GAP 5: Control of statistical surveys - Guiding Principles
for Good Survey Practice1
Establish the need
- Why is the survey necessary?
- How will the results be used?
- Have alternative data sources been considered, e.g. existing surveys,
administrative data or data collected by other organisations?
- If the survey is to meet EU requirements, does it go beyond these? If
Consider the impact on business, the costs and benefits
- What type and number of respondents are proposed?
- What is the impact on smaller businesses?
- What are the costs to business of providing the data?
- What are the costs to Government?
- Do the benefits outweigh the costs?
Minimise the burden
- What is being done to minimise the burden on business?
- Has the number of questions been kept to the minimum?
- For surveys repeated at regular intervals, could the frequency be reduced?
- Is sampling to be used? Is the sample as small as possible?
- What is being done to help smaller firms?
Draw samples efficiently and fairly
- Is the sampling as efficient as possible?
- Is the interdepartmental business register (IDBR2)
being used in drawing the sample where possible?
- Will the sample distribute the burden fairly, i.e. will similar firms
have an equal chance of being selected?
- Will the sample be rotated where this is sensible?
Consult providers and users and test the survey
- Are potential respondents content with the proposals?
- Is the survey being tested on potential respondents?
- Will there be a pilot survey?
Make it easy, especially for small businesses
- Are electronic methods of data collection possible? Can the data be collected
direct from company records? Can accounting or other software be developed
to help with this?
- Are estimates allowable rather than exact figures?
- Are definitions aligned with those used by business? Have potential respondents
- Are smaller firms excluded or are they sampled as lightly as possible?
- Is there a simpler form for small businesses?
Design a good questionnaire...
- Is the form professionally designed? Is it as short and simple as possible?
- Are tick boxes used where appropriate?
- Are the questions easy to understand, in a logical sequence and in language
relevant to the targeted respondent?
- Are guidance notes easy to follow? Are they brief? Are they in the relevant
place for the question?
- Has the form been tested? Has the survey been piloted?
...and a helpful introductory letter
- Does it make clear why the survey is being conducted and whether it is
voluntary or compulsory?
- Does it explain any benefits to businesses taking part?
- Does it give a confidentiality assurance?
- Is the survey aimed at the right management level within the company?
- Does it make clear what is required of respondents, where possible indicating
the likely time to respond? Has enough time been allowed to return the data?
- Does it provide a contact point for guidance?
- Have potential overlaps with other surveys been considered? And avoided
- Are classifications and definitions similar to those used in other surveys
which collect similar information?
Follow survey control procedures
- Are the results as useful and accessible as possible to business?
- Are the classifications and terminology those preferred by business?
- Will summaries of the data be freely available to providers in return
for their help?
1 These principles were put together by the
Office for National Statistics (ONS). Further information on the National
Statistics approach to compliance sosts is given in the Code
of Practice Protocol on Managing Respondent Load, on the ONS website.
2 The IDBR is widely used for selecting samples of businesses for
regular government surveys, and can help to control compliance burden (e.g.
a small firm which has completed a form for one survey can be guaranteed not
to be included in any others for a given period).
For more information, see our note on IDBR use in HSE.