The effects of open government data: Some stylised facts

2.1Affordance theory and open government data

Originating with the ecologist James Gibson (1979) and introduced into technology studies by, among others Hutchby (2001), affordance theory has become widely used for analysing and researching the appropriation of digital technology and for explaining IT’s effects without being either technologically or socially deterministic (e.g. Leonardi, 2013; Markus & Silver, 2008; Volkoff & Strong, 2013). Following Hutchby (2001), we understand an affordance as an opportunity for action that may or may not be actualised depending on the capabilities of the actors who interact with a given digital artefact. (Markus & Silver, 2008; Stendal et al., 2016) and goals (Volkoff & Strong, 2013; Zammuto et al., 2007). While an affordance is usually seen as a facilitator that enables a positive aptitude to accomplish an action (Leonardi, 2011; Majchrzak & Markus, 2012; Zammuto et al., 2007), it can also constrain users from completing an action (Hutchby, 2001; Leonardi, 2013; Majchrzak & Markus, 2012; Pozzi et al., 2014).

Applied to OGD, this often means that OGD is perceived as affording increased accountability, transparency and efficiency of government services as well as greater participation and engagement by citizens (Ubaldi, 2013). However, not all published data sets will yield these benefits, either because the data itself does not allow for a certain use or because only specific usergroups may be able to make sense of it. For instance, granting public access to certain government-produced data may increase a public administration’s workload, owing to more requests for supply of, or retractions of information from public records. Thus, we understand affordance as opportunity for action that users can benefit from and constraint as possible obstacle that users may encounter.

2.2Affordance theory

In line with Pozzi et al. (2014), Fig. 2 illustrates our view on affordance theory implying difference phases in the sequence of existence, perception, actualisation and the effects of affordance or constraint.

Figure 2.

Different phases of affordance theory adapted from Pozzi et al. (2014).

The existence phase refers to a situation where OGD creates an opportunity either for action or creates a potential obstacle, even if the intended user pays no attention to it. In order for this data to be used for more goal-oriented actions, a potential user must perceive the affordance or constraint by making a connection between the artefact and its possible use. However, as we know from other theories (e.g. TAM, UTAUT), perceptions and intentions do not always lead to expected behaviours. What is of interest is when a person takes a goal-oriented action based on published information, i.e. actualises an affordance (in affordance theory’s parlance). For value to be measured, goal-oriented actions need to translate into observable empirical evidence. Thus, we differentiate between an affordance’s effect, as empirical evidence that can be measured from OGD use, and constraint induced by an affordance, as the empirical evidence we can observe that restrains a user from realising their usage intention. For instance, poorly documented data sets can constrain a potential user from actually using OGD. It is crucial to understand the distinction between affordance and constraint as well as existence, perception, actualisation and effect, because as we move toward constructing SFs, we will consider only statements that depict empirical evidence stemming from an actualisation of OGD in practice, rather than intentions, opinions or potentials regarding possible effects resulting from the distribution of government-produced data.

3.1The identification of the case studies

To identify the relevant studies that have reported effects relating to OGD, we conducted a systematic literature review in March 2020. Following the Moher et al. (2009) PRISMA strategy (preferred reporting items for systematic reviews and meta-analyses), we searched three bibliographic databases: Web of Science (WoS), Scopus and the AIS eLibrary. We opted for WoS and Scopus because they are considered the most extensive databases on different scientific domains (Chadegani et al., 2013). Furthermore, we searched a less extensive yet more specific database, the AIS eLibrary, to complement our search with more articles from the IS domain.

To identify the most relevant scholarly publications in WoS and Scopus, we used starting conditions and specific terms in the search strings. We restricted the document type to articles written in English with the following exact keywords in the title: open government data. We used the same conditions for AIS eLibrary, except that owing to the fact that there was a lower number of papers in the specific database yet more that were specific to our research subject, we broadened our search by including all document types written in English with the following keywords in the title: open data.33

Figure 3.

The literature search and assessment process to derive stylised facts adapted from Moher et al. (2009).

As shown in Fig. 3, the search yielded 255 records. We added an additional 95 records identified through backward searching within the references section and ended up with a list of 350 articles. We excluded duplicates and screened the articles in two iterations based on reading titles, abstracts and introductions in case the assessment based on titles and abstracts was insufficient to determine a paper’s relevance. We considered papers identified through the case literature relevant when they included some sort of empirical findings relating to OGD use or constraints on OGD use. Thus, we excluded from our analysis conceptual and theoretical papers and papers that reported no empirical findings. We retained 48 papers (40 from databases’ search and 8 from backward searching), from which we excluded 31 (29 from databases’ search and 2 from backward-searching) based on full texts because they were concerned with theory development, methodology or framework development with either no or a poor empirical basis. As a result of this process, we retained 17 articles for the in-depth analysis (11 from the databases’ search and 6 from backward searching in the references section). Table 1 summarises the case literature we used to derive the SFs.

3.2The extraction of statements and the designation of evidence level

Following the stylisation process outlined by Houy et al. (2015), we first extracted original, unmodified statements from the identified case literature.44 While reading the articles, we distilled 394 statements, including excerpts from personal communication (e.g. with employees from public institutions, citizens or other stakeholders) or statements by the authors on the basis of some empirical material (e.g. survey data, case findings). In step two, we categorised these statements, applying the affordance theory lens to statements that point to the existence, perception, actualisation or effects of OGD affordance or constraint. We categorised a statement as existence when a pure informative declaration was given. We coded personal opinions, beliefs or expectations as perception. This was for instance the case when an interviewee expressed their opinion on OGD without proven experience that supports their claims’ truthfulness. We coded personal opinions supported by personal experiences as actualisation. Finally, we coded a statement as effect if it was supported by broader empirical evidence. To reduce subjectivity in the coding procedure, multiple coders assessed the statements separately and discussed inconsistencies until they reached agreement. Of the 394 statements, we coded five as existence, 126 as perception, 90 as actualisation and 173 as effect; we retained only the latter statements for further analysis.

In step three, we defined the distinct evidence levels (EL) (see Table 2) and allocated each of the remaining 173 statements to one EL in order to categorise the claims’ quality and legitimacy. We based the definition of the EL on Gregor’s (2006) classification of theoretical contributions. Of the 173 statements, we categorised 47 into EL1, 89 into EL2, 37 into EL3 and none into EL4.

4.1The aggregation, abstraction and construction of the stylised facts

Each SF describes a common effect relating to OGD, that is backed up by several statements extracted from the 17 studies we identified in our systematic literature review and that we assigned into a specific EL according to each claim’s quality and legitimacy. Furthermore, each SF can be positive, negative and/or inconclusive. A positive effect refers to empirical evidence emerging from OGD-related use and a negative effect (or constraint) points to empirical evidence hindering the completion of an intended use. An effect is considered to be inconclusive when the empirical evidence reports contradictory findings for example where, even if some empirical evidence points to an affordance, there is other empirical evidence that shows that an intended use was also constrained in some way. Table 3 shows the number of statements each SF is composed of as well as the median EL and the variance. We calculated the median EL and the variance considering the statements and their respective ELs and not the sources, given that the number of statements extracted from the case literature varied (i.e. certain cases reported more effects than others). According to Reiter et al. (2013), the total number of statements is a good indicator of the relevance of the empirical evidence supporting a SF, while the median EL points to the rigour of how this evidence was collected.

Figure 4.

The stylised facts embedded in the open government data value creation process (authors’ own representation).

As shown in Fig. 4, we organised our findings along the value creation process, which we split into data production, data distribution and data use (see Ubaldi, 2013).

4.2From the production to the distribution of open government data

OGD production covers the generation, collection, storage, aggregation and processing of the data. OGD production is usually realised by public sector entities in view of producing valuable data for distribution and enabling potential users to make future use of the data. The aim is to make the data that is generated by public administration activities more suitable for the use by third-parties (e.g. through regular updates and data cleansing) (Attard et al., 2016b). OGD production is a prerequisite of OGD distribution and the empirical evidence related to this stage deals mainly with one topic: OGD quality.

The first SF we derived from the literature can be described as follows: poor governance and resistance in public administration negatively affect OGD quality. Many studies reported a negative effect at the institutional level, where there is a lack of knowledge on OGD and especially of the required standards. As highlighted for instance by Yang et al. (2015, p. 603), “…as some agencies do not really understand the meaning of open data, they tend to provide unstructured data such as pictures and articles rather than structured raw data”.

Another example from Zuiderwijk and Janssen (2015, p. 110) stressed that “some datasets were simply too poorly documented and had insufficient metadata to be able to be interpreted correctly”. Hence, according to the statements retained to construct the SF, as a result of a broad lack of understanding of OGD, public administrations are likely to provide data that are unstructured and incomplete rather than structured and complete (Hermanto et al., 2018; Yang et al., 2015) as well as undocumented data sets rather than data sets accompanied by metadata (Janssen et al., 2012; Zuiderwijk & Janssen, 2015). Lack of standards presents an upstream challenge to interpreting and further using the data, because lower-quality data sets prevent potential users from making optimal use of the data, which in turn undermines the effectiveness of the intended use. As Boudreau (2020, p. 10) reported, “community members found that quality problems prevented them from fully exercising their oversight role”. In another example Hermanto et al. (2018, p. 301) report NGO representatives having problems arising from incomplete data sets on which they were heavily reliant. These cases in the literature demonstrate that if the produced OGD is of low quality, it may be of little value for further use because data quality is a precondition of getting better outcomes from OGD use. Thus, to generate added value when distributed, the OGD quality is crucial, because quality enhances the possibilities of use as well as the effects (Safarov et al., 2017).

4.3From the distribution to the use of open government data

Data production is followed by data distribution. To enable further use, OGD must be distributed to potential users. OGD distribution is realised by specialised public and/or private entities in the role of platform intermediaries. OGD distribution is a direct prerequisite of OGD use, and the empirical evidence resulting from this stage deals with two topics: distributing and monitoring OGD.

SF2, which is about data distribution, can be described as follows: poor governance and resistance in public administration negatively affect OGD distribution. According to what is reported in the literature, poor governance and resistance in public administration often relate either to the data providers’ unwillingness to publicise, license or embargo certain data or to the features of OGD (e.g. data that is inappropriate for distribution because it is not anonymised). These barriers restrain OGD distribution and if OGD are not distributed to potential users, it cannot be actualised. Neither can it produce empirical evidence on the effects generated from its use. As noted by Zuiderwijk and Janssen (2015, p. 110), “approximately only eight percent of the agency’s data might be appropriate for publication” owing to legal constraints, political sensitivity or other reasons, such as a need to protect the organisation’s reputation. Such restricted distribution limits the possibilities for potential users. Thus, good OGD distribution is a precondition of obtaining useful outcomes from OGD use.

SF3, which deals with the monitoring of OGD, can be summarised as follows: supervising OGD’s accessibility has an inconclusive effect on the ability to monitor the data sets. According to the evidence gathered, for specific data sets, governments closely observe which stakeholders use which data sets for what reasons in order to keep a certain level of control so as to prevent misuse (e.g. respecting data protection or legal aspects). At the same time, the statements also show that this is not taking place systematically because the supervision of OGD’s accessibility does not provide the ability to properly monitor the data sets. For instance, interviewees in Yang et al. (2015, p. 10) indicated that “government agencies encounter difficulties in knowing who uses open data and what and how data are utilized”. We found that an inability to monitor the data sets prevents governments from systematically knowing where the demands for access to OGD comes from, how the data has been used, or if it was beneficial for users. The elements mentioned above show that monitoring of the data sets is limited. Supervising OGD (i.e. preventing accessibility to specific data sets) contradicts the essence of OGD as free, accessible, usable, modifiable and sharable (Open Knowledge Foundation, 2018). However, setting up mechanisms that gather feedback on OGD use is fruitful to gain users’ feedback about which data sets are valuable to afford full completion of the intended use.

4.4The effects of the use of open government data

As OGD becomes valuable only when used (Attard et al., 2016a; Janssen et al., 2012), OGD use can be considered the final stage of the value creation process. OGD is used by a wide range of users from the public sector (e.g. public administration, citizens) and the private sector (e.g. businesses, entrepreneurial ventures). The empirical evidence resulting from OGD use addresses six topics: knowledge creation, efficiency, cooperation, communication, transparency and accountability and citizen participation and engagement.

SF4 deals with knowledge creation and can be summarised as follows: OGD has a positive effect on knowledge creation in public administration, yet an inconclusive effect regarding entrepreneurial ventures and business innovation. According to the empirical evidence retrieved from our case literature, while the effect is positive for the public sector, the effect is inconclusive for the private sector owing to conflicting evidence. Inspiration or innovative ideas that were previously infeasible because of the inaccessibility of specific data sets now have the possibility of being realised. In the public sector, public administration can compile new ideas and services from OGD and can therefore harness them to (for example) better understand citizens’ needs. For instance, Neuroni et al. (2013, p. 1917) mention a “possibility for public administration to collect new ideas how to turn data into value through the observation of the use of OGD”. In the private sector, while Nam (2015, p. 560) reported that “many interviewees indicated social and commercial values that businesses using public data create”, the empirical evidence retrieved from the case literature also shows that it is rather difficult to be innovative and to create commercial value by means of a common good. For example, Yang et al. (2015) state that when an organisation intends to use some data, it must interact with local entities, and the data sets obtained are not always standardised. Further, as one interviewee in Janssen et al. (2012, p. 263) puts it: “…use is limited to the happy few, these who are educated and have time to explore new business opportunities”.

SF5, which portrays a transaction economics view of OGD, can be described as follows: OGD has a positive effect on efficiency in public administration, yet an inconclusive effect regarding cost reduction. SF5 is transactional, because increased efficiency is cost-related and starts from the premise that value is usually generated by reducing activities’ costs. In the public sector, increased efficiency relates to the better use of the existing resources as a result of more productive ways of working such as increased inter-agency information-sharing. However, although higher efficiency usually leads to financial savings, we found an inconclusive effect regarding cost reduction. The evidence suggests that OGD can also generate additional labour costs by increasing the burden on public administration’s core mission(s). Furthermore, OGD is free (to users), which can reduce income to an administration because the administration charges low fees. As discussed by Parycek et al. (2014, p. 88), “giving away data free of charge leads to some decline in earnings”. According to their analysis, the inconclusive effect comes from the increased efficiency, leading to financial savings by reducing the activities’ costs; on the other hand, free services and a higher burden on public administration create additional costs.

The transformational effects from using OGD on collaboration are summarised in SF6: OGD has a generally positive effect on cooperation through the digital transformation of public administration. SF6 has a transformational dimension because digital transformation is closely linked to innovative processes. According to the collected evidence, the positive effect emerges mainly from increased collaboration, coordination and interaction between the stakeholders in the public sector. The power of the collective data sets is fuelled by the easy accessibility and the ability to re-use the data for different purposes, creating links and interdependencies between the different stakeholders in the public sector Zuiderwijk and Janssen (2014, p. 23) state that: “most interviewees showed that they knew which other civil servants were working on open data policies in the other government organizations, especially at the ministerial level” and that “handling similar types of data makes it easier to collaborate in developing policies”. Hence increased cooperation may lead to more coordinated actions which in turn generate more favourable opportunities in the public sector.

SF7, which is about the effects on communication and information handling patterns, can be summarised as follows: OGD has a generally positive effect on communication in public administration, yet an inconclusive effect regarding the handling of information requests from other stakeholders. SF7 is transactional because increased efficiency usually reduces an action’s costs. OGD’s positive effect on communication in public administration shows that OGD is grounded in public administration through inter-agency information-sharing. Concerning the requests from other stakeholders such as citizens, we found no conclusive effect. For instance, Zuiderwijk and Janssen (2012, p. 88) report that “opening data causes a decrease in information requests”, but Parycek et al. (2014, p. 87) assess that “the release of OGD has no significant influence on the rate of requests from external stakeholders”. Accordingly, it seems that for other stakeholders, OGD did not prevail and that some of the continuing requests for information come from within public administration.

SF8 summarises OGD’s effects on transparency and accountability: OGD has a specifically positive effect on public administration’ transparency and accountability. SF8 has a strategic dimension, because it directly enhances the general perception of public administration. According to the collected evidence, the positive effect comes from the increased availability and visibility of information. The increased transparency enables citizens to monitor public administration and favours their accountability. For instance, Janssen et al. (2012, p. 260) report that the “availability of information about what governments are doing and why is increasingly recognized as an important precondition to the meaningful exercise of democratic accountability and deliberation”. The statements also show that increased transparency helps prevent corruption. Lourenco (2016, p. 32) found evidence that data distributed through a portal is used for a wide range of purposes, including to hold public agents accountable and to prevent possible cases of corruption. Accordingly, the elements mentioned showed that OGD increases access to government information, which in turn reduces information asymmetry, which in turn enhances transparency and accountability.

Finally, SF9 is about effects on citizen participation and engagement, which can be summarised as follows: OGD has a specifically positive effect on citizen participation and engagement. SF9 has a strategic dimension because enhanced participation and engagement improves collaboration between public value-oriented stakeholders. According to the collected evidence, SF8 and SF9 are closely linked, because the increased insights from and visibility of government activities are likely to increase citizen participation and engagement. As discussed by Hossain et al. (2018, p. 673), “actions and decisions taken by government are now more transparent to the citizens and enable them to become more enthusiastic to participate in political activities”. Thus, OGD increases transparency by making available certain information about governmental activities and this creates synergies with citizens. As a result, citizens are given opportunities to gain more insights into governmental activities which make them keener to get involved in policy or decision-making processes.