First of all, every election season I devour Nate Silver’s blog FiveThirtyEight.com, and I loved his book The Signal and the Noise. So it’s with great pleasure, and maybe a little regret, that I get to show the folks at fivethirtyeight a thing or two about data visualization.
I just read the article Dear Mona, I Masturbate More Than Once a Day. Am I Normal? There are usually two ways to answer this question. The self-righteous ‘no, you’re not normal and should stop committing adultery in your heart.’ Or you can give the psychiatric question for a question: ‘what is normal?’ The author (Mona Chalabi) takes the third route, that I love, by answering with data. Below is the chart she uses to say, well, you’re not normal because only 13.5% of men in their 30’s masturbates more than 4x per week.
Technically, of course, she’s accurate. And the data visualization Ms. Chalabi uses is useful because, as a table, I can look up the exact % that applies to my group. But with added bars, I can also do some basic comparisons between frequencies within my age group (looking down). What this visual does poorly is allow us to compare frequencies going across (for example, how does frequency of masturbation evolve over our lifetimes?).
Below is a visual that attempts to fit all the data together, giving us a sense of proportion within each age group. This makes it somewhat easier to see how masturbation frequency changes over our lifetimes (especially for the most and least frequent groups – dark red/dark blue). Using graded colors allows the questioner to move from left to right with an intuitive understanding of what he is looking at.
Of course I think we can do even better. The visual below allows us to better compare the evolution of masturbation frequencies. Bars are great for comparing quantities but, with five frequencies, we can’t start all the bars at 0% unless we go back to Mona’s initial graph and lose context. Not only that, but because the distribution of frequency by age group is a whole pattern, the connected lines give us a better holistic impression of those patterns.
All these graphs conceptually divide men and women from one another before looking at patterns. In the graph above, at least we’re using the same scale, so it’s more possible to compare men vs. women. But if we wanted to start with how masturbation habits evolve between the sexes, then take a look at another visual:
With this graph we can see how men and women compare across different age groups. Interestingly, the curves for men are flatter the younger they are, meaning that young men are just as likely to masturbate more than 4x per week as not at all. The more we age, the more the men’s curve starts to match that of women. At 70+, where men fit into masturbation frequency groups isn’t terribly different from women. As a side note, there’s an interesting pattern for women in their late 20’s and 40’s in that the “not in past year” % hooks downwards.
There are a lot of ways to visualize even the most simple data sets. The author of the fivethirtyeight.com story chose a table with some added bars to allow us to look up the questioner’s cohort easily. By using other visualization techniques and labeling certain data points, we can do all that in addition to giving readers more context into the all the issues at hand.