The most information in business is quantitative information, specifically values that measure performance, highlight opportunities and give insight into future events. While this information is at the core of a successful business, it is often misrepresented in poorly designed graphs because the creators of these visualizations have not been trained in developing effective graphs. Without learning basic graphicacy skills, oneโs decisions about how they will present information is close to useless.
When developing a graph there are six fundamental stages one can go through sequentially to ensure that they are creating a useful visualization. Step 1: Determine your message and identify your data. This first step emphasizes that it is not useful to just take what data you have and turn it into a graph. You first must fully understand the data and what it means, consider your graphs potential audience, and then decide what you want the graph to say. Step 2: Determine if a table, graph, or both are needed to communicate your message. If the data being used is very specific and it needs to be able to compare individual values than a table is more effective. If the message is contained in the shape of the data , then it is better to utilize a graph. Step 3: Decide the best way to encode the data. In this step one must look at the relationship type between the data points and identify if its a nominal comparison, a deviation relationship etc. After deciding what type of relationship exists within the data one can choose the best encoding method such as a bar chart, lines to emphasize the overall shape of the data, or points. Step 4: Determine where to display each variable. For any graph that involves an interval scale it is important to always position the categorical value on the x-axis and the quantitative variable on the y-axis. If bars are going to be used to represent the data it is up to the creators on how they want to orient them vertically with the categorical scale on the x-axis, or horizontally with the categorical values on the y-axis. If the graph has especially long text labels, and if there are many bars, it is a good idea to orient the graph horizontally. In most cases only a single graph is necessary to convey a message, however, some instances do exist where multiple graphs are required. The best way to take care of these situations is to utilize small multiples, which involves a series of small graphs arranged in a way so that they can all be viewed simultaneously. Each graph among the small multiples should be very alike, with consistent scales, and only differing in their categorical variables. Step 5: Determine the best design for the remaining objects. The first part of step five is to determine the range for the quantitative scale. If one is using bars then they must start the scale at zero, however, if one is using lines they must consider if the overall trend or analysis of the values is more important, and either narrow or lengthen the scale accordingly. The scale should also be adjusted so that it extends slightly above the lowest data values, and slightly above the highest. Other factors that must be taken into consideration are whether a legend is necessary and where it should be placed, whether the scales need tick marks, if grid lines are required and where the quantitative scale will be located. Step 6: Determine if particular data should be featured, and if so, how? The last step in the process of developing an effective graph is deciding if there is a piece of data more important than the rest that should be highlighted in the graph. If such data exists one of the best ways to encode it is to use dark or bright colors, which will clearly stand out if the rest of the graph is mostly soft colors. Other ways to highlight important data are placing borders around bars, making lines that should stand out thicker, and making feature points larger or including fill color in them alone.