(First post in my Data Visualization Diary series, where I explore a data set, try to find the interesting stories that the data contains, and practice techniques for using images to communicate the stories that the data has to tell. Through these experiments, I hope to learn what visuals work well, and which fail to communicate in an interesting, efficient, and understandable way. Comments and suggestions welcome)
Today, I picked a data set that journalists everywhere might find interesting. The 6 month data on magazine subscription and single issue sales were released by the Audit Bureau of Circulations. Only the data for the top 25 magazines in each sales category, circulation and single issue, were released online. You can see the data I started with on their website, here.
So, the first thing I noticed, looking at the data, is that subscription sales are much higher than single issue sales. Also, for the most part, different magazines made the top 25 in each category. Only 7 magazines made the top 25 in both categories:
Also, there seems to be a trend in what kinds of magazines sell better in single issue sales: women’s magazines and celebrity-watching magazines.
However, what might be more important to journalists, like myself, hoping to write for magazines, are the changes in sales during the past year. Some magazines are growing their audience, others, shrinking. I calculated the average of all of the % change in sales from 2010 to 2011, and got a surprisingly positive result: Overall increase subscriptions for the top 25 magazines was a positive 20%. Wow! But, it turns out to be one of those misleading statistics. When I averaged % change, 1 magazine with a really high % change can swing the whole group, even if they are a relatively small portion of the total magazines. So, I averaged by change in the total # of issues sold (by subscription) for all the top 25 titles, and I got -0.78%. So, across the board, magazine consumption of the leading titles went down, slightly less than one percent, from 2010 to 2011. The graph below shows why my first calculation was so misleading:
In case you can’t read the titles, that big winner that threw off the average in Game Informer magazine. I should start pitching stories to them, apparently. But, the above graph is pretty hard to read, and most of the data points are so small, that you can barely see the bars coming off the line. So, I think I need a better graph. To make these data easier to understand, I either need to summarize, or show you the important ones. Below is a graph of just the titles showing significant change in subscription sales.
Or, alternatively, let’s say we don’t care bout specific titles. We’d rather have a summary of how the industry as a whole is doing. For this, I can make a simple pie chart:
From this graph we can quickly see that the largest category is green, representing small decrease in sales. Next is red, for small increase in sales. However, it’s obvious that the two negative categories, green and purple, are substantially larger than the red and blue positive categories, so we get the sense that overall, sales are decreasing, but not dramatically. And some titles are doing really well (seriously, Game Informer, do you need any science stories written by someone who has never played a video game? Because I will write them!!)
So, even from this simple data set, there are a lot of patterns to learn. What I want to do, for a final summary graph, is look at the sales trends for different categories of magazines. But, I’m not sure how to quickly make the kind of graph I want. So, I’ll just show you what I’m thinking. I want to take this graph of the top 25 divided by category:
But, then, instead of the percentages representing the % share of the group, as the do in the graph above, I want to give the % average change in sales for each category slice. That way, you can understand approximate size of each category for the size of the slice and the change within that category from the % given. I will work on solving this puzzle. If you are interested, here’s the breakdown by category: