I am currently at the International Communication Association (ICA) annual conference in Seattle.  I’ll be posting what I learn.

Facebook Graph API:

Anne Oeldorf-Hirsch,  U of Connecticut, USA

1. Create an app (use computer science expertise)
2. Study participants visit study site and are linked to Facebook after disclosure message and consent about data collection
3. On Facebook, participants grant permission for study app to access whatever information the researcher wants (example: friend lists, posts, post comments)
4. You then get a dataset organized by users ID number
5. You can then use that data as you wish, for example, to populate a survey where the user will explain or describe their Facebook activity.

NameGenWeb + Programming in Comm Research:

Nicole Ellison, Michigan State U, USA

Using  the NameGenWeb app to collect Facebook network data

• NameGenWeb (image above) is a Facebook app that collects information about a user’s network
• The app is slow in gathering this information, causing some people to quit the app prematurely
• Like any other app, the user must give the app permission to access their account data on the Facebook platform

Thoughts on programming and communication research

• The benefit is that you don’t need to rely on self-report – you have the user data
• The problem is that it is a skill set many comm researchers don’t have
• Comm grad students should learn programming
• Relying on computer science students creates black box problem, they are unlikely to have substantive expertise in the research question
• You are now at the mercy of the social media company (lack of control over data collection, plug can be pulled completely)

Getting data from Twitter

Deen Goodwin Freelon, American U, USA

• Data scraping: automating collection of digital data
• Pro’s: powerful for speed and convenience, free, start immediately, usually pretty easy, good for class projects
• Con’s: APIs limit availability, can only retrieve data within limited time windows, requires very high local system reliability (ie, internet outages means data collection shits off)
• Tools to use: NodeXL is easiest but not available for PC, Deen showed a nice table of options which I’ll link to (here) when he posts it
• Purchasing data is the best option if you can afford it
• Twitter data vendors: Gnip, Datasift, Sysomos (give them a time period and keywords and they give you the data)
• Trusting bought data: If you can’t validate an analysis, don’t use it (ie, identifying language or gender)
• Data formats: csv, xml, JSON, MySQL (you need to learn how to use them)
• Audit your data:  They might not have included everything that fit your query parameters (time, key words)
• Comm needs computational methods and needs a “development core” (for now, apprentice yourself)

Computation and comm research

Jeff Hancock, Cornell, USA

• Sending grad students to comp sci departments to take classes is not the solution, they come back frustrated, but without skills useful for comm.
• This is because comm and comp sci have different priorities
• programming languages: Java  is useless, Python is great
• NSF is looking for collaboration between social science and computer science