Ensuring the Use of Quality Data
The quality of your data, not your choice of tool, determines the value of your visualization. In a 2013 article in the Harvard Business Review titled “When Data Visualization Works — and When It Doesn’t,” author Jim Stikeleather pointed out three elements that affect the efficacy of data:
Data quality: Obviously, if your data is incomplete or full of errors, your data visualization will be useless. But it’s not always easy to determine what data is missing and, therefore, how reliable the predictions you make with it will be. It’s important to pay attention to the quality of your data up front to make sure that your conclusions are usable. Work with your IT department and major stakeholders to determine as much about your data as you can. Find out about its origins and how often it is updated.
Context: Context refers to your ability to draw conclusions from your data. If you don’t understand how the data was sourced, how current it is, and so on, you risk drawing faulty conclusions from it.
Biases: It’s important to acknowledge any biases you have about the data before you create your visualization. Do you want the conclusions to support a pet theory? Are you making the data visualization look a certain way to support your conclusions? You must divest yourself of these notions before you begin.
Regarding biases, when you look at any data visualization, it’s a good idea to ask yourself whether the data was created by someone who may have a stake in a certain outcome. Sometimes, the developer’s bias may be unconscious. Make an agreement with the major stakeholders that the data you use must be certified by IT so that you avoid any bias that might be introduced when the stakeholders themselves provide the data.