Perceived Response Burden

1
Perceived Response Burden
Dan Hedlin Helen Wahlström Helena
Bäckström Johan Erikson
2
Perceived Response Burden

• Partly funded by Eurostat (ESTAT 200292501005)
• Partners Statistics Sweden, Statistics Norway
  and the UK Office for National Statistics

3
Background

• LEG recommendation no.5
• NSIs should strive to improve their relationships
  with data suppliers
• Research should be conducted on- suppliers
  perceptions- how to enhance their awareness of
  role of statistics

4
Project Output

• includes
• Model for Total Business Survey Burden a
  holistic view on response burdenSee Jones et al.
  (2005)
• Understanding of root causes of burden
• Two question sets, PRB-questions

5
Measuring Perceived Burden

• 10 questions
• Separate data collection from completion of
  questionnaire
• Use for own business and use for society
• Questions refer to one particular survey

6
Reduce burden by enhancing awareness?

• Many respondents do not appreciate worthof
  statistics
• Would their perceptions of burden be altered?
• Cf Gower (1994), Tomaskovic-Devey, Leiter and
  Thompson (1994), Willimack et al. (2002) and
  Haraldsen (2004)

7
Burden Two-sided

• Rewards, e.g.
• Feed-back of survey results (egoistic)
• Help Sweden/society (social appeal)
• Respondents personal interest in the topic or
  survey (personal)

8
Embedded Experiment

• Structural Business Statistics Survey (SBS)
• 3,677 businesses in the experiment
• Grouped in four blocks (by number of employees)

9
Embedded Experiment contd

• Randomised (within block) to one of the following
  treatments
• No special enclosure
• The enclosure How the data are used
• The enclosure Business ratios
• Both enclosures

10
Research Questions defined in analysis plan

• Can we influence respondents perceptions of the
  survey by enclosing supplementary information
  (either Business ratios or How the data are
  used) with the cover letter? Question 1a and
  3a
• Can we influence respondents appreciation of
• their own (potential) use of the statistics?
  Question 7
• the importance of the statistics to society?
  Question 8

11
Analysis

• Proportional odds models cumulative
  probabilities modelled
• Treatment and block are factors
• No inclusion probabilities involved in this
  analysis

12
Proportional odds model
13
Proportional odds model
14
Results

• Response rate (Feb 2005) 68 in SBS and 54 in
  experiment (no matter treatment)
• Statistically significant result the enclosure
  Business ratios enhances appreciation of use of
  statistics for own business

15
Useful to society?
Green enclosure How the data are used Black
not this enclosure
16
What does this mean?

• No panacea
• Many actions required (cf nonresponse)
• Actual burden not affected
• Any connection between enclosures and burden
  established by respondents not pointed out by
  us
• If perceived useful, then thats their conclusion

17
Continuous Improvement

• A tool for measuring perceived response burden
• Potential nonresponse bias is a concern