Martin De Saulles
and similar data. These were considered to be aimed at increasing public accountability. Data sets which included mapping information, transport statistics and other data that could not obviously be used to hold public bodies to account were classed under service delivery. A score of 1 was given against each of the two dimensions being measured giving a maximum score of 2 for each data set. For example, if a data set was presented as an Excel file, was released under the Open Government Licence and contained details of public expenditures it scored 1 as adaptable and 1 for public accountability and zero for the other variables. On the other hand, if it was a PDF document which outlined milk prices in Wales then it would receive 1 on the inert scale, 1 for service delivery and zero for the other 2 variables. 5 of the data sets were excluded from the final result as it was not obvious what the primary purpose of the data was intended to be. Once the data had been collated it was presented as a 4 spoke radar chart, shown in Figure 1 which is based on Yu and Robinson’ s own representation of their framework.
Figure 1: Summary results
6. The results
It is important to remember that the sample being analysed comprised 100 data sets meaning that any conclusions drawn should be tentative. However, the results show a definite bias towards the data having a service delivery bias at the expense of public accountability. The extent to which the data is adaptable for reuse is less distinct with only a slight bias towards adaptability.
Based on this analysis it might be argued that the data. gov. uk initiative has so far produced data which is less aimed at making the workings of government and public bodies more transparent and more focused on improving the delivery of public services. There also appears to be a lot of work required if the data it is making available is fit for the purpose of re‐use by third parties wishing to manipulate it as almost half of the data analysed was inert. It should also be noted that most of the data in the sample considered to be adaptable were presented as Excel sheets and CSV files which, although manipulable, do not conform to the same machine‐readable formats which can be found in XML files or well‐documented application programming interfaces( APIs). If this more rigid definition of adaptable had been applied to the analysis then the vast majority of the data sets would have been considered inert. Dead links were also an issue for 5 of the data sets where a publisher had uploaded a record to the data. gov. uk service that either had an invalid URL linking to the data or had changed the URL at a later date and not updated the record. For anyone attempting to build a data service on the basis of such data this could be a major problem.
7. Conclusions
As already mentioned, this research presents a first step in testing the value of Yu and Robinson’ s framework for evaluating government data. A number of judgements were made during the research process to decide where individual data sets from the data. gov. uk initiative sat on the dimensions of the framework laid out by
164