Open Data, Better Practices

6 minute read

In a recent publication in Policy Options, the IOG raised questions about the prospects and shortfalls of the Open Data movement. The idea that making government data freely accessible would empower government and citizens alike is tantalizing, but probably also flawed – or at least incomplete. The core challenge of Open Data is that data has been made “open” and civil society is not using it, or at least not nearly as much as many expected. In turn, the societal benefits expected to come from Open Data have yet to be realized. This is not to say that the proposed benefits of Open Data are not feasible, but it is now clear that they are not inevitable either.

Some might propose that the shortcomings are due to the government having bungled up its approach to Open Data, but this argument is hard to sustain. By many important measures, Canada is at the top tier for data openness, the federal government recently ranking 2^nd worldwide for its efforts. In fact, it seems as though the major shortcoming of the Open Data movement lies not with releases by the public administration, but rather, with the capacity of the wider policy research community to make use of Open Data. Indeed, the research community is befuddled by data and rife with customs and traditions that may squander the potential for public engagement in the policy process and Open Government more generally.

This argument is explored in greater detail in our publication with Policy Options, but what this publication did not cover was what the public service could do to spur improvement. Since the release in Policy Options, the IOG has received outreach asking after solutions and what constructive action might be taken, specifically on the part of government. While there is no silver bullet or quick fix, there are many things government can do to make open data more accessible. One cannot forget that only a small fraction of the policy research ecosystem is directly administered by the public administration and so improvement will require many systemic and long-term interventions. Here are a few best practices government can keep in mind for instigating constructive change and promoting the use of Open Data.

Speaking Data Fluently

The fact is that not everyone “speaks data.” This is especially true for Canada’s policy community which, due to customs surrounding research and a heavily qualitative policy training regimen, is less capable with data than many peer countries. Building a better constituency for Open Data starts by making clear what data and data fluency can offer researchers. These weaknesses can be resolved by changing norms, developing new habits, and ultimately changing how Canada thinks about policy training.

In some jurisdictions, those receiving policy-training are required to submit assignments using analysis generated from Open Data. Indeed, why fabricate artificial data sets for teachable moments where nearly limitless amounts of Open Data are available? Open Data also provides an opportunity for students to contribute solutions to real-world problems. Again, why tackle simulated problems when Open Data allows the energy dedicated to training exercises to be put to good use in the wider world? In both cases, this not only breeds familiarity with data, but also with specific portals where Open Data can be accessed. Indeed, Open Data is not just a matter of occasional use; the use of data is habit-forming.

Infographics are another good way to bridge the gap, especially from a data consumption standpoint, since they make data and their insights easily accessible, regardless of a user’s level of quantitative literacy. Infographics come with several advantages including the ability to condense information and make it easier to circulate, the ability to convey highlights, and the ability to influence a change of perspective on issues. If used effectively, infographics can also help encourage awareness of Open Data and relevant datasets.

From Numbers to Narrative

A major obstacle to using and understanding data is lack of context; seldom do data come with a story that speaks for itself. The most compelling data are situated within a narrative that is easily accessible to a wide audience and range of skill sets. For further reading, there are good articles on how to make a narrative for data here, here and here. The overarching lessons include decluttering the data so as to demonstrate clearly the trend under investigation, to present trends in comparative perspective, to highlight relationships that might not be immediately apparent, and to write with a specific audience in mind.

For a case in point about the accessibility of data, the image below represents roughly 5% of a dataset about global life expectancy, or roughly 2.5% of a combined data set comparing life expectancy with GDP per capita. Most Open Data is presented in a similar fashion, although many of the tables available are even more obscure, with difficult to understand labeling conventions and categories, or explanations of technical jargon that are buried deep in appendices or on other websites. Overall, this is very inaccessible even for those who “speak data” and downright intimidating for novices.

But when these data are compared, contrasted and analysed, it is immediately clear that there is a strong relationship between increasing health (life expectancy) and wealth generation. In other words, these data say that first people get healthy, then they get wealthy. Applying this overarching narrative to the data is widely useful and in turn contributes to the usefulness of the dataset; providing an avenue for exploration. Having established a narrative through which to understand the data, the dataset itself is more widely useful since prospective users know what trends to look for, what might represent exceptions to the rule, and where interesting variations might lie.

Easy Analysis

There are several strong examples of how to make data analysis easier and more accessible. Gapminder is the most widely-celebrated global best practice for promoting accessibility of Open Data and improving ease of data analysis. Gapminder collects data into a free, open and interactive system which is easily accessible to data novices, highly compelling in its presentation of data, and renders accessible an extraordinary amount of data tables.

The World Bank Open Data Portal is another good example, since its landing pages automatically display several trends from its Open Data as a guide to the capabilities of its datasets and to inspire potential avenues of inquiry. StatsCan’sThe Daily is another good example, although a less visual one, as it provides meaningful highlights from new research while linking back to relevant studies and datasets.

All the Right Places

Not all users will search for Open Data in the right places. The fact is that users may not know what data they want. For instance, I may be looking for data on research and development in artificial intelligence, but not know that Canada is a leader in this space. If the relevant data on Canada are available exclusively through Canadian open data portals, they may never get found by those it could benefit.

Open Data should be made as widely available as possible by making sure not only that it is accessible through search engines and the like, but that it can also be made available through, third-party networks. To use one example, the Open Data Network aggregates and collects Open Data from a variety of sources, but many levels of government in Canada do not display their Open Data there (or on competing sites) due to technical issues.

Best Practices

Undoubtedly, the best practices for circulating and promoting Open Data have not yet been conclusively determined. It is up to readers of this and other thought pieces on Open Data accessibility to come up with novel ways of bolstering the use and uptake of their Open Data offerings. Indeed, in the age of big data, exponential computer power, and increasingly affordable artificial intelligence, the hardest limit on the potential for Open Data, is the knowledge and creativity of the user.

(Image Sources in order of appearance: Reddit, StatCan, Visual Capitalist, Various, accumulated by Gapminder)