Help Your Audience Parse the Information, Not Your Graphics
Data visualization: the art of showing multiple points of data in a format that your audience can make sense of it in a reasonable time frame. Despite TV programs like The Librarians or Limitless or further back in time, Numb3rs, showing people parsing large amounts of data in seconds, most of us don’t. We are, however, hardwired to see patterns.
Patterns make learning easier, allow us to group information for easier recall, and allow us to make predictions based on what has already happened. Scientists know that we use different parts of our brains for finding and recognizing patterns than we do when we can perceive no discernable pattern.
Creating a visual representation of quantitative content allows our brains to see patterns that we might not recognize, nor maybe even be able to take in if the data is in raw form. The solar industry is chock-full of data. From solar insolation calculators to PPA rates and risks to MW/GW/TW forecast models, we generate a lot of data. And that doesn’t even cover data generated for dashboards for customers or financial models.
When presenting this data, your job is, in Edward Tufte’s words, “to assist in thinking.” That means using data to help your audience draw their conclusions. Data visualizations should be useful, visually appealing, but never misleading. We’ve gathered nine tips below to help you when working on your own data and presenting it. Note: all of the data used in the visuals below is for illustrative purposes only. It’s made up. Not real.
#1: Do you have fewer than five (5) points of data? Use a table or column, not a graph.
If you have fewer than five points of data show it in a table or column of information, not a graph. It takes the human mind a bit of time to figure out the “code” used to create a graph. What are the units of measurement? X-axis? Y-axis? Time period? Color coding? Percentages? Etc. All of that takes time for the observer to figure out and make sense of what they are viewing. If our goal is to “assist in thinking”, then it would be much better when we have small data sets to allow the viewer to spend that mental capital on taking in the data itself. A short table will allow the viewer to take in the data in a single eye-span (see tip six below).
In this table, a few points of data can be compared quickly without having to interpret the data.
This data shows the same information as the table, using a standard layout from PowerPoint. The observer will have to flit back and forth between the key and the graph, and the data layout makes a false comparison between carbon saved and cars removed.
#2: Label your data. Don’t separate labels into a key or legend.
This idea goes against all of those default Excel and PowerPoint graphics you’ve seen over the years, but after reading this advice, then seeing it in action in various ways, it makes a lot of sense. Effective implementation does take some design patience. Again, this tip goes back to the idea that we are assisting in thinking. If we can take mental burden off the viewer so they can spend energy on reviewing the data versus figuring out the code of the data, then we should. By labeling the data, instead of adding a legend, we are saving the observer mental energy from going back and forth between the key and the data and figuring out that red means 2018 and green means 2017 and yellow means 2019. Instead, we have labeled the data and the viewer can take it in in one group.
Adding data points to the chart allows the viewer to read the numbers if they would like the full picture while not having to guess based on approximate placement compared to the Y-axis numbers.
#3: Consider aspect ratio.
Sometimes, this can be a bit tough. We have “this much” room to show the data, therefore, the X and Y axis need to be “this” big. However, as we noted before, our brains are wired to detect patterns. Taking that a step further, our brains love specific angles, particularly a 45° angle. Size your line charts to closely match a 45° angle if at all possible.
These two graphs show the same graphed numbers. Our eyes will make more realistic judgements of the data from the chart on the left, using approximate 45° angles.
#4: Skip excess colors, boxes, 3D effects, etc.
Just because you can, doesn’t mean you should. Sometimes, an awesome 3D effect adds a needed punch to a great poster for the solar battle of the bands. In data presentation, it adds complexity that forces the observer to work harder at parsing what they are looking at. Unless absolutely necessary, skip the 3D effects, extra boxes, drop shadows, etc. Be judicious in color choices. Keep it simple.
#5: Color choice.
There is a lot of advice on color out there. Salesforce studied color in dashboards and discovered that varying, brighter colors had the strongest impact on data observation. On the other end of the advice spectrum, Edward Tufte, in a recent class on data visualization, noted that light, pastel colors with gradients denoting different data points as one might see on a topographic map, allowed the viewer to take in a lot of data while still being able to easily read overlaid labels. Add to that, the layer of people who have vision impairment and color is not helpful for them, and what you get is a great big mess of advice. Our advice? Know your audience and what makes the most sense when presenting to them. Will they be reading this data on a mobile dashboard? Or viewing it on a large screen in a conference hall? Can you work with your brand colors (always our first go-to)? And finally, be consistent. If you have six graphs that all have a data point of 2019, then use the same color to represent 2019 in all six graphs. This will lighten that cognitive load we discussed in earlier tips.
#6: Consider eye span.
Eye span, or vision span, is the area that we can take in with one sweep of the eye. We will quickly scan an area, then go back to take in more detail if we are interested and if we have the time (i.e. in a presentation). If when presenting data, you break up two comparative charts over two slides, then you lose the advantage of eye span for the viewer to take in that data and make their own leaps in comparison. If possible, group comparative data together or even on the same chart.
#7: Beware of creating bias or distorting data by presentation method.
Remember in tip #4, we talked about skipping the 3D effects and keeping it simple? There are times that using these effects can distort your data, which will create bias in your viewer. That is the last thing we want. Data visualization works when it presents the data, not a point of view. The point of view is created by dialogue around the data. Label X and Y axis, work to show the whole picture (Q1 was crap, but Q2 and Q3 were great? Show Q1 anyway… it will help make informed decisions). Take a look at these examples:
These two pie charts, created using default PowerPoint options, show the same information. The chart on the left distorts the data by using a 3D effect with no reference point, making the slices of pie look almost identical in size. The pie chart to the right give a more accurate view of each slice, and labels the slices, giving them further context.
#8: Comparisons, contrasts, differences, causality.
One of our favorite sayings around here is correlation is not causation. However, as we noted earlier, our brains are wired to look for patterns. Show the patterns if your data has them. Show the comparison (being cognizant of eye span), show where the contrasts and differences are clear. And if you can, show causality. What was the cause of the spike in solar purchases last quarter? Was an incentive ending? These points will enrich your data and the story that will emerge when you add this context.
#9: PowerPoint presentation? Split it up into a handout and slides.
“We will provide the slides after the webinar <presentation, talk, conference, etc.>” When we hear this, we hear: Our slides will have a lot of text and explanatory data on them because we know the observers will be reading them on their own and all of our points will need context. Bluntly, creating a presentation in this way is convenient for the presenter, but very inconvenient for the audience. If our intention is to assist in thinking, then our slides should assist in our talk, not reiterate our talk. Present the data, talk about the data, make your point verbally – not on the screen. This divides the viewers attention between what you, as the presenter are saying, and trying to read the text on the screen. Take the extra time to provide a handout that discusses the points you made in your presentation. Yes, it does take extra time to do this, but it will aid in your presentation immensely. And you will be doing the audience a huge favor by providing them with information in a way that they can learn and process.
And finally, this should go without saying, but we’ll say it: show your data honestly. Be in integrity with your audience.
If you would like to delve more into the world of data visualization, here are a few sources:
- Beautiful Evidence, Edward R. Tufte, 2006 https://www.goodreads.com/book/show/17743.Beautiful_Evidence
- Tableau blog: https://www.tableau.com/about/blog
- Data Visualization – Best Practices and Foundations: https://www.toptal.com/designers/data-visualization/data-visualization-best-practices
Interested in working with Corbae on your next presentation or white paper to present your information in the best way possible? Reach out. We would love to chat with you. Know someone presenting at SPI? Forward this article on.