=Paper=
{{Paper
|id=Vol-1498/HAICTA_2015_paper99
|storemode=property
|title=Open Data in Regions from the Users’ Perspective: an Analytical Study
|pdfUrl=https://ceur-ws.org/Vol-1498/HAICTA_2015_paper99.pdf
|volume=Vol-1498
|dblpUrl=https://dblp.org/rec/conf/haicta/UlmanSMSSB15
}}
==Open Data in Regions from the Users’ Perspective: an Analytical Study==
https://ceur-ws.org/Vol-1498/HAICTA_2015_paper99.pdf
Open Data in Regions from the Users’ Perspective: an
Analytical Study
Miloš Ulman1, Edita Šilerová2, Jan Masner, Michal Stočes2, Pavel Šimek2, Petr
Benda2
1
Department of Information Technology, Faculty of Economic and Management, Czech
University of Life Sciences in Prague, Kamycka 129, 165 21 Prague Suchdol, Czech Republic,
e-mail: ulman@pef.czu.cz
2
Department of Information Technology, Faculty of Economic and Management, Czech
University of Life Sciences in Prague, Kamycka 129, 165 21 Prague Suchdol, Czech Republic
Abstract. The paper is focused on open data in regions from the users’
perspective and presents the perceived usability of various public data sets in
the Czech Republic. There were 265 data sources assessed by respondents
according the format, suggestions of further use, number of views, and
usability for citizens, businesses, officers and other users. Only 14 datasets
from national public agencies were already provided as open data, but none at
the local level. The most frequent formats of data were DOC, HTML, PDF and
XLS. The respondents came with 36 different suggestions for the further use of
data sets. Currently, the citizens cannot see the difference in usability of open
and non-open data unless there are particular applications available.
Nationwide data sets were assessed in usability on average better (1.57) than
local data sets (1.93). The usability of data was evaluated similarly across
observed regions beside those provided by national institutions.
Keywords: Open data, open government data, user’s perspective, region, local
government, usability.
1 Introduction
Open data is the term that echoes from local and world media with growing
frequency. Open data are complete, easily accessible, machine readable, using open
standards (e.g. CSV or XML) and published with an open license (Auer et al, 2007).
The term of open data firstly gained popularity in academia where it denotes effort to
publish academic data under free access in special digital depositories (Murray-Rust,
2008). Now the idea of open data is mainly perceived with political meaning,
especially, due to the launch of government open data portals data.gov and
data.gov.uk in the U.S. and the U.K. (Kassen, 2013) and with the start of initiatives
such as Open Government Partnership. Making public sector information freely
accessible in open formats is also referred as open government data (OGD)
(Kalampokis et al, 2011; Shadbolt et al, 2012).
902
Benefits and advantages of using open data provided by government and public
agencies (or open government data) are promising for the public administration and
for private sector (World Bank, 2014). But most public organizations have no or
limited interaction with data users and are often very selective in communication
(Susha et al, 2015). Some authors propose that open data rather should be a way for
government to interact with citizens (Sieber & Johnson, 2015).
The Czech Republic has committed to implement open government data by
joining the Open Government Partnership in 2011. Since then a number of OGD
activities have been started at different levels of the public sector as well as in
academia and other domains. Although there are some challenges that the Czech
Republic needs to face – like the missing official OGD catalogue, some public sector
bodies have already started to publish open government data (Chlapek et al, 2014).
There are several pioneering organizations that started as first and have showed
already some tangible results. Among these organizations belong national state
agencies (namely the Ministry of Finance, Czech Trade Inspection or Czech
Telecommunication Office), regional administrations (e.g. Vysocina region) and
municipalities (e.g. Prague, Decin, Opava). The national open data catalogue of the
Czech Republic has been recently launched within the main government portal
portal.gov.cz. There are concentrated all datasets from national public institutions
that have opened their data so far. The main resource is the register of public
contracts that provides scanned copies of all public tender contracts with metadata in
XML for automated processing.
Enormous data volumes are generated on daily basis by municipalities and
regional authorities. The demand for public data in open format and the number of
relevant applications are expected to be steadily growing as an effect of the increase
in the level of digital skills and demand for information for example among
inhabitants in rural areas (Vaněk et al, 2011) and small farmers (Kubata et al, 2014).
The extend of open data sources could have a positive impact on areas such as
tourism (Šilerová, 2013). Some authors argue that the potential of the open data
concept can be realized at the local level (Kassen, 2013). Contributions and particular
impacts of open data on rural life should be examined and researched. The
Department of Information Technologies FEM CULS Prague has started cooperation
with Czech Ministry of Agriculture on the research of open data in agriculture and
rural regions.
There are two aims of the paper: firstly, to estimate the potential of open data
sources at the national and regional level, and secondly, to evaluate user’s view of
open data usability in the Czech Republic.
2 Materials and Methods
Particular research questions that were examined in the survey were stated such as:
1. The number of potential data sets identified for opening and the number of
suggestions for the further use of open data was rather higher at local or
regional administrations than at the national level.
903
2. There are differences in estimated usability related to the type of the
municipality and the user group.
3. There is a significant difference in usability of open data and data that are
not open.
4. There are differences in estimated usability related to the size of the
municipality population.
Based on research questions further statistical processing was conducted. The
methodological approach used in the paper is both of qualitative and quantitative
nature and includes literature review, user interviews and descriptive analysis. The
data were processed with association tables, descriptive statistics and tests to
compare differences in values.
3 Results and Discussion
Data were collected through questionnaire form administered to citizens living in
various regions. In total, there were 265 data sets assessed by respondents in terms of
format, suggestions of further use, number of views, and usability for citizens,
businesses, officers and other users. The perceived usability of data sets was ranked
on the scale from 1 (the best) to 5 (the worst). Basic data are summarized in Table 1.
Table 1. Data sets overview from the survey.
Location Data sets Open data Suggestions
Local 216 0 26
National 49 14 10
Total 265 14 36
Source: self-authored, 2015.
The collected data represent 52 municipalities from 9 regions (including the
capital Prague), 8 public institutions (7 schools and 1 school dormitory) and 11 state
bodies in the Czech Republic. The most frequently evaluated types of data sets were:
obligatory information published on public office website (also called as “e-desk”)
(33 %), budget information (33 %), annual report (13 %), newsletter (10 %) and
decision of representatives (10 %). In the survey focused on the openness to
information disclosure among between 395 municipalities (with up to 2,000
inhabitants) in one region of the Czech Republic in 2009 (Bachmann, 2012), it was
found that 71 % of municipalities published electronically minutes from the
municipal council meetings, and 28 % electronic periodicals (newsletter). In the
survey between 400 municipalities across the whole Czech Republic in 2012, there
were 27 % of municipalities publishing resolution from the municipal council
meetings, 17 % resolutions and minutes, and 40 % of municipalities publishing
newsletter. The procurement information was published only in 30 % of cases.
However there is a significant correlation between the size of municipality
population and its information openness (Bachmann & Zubr, 2014).
904
There were 16 various formats of data sets identified such as DOC, HTML, XLS,
PDF, XML, CSV and other. Additional 6 various combinations of certain formats
were used and some datasets were already provided in the form of a web application.
Respondents came with particular ideas of further use of data sets in open format in
36 cases (13.6 %), while 14 (5.3 %) data sets were already provided according to
open data principles. The remarkable finding is that all open data sets were provided
by state administration bodies and none by local authorities. However more
suggestions came for local data sets (26) than for national (10). The most suggestions
for the further use of data were declared with files in PDF (9 suggestions), HTML
(8), XLS (5), HTML combined with PDF (4) and CSV (4). The first research
assumption was confirmed since the availability and the number of inputs for data
sets use is prevailing at regional level.
The second question was explored by conducting a statistical analysis. Regarding
the nature of gathered data differences among mean values were examined by the
analysis of variance and several non-parametric tests. We found that data usability
evaluation does not have normal distribution and samples are of different size.
ANOVA test assumptions such as normality and symmetry of sample distribution
were not fully satisfied, so non-parametric tests were also employed, namely
Kruskal-Wallis, Leven’s test (means) and Welch’s test (Lantz, 2013; Zimmerman,
2011). The statistical hypotheses that were tested are presented in Table 2.
Table 2. Statistical hypotheses related to data sets usability evaluation
Hypotheses
H1 There is no significant difference between the type of format of the data set and its
usability (from the citizen’s perspective).
H2 There is no significant difference between the local and national providers and the
estimated usability (from the citizen’s perspective).
H3 There is no significant difference between regions and the estimated usability (from the
citizen’s perspective).
H4 There is no significant difference between the evaluation of the data set usability and the
user group.
H5 There is no significant difference in usability between data provided as open data and
non-open data.
H6 There is no significant difference in usability between municipalities according to size of
population.
Source: self-authored, 2015.
Basic descriptive statistics and results of hypotheses testing are presented in
following tables (Table 3 – 12).
Firstly, we focused on differences between data format and perceived usability of
data. Based on results in Table 4, we can conclude that there are no significant
differences in the means and the type of format has no influence on the perceived
usability of data. It is needed to add that only three data formats were included
because the other had less than 10 evaluations that might affect reliability of results.
Selected formats (DOC, HTML, PDF, XLS) are also typical for publishing data on
web, however technically, PDF is considered to be one-star data according to the
five-star data concept (Berners-Lee, 2010).
905
Table 3. Formats and usability from the citizen’s perspective - descriptive statistics.
Groups Count Sum Mean Variance SS Std Err
DOC 17 35 2.0588 0.6838 10.9412 0.2284
HTML 81 150 1.8519 0.9528 76.2222 0.1046
PDF 77 144 1.8701 0.7724 58.7013 0.1084
XLS 29 58 2 1.125 31.5 0.1749
Source: self-authored, 2015.
The interesting finding brings the frequency analysis of formats usability
evaluation. Data sources were assessed with 1 or 2 in usability in particular formats:
DOC (64.7 %), HTML (75.9 %), PDF (80.5 %) and XLS (75.9 %). So having data in
one of these formats is likely to be perceived positively.
Table 4. Formats and usability from the citizen’s perspective – differences testing, α=0.05.
ANOVA Kruskal-Wallis Welch’s Levene’s
p-value 0.7777 0.6798 0.7671 0.5844
significant no no no no
Source: self-authored, 2015.
Many examples of successful open data applications are based on data sets
provided by nationwide authorities such as police or national health agency, but we
suppose that there are myriads of opportunities to exploit data from local or regional
resources. The respondents spotted 216 various local data sources (see Table 5).
However, none of local data sets was served in an open data format. There were
some significant differences in average values (and medians according to the
Welch’s test) of usability between national and local data sets (see Table 6). Thus,
the hypothesis no. 2 (H2) has to be rejected.
Table 5. Usability of local vs. national open data from the citizen’s perspective - descriptive
statistics.
Groups Count Sum Mean Variance SS Std Err
Local 216 418 1.9352 0.9260 199.0926 0.0653
National 49 77 1.5714 0.9063 43.5000 0.1372
Source: self-authored, 2015.
The evaluation of the usability was positive (1 or 2) at local resources in 75.5 % of
cases and at national resources in 89.8 % of cases.
Table 6. Usability of local vs. national open data from the citizen’s perspective – differences
testing, α=0.05.
ANOVA Kruskal-Wallis Welch’s Levene’s
p-value 0.0174 0.0049 0.0185 0.8265
significant yes yes yes no
Source: self-authored, 2015.
906
Collected answers were categorized according the municipalities and regions
where data were originated or related. In Table 7, there are four regions that provided
at least 10 scores: whole Czech state administration (CZ), Olomoucký region (M),
Plzeň region (P) and Ústí nad Labem region (U). The differences between regions
were not statistically significant with 95 % probability (see Table 8). The other
regions that were presented in responses, but not included in computation, were:
South Bohemian region, Pardubický region, Vysočina region, Karlovarský region
and Moravia-Silesian region.
Table 7. Regions and usability - descriptive statistics.
Groups Count Sum Mean Variance SS Std Err
CZ 49 77 1.5714 0.9063 43.5 0.1331
M 38 76 2 1.13514 42 0.1512
P 139 255 1.83453 0.80576 111.1942 0.0790
U 11 20 1.81818 0.56364 5.6364 0.2810
Source: self-authored, 2015.
The frequency of positive evaluation (1 or 2) of data sources usability across
regions was distributed such as: national (89.8 %), Moravian-Silesian region (76.3
%), Plzeň region (78.4 %) and Ústí nad Labem region (81.8 %).
Table 8. Regions and usability – differences testing, α=0.05.
ANOVA Kruskal-Wallis Welch’s Levene’s
p-value 0.1836 0.0972 0.2533 0.9122
significant no no no no
Source: self-authored, 2015.
The usability from the perspective of citizen, business and public administration
(PA) staff was examined and summarized in tables 9 and 10. However, the results
were not unanimous and we cannot simply tell whether or not there are any
differences in usability evaluation. From the descriptive statistics (see Table 9), we
can see that the variance and mean values from business users remarkably differ.
ANOVA and Levene’s tests that examine differences in means signalize that there
are no significant differences, however, Kruskal-Wallis and Welch’s test indicate the
opposite (see Table 10). Actually, the responses were given by citizens rather than by
business people or public officers. This fact might have significant impact on data.
To answer the second research question supposing difference in data usability
among regions and user groups, we can conclude that the usability was evaluated
similarly across observed regions and between user groups. On the other hand, the
data sets provided by national institutions were ranked with better score than regional
sources (see Table 5).
907
Table 9. Usability of open data according user groups - descriptive statistics.
Groups Count Sum Mean Variance SS Std Err
Citizen 265 495 1.8679 0.9389 247.8774 0.0640
Business 60 135 2.25 1.8178 107.25 0.1344
PA staff 58 109 1.8793 0.9852 56.1552 0.1367
Source: self-authored, 2015.
The evaluation of usability according user groups ranked 1 or 2 in 78.1 %
(citizen), 68.3 % (business) and 79.3 % (PA staff) of cases.
Table 10. Usability of open data according user groups – differences testing, α=0.05.
ANOVA Kruskal-Wallis Welch’s Levene’s
p-value 0.0351 0.2814 0.1204 0.0021
significant yes no no yes
Source: self-authored, 2015.
The third research question was to reveal the users’ point of view on usability of
data provided in open format and data that are not open yet. The descriptive summary
is presented in Table 11. There were only 11 open data sets evaluated by users
against 254 non-open data sets. Some significant discrepancy in mean values was
confirmed only by Levene’s p-value, but other test results including ANOVA
prevented from rejecting the hypothesis about equality of mean values (see Table
12).
Table 11. Usability of open data and non-open data from the citizen’s perspective -
descriptive statistics.
Groups Count Sum Mean Variance SS Std Err
Open data 11 25 2.2727 2.6182 26.1818 0.2916
Non-open data 254 470 1.8504 0.8688 219.8150 0.0607
Source: self-authored, 2015.
There was also no remarkable difference in the frequency of a positive evaluation
of data usability between open data (63.6 %) and non-open data (78.7 %).
Table 12. Usability of open data and non-open data from the citizen’s perspective –
differences testing, α=0.05.
ANOVA Kruskal-Wallis Welch’s Levene’s
p-value 0.1574 0.7800 0.4102 9.1636E-05
significant no no no yes
Source: self-authored, 2015.
We can conclude that citizens can hardly see any difference between open and
non-open data in terms of the usability. Unless there are particular applications based
908
on open data much interest cannot be expected. The value of open data materializes
only upon its use (Susha et al, 2015).
Table 13. Usability and size of municipality population – differences testing, α=0.05.
ANOVA Kruskal-Wallis Welch’s Levene’s significant
Citizen 0.0292 0.0876 0.0636 0.9554
no
Business 0.1766
0.5432 0.0139 0.0002 no
PA staff 0.3064 0.4517 0.1777 0.0028 no
Source: self-authored, 2015.
The fourth research question and resulting final hypothesis were to examine
differences among municipalities of different size. In the Table 13, we can see
comparison of testing differences in usability between municipalities (without
schools and national institutions) according to size of population. There were four
different categories of the population size: under 2,000 inhabitants, between 2,000
and 4,999, between 5,000 and 9,999, and over 10,000. The categories follow
guidelines of the Czech Statistical Office (CZSO, 2011). Because there was no
sample where all test criteria would be significant, we could not reject the null
hypothesis and we should state that there are no significant differences between mean
values of the usability evaluation. The benefit from using data was not affected by
the size of municipality population from the perspective of respondents. However,
the respondents did not need to be residents in the respective municipalities. The
number of population had some significant effect in the evaluation of municipality
openness to information disclosure (Bachmann, 2012; Bachmann & Zubr, 2014).
4 Conclusion
Based on the survey, it can be concluded that data sets provided by national state
agencies and authorities are more frequently in line with open data principles, while
at the regional level there are still large resources that could be unleashed under open
data concept. Citizens and businesses come in contact rather with local than national
authorities and they could take better advantage of novel applications and services
based on open data if they exist.
The main highlights of the survey findings are summarized under:
• In the sample of 265 data sets, the most frequently published data were:
obligatory information published on public office website (also called as “e-
desk”) (33 %), budget information (33 %), annual report (13 %), newsletter
(10 %) and decision of representatives (10 %);
• Out of 265 data sets, there were 216 data sets provided by local municipalities
or institutions (81.5 %) and 49 by national bodies (18.5 %);
• Nationwide data sets were assessed in usability on average better (1.57) than
local data sets (1.93);
909
• HTML (30.5 %), PDF (29.1 %), XLS (10.9 %) and DOC (6.4 %) are the most
used data formats on municipality and public authority web sites. Their
usability was assessed as positive (score 1 or 2) in at least 64.7 % cases;
• The usability of data was evaluated similarly across observed regions beside
those provided by nationwide institutions;
• The usability of data was assessed from three perspectives: citizen (265
evaluations), business (60) and public administration staff (58);
• The average usability was the best from citizen’s point of view (1.86) and the
worst from the business point of view (2.25), but most of respondents were
rather citizens than business people, which might have impacted results;
• Currently, the citizens cannot see the difference in usability of open and non-
open data unless there are particular applications available;
• The benefit from using data was not affected by the size of municipality
population from the perspective of respondents, but those respondents did not
need to be the actual residents in those municipalities.
The presented results are limited due to the scope of the survey. More details
about opinions of users coming from regions and local communities should be
investigated. Also the willingness of people to use tools based on open data and to
interact online with their local representatives should be examined thoroughly
(Office, 2011; Susha et al, 2015).
Acknowledgments. The results and knowledge included herein have been obtained
owing to support from the Internal grant agency of the Faculty of Economics and
Management, Czech University of Life Sciences in Prague, grant no. 20151039
"Využití otevřených dat v agrárním sektoru" (The use of open data in agricultural
sector).
References
1. Auer, S., Lehmann, J., Bizer, C., Kobilarov, G., Cyganiak, R. & Ives, Z. (2007)
DBpedia: A nucleus for a Web of open data. Translated from English by, 4825
LNCS.
2. Bachmann, P. (2012) Openness to information disclosure: the case of Czech rural
municipalities. AGRIC. ECON. - CZECH, 58(12), 580-589.
3. Bachmann, P. & Zubr, V. (2014) What affects the information provided on the
web? Case of Czech rural municipalities. Acta Universitatis Agriculturae et
Silviculturae Mendelianae Brunensis, 62(6), 1221-1231.
4. Chlapek, D., Kučera, J., Nečaský, M. & Kubáň, M. (2014) Open data and PSI in
the Czech Republic. European Public Sector Information Platform.
5. CZSO (2011) Preliminary results of the Census 2011 Population, 2011. Available
online: https://http://www.czso.cz/csu/sldb/population [Accessed.
910
6. Kalampokis, E., Tarabanis, K. & Tambouris, E. (2011) A classification scheme
for open government data: Towards linking decentralised data. International
Journal of Web Engineering and Technology, 6(3), 266-285.
7. Kassen, M. (2013) A promising phenomenon of open data: A case study of the
Chicago open data project. Government Information Quarterly, 30(4), 508-513.
8. Kubata, K., Tyrychtr, J., Ulman, M. & Vostrovskỳ, V. (2014) Business
informatics and its role in agriculture in the Czech Republic. Agris On-line
Papers in Economics and Informatics, 6(2), 59-66.
9. Lantz, B. (2013) The impact of sample non-normality on ANOVA and alternative
methods. British Journal of Mathematical & Statistical Psychology, 66(2), 224-
244.
10. Murray-Rust, P. (2008) Open data in science. Serials Review, 34(1), 52-64.
11. Office, C. S. (2011) Preliminary results of the Census 2011 Population. Available
online: https://http://www.czso.cz/csu/sldb/population [Accessed.
12. Shadbolt, N., O'Hara, K., Berners-Lee, T., Gibbins, N., Glaser, H., Hall, W. &
Schraefel, M. C. (2012) Linked open government data: Lessons from data.gov.uk.
IEEE Intelligent Systems, 27(3), 16-24.
13. Sieber, R. E. & Johnson, P. A. (2015) Civic open data at a crossroads: Dominant
models and current challenges. Government Information Quarterly.
14. Susha, I., Grönlund, Å. & Janssen, M. (2015) Organizational measures to
stimulate user engagement with open data. Transforming Government: People,
Process and Policy, 9(2), 181-206.
15. Šilerová, E., Maneva, S., Hřebejková, J. (2013) The Importance of Congress
Tourism for Regional Development. Agris on-line Papers in Economics and
Informatics, V(3), 79-86.
16. Vaněk, J., Jarolímek, J. & Vogeltanzová, T. (2011) Information and
Communication Technologies for Regional Development in the Czech Republic –
Broadband Connectivity in Rural Areas. AGRIS on-line Papers in Economics and
Informatics, III(3), 67-76.
17. Zimmerman, D. W. (2011) A simple and effective decision rule for choosing a
significance test to protect against non-normality. British Journal of Mathematical
& Statistical Psychology, 64(3), 388-409.
911