaining in-progress cases.
As part of the recently deployed collection platform, rules to deliver cases
according to the highest priority have been introduced to govern work in
Statistics Canada’s five call centres. These rules have various levels. For example,
at one level, the rule assigns cases to call centre agents so that they work only
on a given survey, or in proportions x, y and z on several given surveys. The call
centre would pay attention to these particular surveys on that day. Next, the
prioritization system targets specific operations, such as non-response follow-
up or refusal conversion, within the priority surveys.
The allocation of interviewer efforts is related to case prioritization.
Research at Statistics Canada had shown that staffing levels were not always
well aligned with the workload sample and expected productivity (Laflamme
(2008a); Laflamme (2008b)). In response, Statistics Canada has optimized
interviewer efforts on cases where they will be more efficient.
Another initiative is to automate the delivery of specific cases, such as those
eligible for responsive collection design.
Responsive collection design
Responsive collection design (RCD) is a technique Statistics Canada has
used in production for all computer-assisted telephone surveys since January
2015, following a series of experiments in previous years.
Using RCD at Statistics Canada resulted in higher response rates and improved
data quality, without increased costs or burden to Canadians. A typical RCD
approach divides the collection operation into phases. The earliest phase
begins the survey with a traditional, randomized allocation of questionnaires
to interviewers. Next, the interviewers are asked to complete certain cases that
265 | I S I W S C 2 0 1 9