Tom Pepinsky

Category: Indonesia

  • Pie Charts or Bar Charts for Simple Comparisons?

    My post on Islamism and Elections in Southeast Asia got some nice feedback, and I’m convinced that the simple graphics helped. I was surprised, though, at the amount of offline feedback that I got about (1) the fact that I chose pie charts and (2) that I did them in Excel.

    An admission: I spend a huge amount of time working on my graphics. JMP calls it “fiddling” and “tinkering,” but in a positive way. It might have to do with the fact that (like George Costanza) I wanted to be an architect when I was younger, but it turns out that the artistic gene skipped me. I’ve also been influenced by my neighbor, who has taught me some about design over the past couple of years.

    I use Excel to make graphs when I need something fast. But for professional things I’ve been trying to move to R for plotting and graphing—it looks cleaner, and your R script allows anyone to recreate your exact figures (impossible with Excel). After I wrote that quick post last night, I had the off-hand thought that people might not take me seriously at my presentation this Friday because I used Excel. I’m not sure that’s the right conclusion, but I did get feedback that the charts were a bit ugly.

    But that’s a minor issue compared to the pie charts: people have very strong opinions about pie charts! I made a special point to issue a caveat in the last post, but even so, the pie chart haters abound. Here’s Edward Tufte, whose opinions of pie charts seem to have set the tone of the discussion:

    If you want more…just google Tufte pie chart and see what comes up.

    Motivated by my sense that figures created in R would probably look better, I spent some time today creating some new figures using ggplot2. Let’s compare the same information displayed two ways. First, the pie chart:

    indonesia-pies

    Now, the stacked bar chart:

    indonesia-stacked

    For now, just ignore the fact that I couldn’t figure out how to didn’t label the pie slices with their party names. The main thing to focus on is that both convey the exact same information. Comparing the two, I honestly don’t see what’s so bad about the pies. My main point is simply to show that the green has not encroached on the red between 1955 and 2009, and that seems to come through nicely. JMP claims that she likes the pie charts better than the stacked bar charts, although she thinks that I shouldn’t outline the slices in black for different parties within each category, for something like this:

    indonesia-pies2

    I see her point, but I do like the idea of highlighting the number of different parties within each category.

    Anyway, for Friday, I’ll almost certainly go with stacked bar charts, in R. If you’re there, tell me what you think. Oh and also, if you have some thoughts on the substance of the presentation, I’d like to hear those too.

  • Elections and Islamism in Southeast Asia

    Here are two interesting graphics that I’ve made while preparing for a presentation on Islam and elections in Southeast Asia. For both Indonesia and Malaysia, I have taken the results from the first parliamentary election after independence and the most recent parliamentary election, and compared the results for self-identified Islamic parties (green) versus other parties (red in Indonesia, blue in Malaysia). Pie charts are usually bad, but in this case they highlight the qualitative comparisons nicely.

    First, the Indonesia results.
    indonesia votes
    The continuity between the 1955 and 1999 elections has been widely noted, but these results show that 10 years of democracy hasn’t changed much either.

    Now, the Malaysia results.
    malaysia votes
    To my knowledge, this is the first time that the parallels between 1959 and 2013 have been highlighted this way, and also the first time that that continuity has been explicitly compared to Indonesia.

    What’s the point I hope to make? Simply this: the popular conversation about Islam in Southeast Asia holds that Islamism is “on the rise.” These figures complicate that simple conclusion: they highlight that electorally, explicitly Islamist parties are not gaining ground—if anything, they are losing ground. If we are to argue that Islamism is on the rise, we must find a way to explain this. I have some ideas about how we might do that, but suffice it to say these ideas consider the rise of Islamism as a complex phenomenon which is interrelated with a lot of other concomitant social, economic, and political transformations.

    This post also stems from a conversation I had while at a workshop at the American Center for Oriental Research last week. An audience member approached me after my presentation comparing Indonesia to Egypt and Tunisia, and insisted that Indonesia has a lot more Muslims than it used to. The exact quote: “87% Muslim? I don’t think so. Unless something has changed recently.”