Pet Peeve: Bad Graphs in Major Media

One of my biggest pet peeves in this “data journalism” movement that is afoot are the crappy graphs the pop up on major news sites. The biggest perpetrator is the Wonkblog, a policy blog on the Washington Post. They love using Excel or Google Docs, a byproduct of needing a quick and easy statistical program in order to keep posting content.

Most of the time, Excel of Google charts offend my sensibilities because they are not visually pleasing. Other times, they offend because they are wrong. For instance, in this recent Wonkblog post from February 18, Zachary Goldfarb included a graph from Health Affairs whose y-axis is completely wrong.

wonkblog incorrect chart

An incorrect chart used in a February 18 Wonkblog post. The right y-axis is labeled incorrectly.

I’m pretty sure this graph doesn’t even need the right y-axis, but I don’t know. It’s just sloppy, and I can’t tell if its the fault of the source, Health Affairs, or WaPo because the article it comes from is behind a paywall.

Another instance of an incorrect or misleading graph is in a seemingly innocuous New York Times piece about House of Cards viewership through the lens of Twitter mentions. The Times hired a social analytics company named General Sentiment to analyze the social media impact of the release of House of Cards. Netflix doesn’t release their viewership numbers, so one has to scrape data from Twitter to get some idea. (Side note: Since Netflix is a public company, shouldn’t shareholders force management to show views?)

Looking at the Twitter stats is semi-interesting, but the Excel graphs that General Sentiment generated are lacking for the NYT:

House of Cards twitter analysis

From General Sentiment analysis of tweets generated by the first season of House of Cards during the first 11 days of its release

House of Cards; NYT Twitter Mentions

From General Sentiment analysis of tweets generated by the second season of House of Cards during the first 11 days of its release

  1. The graphs are comparing the same thing – tweets about House of Cards during the first 11 days of the release of seasons one and two – but the x-axes and y-axes are different.
  2. The Season One graph (the blue bars) starts at 10,000 instead of zero.
  3. I can understand using the dates for the y-axis, but using Day 1, Day 2, etc., would probably be better since, like I said before, they are comparing the same thing.
  4. Separating the seasons doesn’t make much sense because, everyone say it, they are comparing the same thing.

Below would be my Excel graphs of the same information in two different ways. I had to ballpark the numerical values. The article did mention totals over the 11 days, and I came close to those.

House of Cards twitter

How I would’ve graphed General Sentiment’s Twitter analysis of House of Cards

House of Cards twitter

Another way to look at General Sentiment’s House of Cards Twitter analysis

It’s great that news outlets are adding more and more graphs to their articles. They are really helpful to understand information that uses numbers. I hope that they pay just a little more attention to clarity, especially if they are relying on an outside source for their visualizations.


About Jay

Health Policy Wonk. Slow long distance runner. Returned Peace Corps Volunteer, Cameroon '06-'08. Former Night owl.
This entry was posted in Current Events and tagged , , , , , , , , , , . Bookmark the permalink.

Leave a Reply

Fill in your details below or click an icon to log in: Logo

You are commenting using your account. Log Out /  Change )

Google+ photo

You are commenting using your Google+ account. Log Out /  Change )

Twitter picture

You are commenting using your Twitter account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )

Connecting to %s