Work in progress paper "Heaven for grumblers or the road to better public services? A study of critical reports from a Swiss citizen co-production system"

1. Heaven for grumblers or the road to better
public services? A study of critical reports
from a Swiss citizen co-production system
Presentation at the XXI. International Research Society for Public
Management (IRSPM) Annual Conference in Budapest (Hungary),
Corvinus University, April 19-21, 2017
Matthias Stürmer, Oliver Neumann, Tim Loosli
Research Center for Digital Sustainability
Institute of Information Systems
University of Bern | www.digitale-nachhaltigkeit.unibe.ch

2. UK platform FixMyStreet.com
2

3. Open source project FixMyStreet.org
3

4. Other cities using FixMyStreet.org
4

5. Züri wie neu («Zurich as good as new»)
5

6. Example report
6

7. Example report
7

8. Example report
8

9. 9

10. Research questions
1. How can the value of reports by a
citizen-sourcing system such as
FixMyStreet be measured?
2. What is the role of an information
system regarding the value of such
reports?
10

11. Literature review
> Involving citizens in public service
— Citizen co-production (Pestoff 2008, Johnston and Hansen 2011)
— Citizen-sourcing (Linders 2012, Nam 2012, Schmidthuber and Hilgers 2017)
— Open government collaboration (Lee and Kwak 2012)
> Measuring quality in crowdsourcing initiatives
— Wikipedia (Kittur and Kraut 2008)
— OpenStreetMap (Haklay 2010)
> Measuring the value of reports
— Triage in war medicine (Blagg 2004, Falzone et al. 2017)
— Assessing programming errors (Peng and Wallace 1993)
> Information systems (IS)
— IS success model (DeLone and McLean 1992, 2003)
11

12. DeLone and McLean IS Success Model
12
Delone, William H., and Ephraim R. McLean. "The DeLone and McLean model of information systems success:
a ten-year update." Journal of management information systems 19.4 (2003): 9-30.

13. Measuring the value of a report
Rating Severity Range Quality
1 No threat Very few people affected Insufficient quality
2 Little threat Few people affected Low quality
3 Intermediary threat
Intermediary number of
people affected
Satisfactory quality
4 Substantial threat Many people affected High quality
5 Serious threat
Very large number of
people affected
Maximum quality
13

14. Measuring the value of a report
Rating Severity Range Quality
1 No threat Very few people affected Insufficient quality
2 Little threat Few people affected Low quality
3 Intermediary threat
Intermediary number of
people affected
Satisfactory quality
4 Substantial threat Many people affected High quality
5 Serious threat
Very large number of
people affected
Maximum quality
14

15. Measuring the value of a report
Rating Severity Range Quality
1 No threat Very few people affected Insufficient quality
2 Little threat Few people affected Low quality
3 Intermediary threat
Intermediary number of
people affected
Satisfactory quality
4 Substantial threat Many people affected High quality
5 Serious threat
Very large number of
people affected
Maximum quality
15

16. Measuring the value of a report
Rating Severity Range Quality
1 No threat Very few people affected Insufficient quality
2 Little threat Few people affected Low quality
3 Medium threat
Medium number of
people affected
Satisfactory quality
4 Substantial threat Many people affected High quality
5 Serious threat
Very large number of
people affected
Maximum quality
16

17. 17

18. 18
Student rating
Severity: 4
Range: 3.5
Quality: 4

19. 19
Student rating
Severity: 1
Range: 1
Quality: 5

20. 20
Student rating
Severity: 4
Range: 5
Quality: 4

21. Descriptive statistics of the data set
Description Value
Interval of reports From 31 May 2013 until 13 June 2016
Total number of reports 7501 reports
Distinct citizens creating reports 2658 persons
Sample of reports for manual coding 990 reports
Distinct citizens of the sample reports 400 persons
Average severity 3.09 out of 5
Average range 2.91 out of 5
Average quality 2.68 out of 5
21

22. Hypothesis
22
Linear OLS Regression Models
Model 1:
Severity
Model 2:
Range
Model 3:
Quality
a) System quality 0.088401 -0.054025 0.167109 **
b) Information quality -0.273306 * -0.194604 -0.145850
c) Service quality 0.073678 -0.309793 ** -0.052500
d) User satisfaction 0.115979 0.440310 *** 0.073312

23. Control variables (excerpt)
23
Linear OLS Regression Models
Model 1:
Severity
Model 2:
Range
Model 3:
Quality
Sex (female) 0.092765 -0.062642 -0.110503
Age (in years) -0.003669 -0.003053 -0.006557 *
Living in Zürich -0.057820 -0.080970 -0.075354
Education level -0.043854 -0.026397 0.006815
Currently not employed -0.010545 -0.277074 -0.407320 *
Employed in NPO sector -0.041190 -0.502277 * 0.271671

24. Discussion of preliminary results
1. Severity of reports is negatively related to information quality: The
better citizens understand the information and trust the quality of “Züri wie
neu”, the less severe are the reports they submit.
2. Range of reports is negatively related to service quality:
Citizens who are happy with the service provided by the city of Zürich, are
more likely to report issues that concern only a few people.
3. Range of reports is positively related to user satisfaction:
People who are convinced that the city administration efficiently processes
the reports and acts according to the reported issues, these citizens are
more likely to report problems relevant to many others.
4. Quality of reports is positively related to system quality:
If someone is content with the tool, he or she is more likely to create a high
quality report.
24

25. Next steps with our paper
> Elaborate theory on co-producton, citizen-sourcing, and IS success model
> Deepen theoretical foundation of the value of reports
> Clarify coding schema of severity, range, and quality
> Increase number of manual report codings
> Improve statistical models regarding control variables etc.
> Verify the findings within the data
25
What else should
we improve in
your view?