Canadian Government Executive - Volume 23 - Issue 07

October 2017 // Canadian Government Executive / 19 • Accuracy refers to how well the data de- scribes what is being studied. Unrelated data sources may show similar patterns to what the user expects, but this may represent a bias. If graphing the data shows what you would expect (with the arrow always charting upwards) it is important to consider if what is being mapped actually represents the value you are studying, or if it’s something else entirely. • Timeliness of data relates to the period in which the data was collected and whether it is still relevant to the question being asked. While in some cases data that was collected decades ago may still be relevant, in many cases the age of the data decreases its relevance and quality. When working with technology, health, and many other fast changing fields, the latest data sources can often provide the best insights. Basing assumptions on out- dated data will lead to solutions, which are flawed and out of date. • Accessibility relates to the difficulty as- sociated with obtaining the data, the cost involved, and other factors, such as having the appropriate technologi- cal and structural resources. While bet- ter decisions may be arrived at with the best possible data, this can be prohibi- tively expensive and unrealistic. The best solution is a compromise, which aims to use the best possible data avail- able, given organizational limitations. • Interpretability refers to the notes and other information that is available to understand the context of the data. Without information relating to how the data was collected, the methods uti- lized, and so on, it becomes difficult to understand the data, let alone compare and utilize many data resources to see the “big picture” for analytical purposes. • Coherence involves the way in which the data was collected. Using standard- ized tools and measurements can make data relatable as they use the same basic concepts to underline what the numbers represent. Using data from completely different contexts and cre- ated with disparate methods may make it impossible to relate one set of data to another. While there may appear to be high quality data sources available to support decision- making and evaluation, the appropriate- ness of the data should always be ques- tioned. Knowing how various data sources come together to form a larger picture, and understanding it’s relevance, contributes to the data quality and reliability. There is a real danger in not considering these fac- tors, as the data type, the data collection methodology and its relevance, can signifi- cantly change over time. While data may initially appear to be detailing the same phenomenon over a time span, the mean- ing of what is being collected may change, and the final analysis could end up compar- ing apples and oranges, even though they are all labeled as apples. The solution is to explore and map out available data sources and understand the story the individual elements con- tribute to the big picture. Most data sets come with notes explaining the weight- ing of data, and what methods were used to collect and calculate the results. These notes are a key first step to mak- ing a quality determination, and assess- ing the data as a potential element in the analysis. Doing so will allow for use and selection of relevant data, of the right utility and in the right context. Rather than being a one-time function, ensuring data quality should be an essential ele- ment to every evidence based decision- making process. Big Data � � � � � � � � � � � � �M�a�g�-�C�G�E�-�C�M�Y�K