Health Media Collaboratory

Health Media Collaboratory: Social data for public health insight


Tableau: Can you describe a little bit about what you do at the Health Media Collaboratory?
Glen Szczypka, Deputy Director, Health Media Collaboratory: Our mission is data for public good. Over the last ten years, with the advent of the Internet, the advent of social media, we now have screens in front of us at all times.

We get inundated with these data messages. And those messages can lead us to making really bad health choices. So we want to study that data, harness that data, and use it to make people healthier.

Tableau: If I’m new to analyzing social data, what should I keep in mind? Any tips for working with social data in Tableau?
Glen: The first thing you need to realize is social data is dirty data. And just because you use a key word, you can't assume that your question that you're asking is going to be contained in a tweet.

You need to know that the tweet you're looking at is the behavior that you're trying to study. So you really need to clean your social media data before you even put it into Tableau.

With the front end of a tweet, there are about four sources of information that you can get. But at the back end of a tweet, there could be 20 to 25 different types of metadata.

And Tableau is great with tweets. You get longitudinal, latitudinal data. And Tableau works great with that. You can map out where the tweets are into these great cluster circles. It works very well with the metadata variables on the backside of a tweet.

Tableau: What sort of data are you looking at?
Glen: We collect from a variety of social media platforms, Tumblr, Twitter, Facebook, You Tube, and WordPress. Our next platform is Foursquare. And Foursquare is all about geolocation, so we're really excited about working with that. It's a rapidly-changing environment for social data. New platforms are available. Anytime they become available, we try to to collect data there.

Health Media Collaboratory analyzed a year’s worth of tweets referencing cigarette smoking to identify peak opportunities for outreach.

Tableau: And how are you using Tableau with this social data?
Glen: A lot of our grant funders—the CDC, the National Cancer Institute— they're able to see the graphs. They want quick answers. They don't want to read a 20-page or 30-page report, they just want to see that graph, be able to look at it, and with some quick points, understand it.

A lot of our grant funders—the CDC, the National Cancer Institute—they're able to see the graphs. They want quick answers. They don't want to read a 20-page or 30-page report, they just want to see that graph, be able to look at it, and with some quick points, understand it.

Tableau: Can you give us an example of the sort of insight you’re able to gain by visualizing social data.
Glen: We're doing this analysis on people talking about quitting smoking on Twitter. So we took a year and charted it out on a histogram. And you could see the peaks.

We're telling those tobacco control organizations, ‘This might be a good time to reach out to these people.’ They can effectively use their resources by looking at the behavior of Twitter users.

Tableau: What is the value of visualizing social data?
Glen: The value is the insight. It's the insight that when you look at a graph. You get an immediate response to taking 1.7 million tweets and sticking it on a histogram and charting it weekly to see where the peaks are. It just pops out at them.

And I'm big about making things look good. I think it really matters. That might be vain in a way, but I think the way you present your data really matters.