Designing and Conducting Business Surveys for Official Statistics, Tbilisi, Georgia, November 2016 Business Survey Planning Haraldsen & Snijkers, EFTA course, Tbilisi, Nov. 2016 1
Designing and Conducting Business Surveys for Official Statistics, Tbilisi, Georgia, November 2016
Business Survey Planning
Haraldsen & Snijkers, EFTA course, Tbilisi, Nov. 2016 1
Remember this: Process map: Business Survey Life Cycle
1. Specifying Target population
Sample frame (obs)
Sample
Indicators/q’stions
Questionnaire Data collection mode
Fieldwork
Survey communication
Analysis & estimation
Dissemination/Archive
Evaluation (PDCA)
Required output
Data cleaning
2. Designing Building, Testing survey components
3. Collection
4. Processing
5. Analysis, dissemination, archiving
6. Evaluation
Information needs
Reporting unit(s) Response
Testing
Constraints: org.&bus.
Haraldsen & Snijkers, EFTA course, Tbilisi, Nov. 2016 2
1. Specifying Target population
Sample frame (obs)
Sample
Indicators/q’stions
Questionnaire Data collection mode
Fieldwork
Survey communication
Analysis & estimation
Dissemination/Archive
Evaluation (PDCA)
Required output
Data cleaning
2. Designing Building, Testing survey components
3. Collection
4. Processing
5. Analysis, dissemination, archiving
6. Evaluation
Information needs
Reporting unit(s) Response
Testing
Constraints: org.&bus.
Haraldsen & Snijkers, EFTA course, Tbilisi, Nov. 2016 3
Suppose, within your office, you are commissioned to do a business survey. Data are needed on a new topic, and those data are not available in secondary sources (like registers), so you need to design and conduct a business survey: What is the actual first thing you will do?
Remember this: Process map: Business Survey Life Cycle
Question:
Key Statement: survey planning
• The survey data are not achieved by itself
• All planning activities are aimed at
producing a pre-agreed level of
quality, at minimum costs and
response burden
• Plan for quality within constraints:
Data quality is not achieved by itself
Quality management
• Identify risks: what can go wrong?
Risk management
Haraldsen & Snijkers, EFTA course, Tbilisi, Nov. 2016 4
A Survey Project Plan is one of the survey components!
Overview & Take aways
• Planning a survey – basics: - What is a project?
- Project management basics
- Source of project failure
• Planning a business surveys within constraints: - Constraints to consider, specific for business surveys
- the planning process
• An example
• Planning and progress tool
Take aways:
• Issues to take into account when planning a business survey
• Detailed planning issues will be discussed with the various
survey components
Haraldsen & Snijkers, EFTA course, Tbilisi, Nov. 2016 5
What is a project?
• A project is a goal-oriented set of activities aimed at
delivering a pre-agreed, concrete and clearly specified
result
• Which has to be realized within constraints: money, time,
resources
• Which is produced by a temporary organization unit,
lead by a project manager
• While facing high levels of uncertainty
• A project has a clear stakeholder/client
Haraldsen & Snijkers, EFTA course, Tbilisi, Nov. 2016 6
Clear beginning and ending
In contrast to an ongoing process, but also applies to recurring surveys (tile structure of linked projects)
Examples of business survey projects
• (Re)designing and conducting a business survey
• (Re)designing a questionnaire
• Introducing a new mode of data collection
• (Re)designing a business survey communication strategy
• (Re)designing a new data collection system
Haraldsen & Snijkers, EFTA course, Tbilisi, Nov. 2016 7
Project management basics (§4.2)
1. Start with the planning the survey as a project, then
comes the “doing” • Plan people, resources, activities, and money in time, to make
sure that all survey components are ready in time, for the desired levels of data quality (e.g. response rates)
• Conduct risk assessment: an important part of the production process is outside the control of a project manager, and put in place mitigating plans and actions
• Deal with the project environment: balance between specific needs, established standards and production constraints
• Make expectations clear from the beginning
Haraldsen & Snijkers, EFTA course, Tbilisi, Nov. 2016 8
Relevant for
planning business surveys
Project management basics (§4.2)
2. Manage the project according to the plan: after the
planning comes the doing
3. Control and monitor the survey project
4. Communicate project progress
5. Improve the quality of processes:
apply the PDCA-cycle: Plan – Do – Evaluate – Adjust
Haraldsen & Snijkers, EFTA course, Tbilisi, Nov. 2016 9
The Project Manager role (§4.2)
• Make things happen: plan, start, and manage the project
• Control and monitor the survey project
• Focus on management tasks
This should be a separate role!
Haraldsen & Snijkers, EFTA course, Tbilisi, Nov. 2016 10
Three sources of project failure:
The project management principles are aimed at minimizing
project failure, caused by:
1. Organizational factors - Determine internal constraints - Conduct risk assessment and put in place mitigating plans and actions
2. Poorly identified customer needs and inadequately
specified project requirements
- Steps in the planning process
3. Poor planning and control
- SMART planning - Using paradata (process data) to monitor
progress and achieved levels of quality (using indicators)
Haraldsen & Snijkers, EFTA course, Tbilisi, Nov. 2016 11
“Quality is established through processes,
systems and procedures” (Biemer & Lyberg, 2003)
So, what does planning a survey involve?
1. Specifying Target population
Sample frame (obs)
Sample
Indicators/q’stions
Questionnaire Data collection mode
Fieldwork
Survey communication
Analysis & estimation
Dissemination/Archive
Evaluation (PDCA)
Required output
Data cleaning
2. Designing Building, Testing survey components
3. Collection
4. Processing
5. Analysis, dissemination, archiving
6. Evaluation
Information needs
Reporting unit(s) Response
Testing
Constraints: org.&bus.
1. Project plan
Haraldsen & Snijkers, EFTA course, Tbilisi, Nov. 2016 12
Planning the survey, includes:
1. Specifying the survey
2. Planning the survey process in detail
3. Specifying required resources: money, people, materials
4. Specifying the timetable
5. Project management issues
6. Risk management issues
Haraldsen & Snijkers, EFTA course, Tbilisi, Nov. 2016 13
Structure of project plan
Step 1: Specifying the survey (§4.3.1)
Haraldsen & Snijkers, EFTA course, Tbilisi, Nov. 2016 14
Step 1: Specifying the survey
1. Specifying Target population
Sample frame (obs)
Sample
Indicators/q’stions
Questionnaire Data collection mode
Fieldwork
Survey communication
Analysis & estimation
Dissemination/Archive
Evaluation (PDCA)
Required output
Data cleaning
2. Designing Building, Testing survey components
3. Collection
4. Processing
5. Analysis, dissemination, archiving
6. Evaluation
Information needs
Reporting unit(s) Response
Testing
Constraints: org.&bus.
Haraldsen & Snijkers, EFTA course, Tbilisi, Nov. 2016 15
Step 1: Specifying the survey (§4.3.1)
• Consult with stakeholder, determine: - Survey objectives, target population and information needs
- Deliverables and desired levels of quality (e.g. response rates)
- Budget, timeframe, and response burden
- Survey specifications
Signed tender
• Consult with survey organization: - Determine constraints
Establish business case: pre-plan go/no-go?
• Appoint project manager
Haraldsen & Snijkers, EFTA course, Tbilisi, Nov. 2016 16
Constraints to take into account
• Survey deliverables: the quality of the results, survey
specifications, delivery date Check if this can be achieved
• People, material resources, money, time Apply project
management principles
• Quality considerations: effects of methodologies on data
quality Have the quality framework in mind
• The business context: response process and response
burden Study the process that is outside of control and
reduce risks upfront
• Survey production environment Check and consider
upfront
Haraldsen & Snijkers, EFTA course, Tbilisi, Nov. 2016 17
Especially relevant when planning
business surveys
Study the business context
What do want to know upfront about the businesses you are
going to study?
• What data are available Questionnaire
• What terminology and definitions are used
• Where are the data (one/many departments)
• Format of the data (spread sheet, other format) & Mode
• Whom to contact Communication strategy
• Arguments for getting survey participation
• Other surveys they get (multi-surveyed businesses)
• When to inform businesses about the survey, so they have enough time to get prepared
• When are the data available Fielding survey & Frequency
Haraldsen & Snijkers, EFTA course, Tbilisi, Nov. 2016 18 Remember: business survey response process
Input for:
Study the business context
Example:
Redesign of Survey on Financial Position of businesses
• 2013-2014: new data model was developed, based on data
requirements
• End of Q1 in 2017: Fielding of new survey
• At the end of 2014: Planning of survey was discussed
- What is needed to reach this deadline?
- Including a feasibility study (business context)
Haraldsen & Snijkers, EFTA course, Tbilisi, Nov. 2016 19
This has to be planned in and carried out at an early stage, in order to reduce risks. Otherwise risks will be pushed downstream.
The survey production environment
(§4.3.2.2)
If working in a survey or statistical organization,
you will need to consider the survey production
environment that you are working in.
Haraldsen & Snijkers, EFTA course, Tbilisi, Nov. 2016 20
Haraldsen & Snijkers, EFTA course, Tbilisi, Nov. 2016 21
Constraints
Elements of a business
survey production
environment:
• These constraints have
to be checked upfront
• They are input for
(planning) the
survey process
• See pp. 139-140
Haraldsen & Snijkers, EFTA course, Tbilisi, Nov. 2016 22
Haraldsen & Snijkers, EFTA course, Tbilisi, Nov. 2016 23
Haraldsen & Snijkers, EFTA course, Tbilisi, Nov. 2016 24
Haraldsen & Snijkers, EFTA course, Tbilisi, Nov. 2016 25
This is the picture for one survey! However:
for business surveys, the designs are not independent of each other, but related!
Haraldsen & Snijkers, EFTA course, Tbilisi, Nov. 2016
The survey production environment
Investigate what other surveys your
businesses get and how these designs are
related:
• Data asked for (overlap in data requests)
• Timing of surveys
• Response Burden (workload) of surveys
26
Take into account that businesses may be multi-surveyed!
Haraldsen & Snijkers, EFTA course, Tbilisi, Nov. 2016 27
Haraldsen & Snijkers, EFTA course, Tbilisi, Nov. 2016 28
Plan for quality within constraints
Identify:
• Optimal survey design
• Trade-offs between quality and constraints (costs and
response burden):
- How do constraints affect the survey design?
- The consequences of these trade-offs regarding quality
• Decisions that need to be taken, and identify the risks for
the various options,
document
these decisions
Haraldsen & Snijkers, EFTA course, Tbilisi, Nov. 2016 29
Step 2: Plan the survey in detail (§4.3.2)
Haraldsen & Snijkers, EFTA course, Tbilisi, Nov. 2016 30
Step 2: Plan the survey in detail
• Plan all stages of the survey: Plan the entire survey process up to and including survey
deliverables and survey evaluation, including how it will be
monitored and managed.
• Write the project plan
Haraldsen & Snijkers, EFTA course, Tbilisi, Nov. 2016 31
Step 2: Project plan (Fig. 4.3)
Results of step 2: project plan, with the following sections: 1. Statement of the survey
- Objectives, deliverables, specifications, constraints
2. The survey in detail
- Work breakdown structure of activities
3. Required resources
- needed human and material resources,
money based on work breakdown structure
4. The timetable
- who is doing what and when core of the plan
5. Project management
- how the survey will be planned, managed, controlled and monitored
6. Risk management
Haraldsen & Snijkers, EFTA course, Tbilisi, Nov. 2016 32
Step 2: Plan the survey in detail (Fig. 4.2)
What the
business survey
project plan
needs to cover
Haraldsen & Snijkers, EFTA course, Tbilisi, Nov. 2016 33
Planned in step 2 (discussed in next sessions)
Desired quality Business context Response burden
Step 2: planning the survey
1. Specifying Target population
Sample frame (obs)
Sample
Indicators/q’stions
Questionnaire Data collection mode
Fieldwork
Survey communication
Analysis & estimation
Dissemination/Archive
Evaluation (PDCA)
Required output
Data cleaning
2. Designing Building, Testing survey components
3. Collection
4. Processing
5. Analysis, dissemination, archiving
6. Evaluation
Information needs
Reporting unit(s) Response
Testing
Constraints: org.&bus.
1. Project plan
Haraldsen & Snijkers, EFTA course, Tbilisi, Nov. 2016 34
Step 2: Planning the survey in detail
Combine the planning for all survey components and stages:
1. List and number all activities in a work breakdown structure
table with activities
start with last activity: delivery of survey results, work backward
2. Determine critical path: no delays afforded
questionnaire design: start ASAP!
3. Prepare a timetable with milestones:
final delivery date of survey results
start and end of fieldwork
start and end of development of survey components
4. Plan the survey on a real-life calendar
5. Communicate the plan to those involved and revise, get sign-off
Remember: the plan is likely to change!
Haraldsen & Snijkers, EFTA course, Tbilisi, Nov. 2016 35
Three rules of thumb
• Do things right from the start, otherwise you are pushing
problems and costs downstream!
• Do not underestimate the time needed for DBT activities:
multiply estimated time by 1.5 or 2
• Plan for quality within constraints: quality is not achieved by
itself use the total survey error framework
Haraldsen & Snijkers, EFTA course, Tbilisi, Nov. 2016 36
Survey planning and Total Survey Framework
Haraldsen & Snijkers, EFTA course, Tbilisi, Nov. 2016 37
To conclude: planning a survey is …
• a complex process: planning and developments efforts rise
with survey complexity
• not a linear process: it is an iterative process, going back and
forth, making amendments to the plan. The project plan is
very likely going to change
• It requires communication skills, ingenuity, and creativity
Haraldsen & Snijkers, EFTA course, Tbilisi, Nov. 2016 38
After the planning comes the “doing”
Finalizing the project plan:
• Discuss with everybody involved
• Get commitment and approval
Getting started:
• Kick-off meeting
• The project is running
• Everybody is expected to do their jobs according to
the plan
Haraldsen & Snijkers, EFTA course, Tbilisi, Nov. 2016 39
Planning and progress tool (Fig.4.7)
To summarize:
A tool to assist the planning process and monitor progress
Progress Status Overview:
• Work breakdown structure for a business survey:
- are all activities planned?
• Overview of milestones:
- to monitor the survey progress: to be ticked off during the
various survey stages
Haraldsen & Snijkers, EFTA course, Tbilisi, Nov. 2016 40
Planning and progress tool (Fig.4.7)
Haraldsen & Snijkers, EFTA course, Tbilisi, Nov. 2016 41