Cost-Efficient Framework for Data Collection for CATI Surveys

1
Cost-Efficient Framework for Data Collection for
CATI Surveys

• Social Surveys Collection Research
• Steering Committee (SSCRSC)
• François Laflamme
• December 2010

2
Past and current research

• Paradata research has mainly directed its effort
  in improving the current data collection process
  and practices
• Paradata research has also suggested that the
  interviewer staffing levels are not always
  optimally allocated with respect to the workload
  sample and the expected productivity

3
Challenges and objective

• One of the main challenges is to collect
  cost-effective data while maintaining a high
  level of quality
• According to the forecasted budget reductions,
  the pressure has increased and will continue to
  grow
• The objective of the project is to propose a
  cost-efficient framework for data collection for
  CATI surveys

4
Cost-efficient framework

• Five 5 dimensions
• Metrics used for costing and budgeting surveys
• Resources allocation within surveys
• Resources allocation between surveys
• Collection process and practices
• Operational constraints

5
1. Metrics used for costing and budgeting surveys

• Should be based on objective, measurable and
  comprehensible assumptions
• Should be monitored throughout the survey
  collection
• Budget too high - low productivity
• Budget too low - more difficult to reach targets
• Budget often based on previous survey cycles or
  comparable surveys
• But it is exactly what we try to improve (e.g.
  low TPU)

6
2 3. Resources allocation within and between
surveys

• Within survey
• Interviewer staffing levels should consider
  sample workload and collection progress in taking
  advantage of the entire collection period
• Survey productivity decreases for all CATI
  surveys during collection
• Between surveys
• Many concurrent surveys compete for data
  collection resources

7
Resources allocation within and between surveys

• What is the best strategy within survey?

8
Resources allocation within and between surveys

• The main idea is to try to maintain the survey
  productivity for all surveys for a given RO (red
  line) between reasonable boundaries around the
  average productivity
• How?

9
Resources allocation within and between surveys

• The strategy is to distribute system time for a
  given period (day or week) in function of the
  inverse of the proportion of the budget spent
  (inversely correlated to productivity) to
  maintain productivity at RO level

10
Resources allocation within and between surveys

• In practice
• Need to get consistent budget figures for all
  surveys
• Need to get daily or weekly capacity system time
• Need to consider the active collection period of
  each survey
• Need to take into account operational constraints
• Is it feasible? Can it be operationalized and
  automated?
• The objective is to produce guidelines for RO to
  help them allocate collection resources only at
  the aggregate level.

11
4. Collection process and practices

• Focus of the past research
• Some examples
• Best time to call
• Call scheduler features
• Cap on calls, customized time slices, Z-groups
  etc
• Increased the proportion of evening shifts versus
  day shifts to improve contact rate and
  productivity
• Active management
• Responsive collection design
• No significant efficiency if budget and resources
  allocation remain the same.

12
5. Operational constraints

• RO capacity
• Varies by day (interviewer availability)
• LFS week
• Holidays and special event
• Average productivity
• Varies by RO and survey
• Interviewer schedule in advance
• Interviewer training across surveys
• Etc

13
Comments

• 3 Separate projects
• Costing metrics
• Resources allocation within and between surveys
• Collection process and practices
• All need to take into account operational
  constraints
• All dimensions have to be considered
  simultaneously to succeed in improving the
  cost-efficiency of data collection.

14
For more information, please contact Pour plus
dinformation, veuillez contacter
François Laflamme francois.laflamme_at_statcan.gc.ca