It has been a little while since I last posted, work, life, house and all that– or as my students would say “Adulting”. Anyway recently I came upon a technology issue (well multiple, but I will limit the note to just one), that seems to be addressed or even discussed in only a limited fashion. The issue is about how we construct visuals to explain/show/display our data. Not so much in how we choose data, but in how we actually make them. I’m talking this time about the tools.
For a recent article, I was charged with doing the data wrangling (ok, so maybe I volunteered). Anyhoo, the data turned out to be best represented by a series of line charts depicting a trend over time for various items. My thought was to show the charts stacked one above the other with the same x-axis and y-axis. This would enable the viewer to see changes and compare them among each other relatively easily. No I know I have seen a chart like this before, possibly many times. So much so that I immediately thought well, some tool must produce this kind of chart. After all, it’s not rocket science. Identify the x-axis range, identify the y-axis range. Select the data to be displayed in multiple stacked and/or sequenced graphs. Obviously, more options than the normal “what color is your line?”, might be how close you want to graphs to be stacked, whether they share the same title or multiple, what order, verticle or horizontal stacking, etc.
As my mind pleasantly settled into these ideas, I just knew where these lovely visualizations were being created. I had just secured a coveted piece of software that specializes in dataviz. (I say secured because as an educator, getting it free doesn’t count as a purchase.) This software had to be it! So I jump right in there.
Now, I’m not the brightest bulb in the cabinet, nor am I the dimmest. I futzed around with the software for hours, barely getting it to make one chart, let alone 5 (what does that say for the usability of the software, well that is another blog post I’m afraid). After an inordinate amount of time, I jumped onto google to find my answers. Well, it turns out it can make this 5 stacked line chart, sorta. AND, after the bad experience, I had with using the software, I tossed out the idea and moved to Excel, made the charts, copied the charts, made a table in Word, pasted them into a table in Word, and voila 5 stacked charts! That took 15min (see Figure 1)
My story doesn’t end there. The following month, I was faced with a similar issue. I had an article it needed to make complex data easy to understand. The reviewers (and let’s face it the authors, too) were struggling with understanding the data. Our questions on the survey were complex, and the answers back were hard to show. We started with line graphs and bar charts and a couple of pies to represent data collected with Likert scales, choose the top three ratings, and other more complex questions. As time went by the article was accepted on R&R and I knew a large part of the issue was the incomprehensible data. So, like the stacked line graph idea, I had recently seen a serial donut chart. It was remarkably clear in the data it showed. Not only that, this report had a number of ways to show complex data. I was sure if we could use these techniques to show our data, it would be far more clear.
I turned to the software, as I had in the past. AND this time I knew to go directly to google to make sure it was a workable solution. Well, yes and no. It had the same, you have to “hack it to make it work” answer. So, I jumped back to Excel and made it work there (Figure 2).
So, I am writing to say. How long will it be before we have software that actually does dataviz? I have seen some amazing examples of data displayed using complex tools. Sure, people can program anything to work, we can through Illustrator at a problem and fix it. But when will there be software that does simple stacking and series graphs?
When will I have my flying cars?