How a researcher uses NVivo to capture and analyze conference tweets
In terms of data collection, Silverman (2013) talks about our over reliance on interviews in our qualitative research, and suggests we should take more advantage of the “naturally occurring data” around us.
This data includes social media posts, a somewhat new form of data that is easily accessible and publicly available. Analyzing social media data can help to identify trends in local and global conversations, and has proven useful when looking at natural disasters such as tsunamis and floods (Bruns and Liang, 2012), and also in times of crisis such as the UK riots (Procter, Vis and Voss, 2013; Vis, 2013).
Conference collaboration via Twitter
In a different context, Twitter is being used at academic and industry conferences, providing another channel for conference participants to engage.
I have found this to be a positive experience when able to ask questions of the presenter without feeling embarrassed when I was a junior researcher. The medium also enables the ability to discuss controversial topics and speakers in a collaborative manner.
Grace Hopper Conference #GHC14
In the following example I will discuss my experience engaging in Twitter in October 2014 at the Grace Hopper Conference in Phoenix, Arizona.
The conference hosted more than 8000 women working and studying in computing. As I didn’t know anyone at the conference (or so I thought), I found Twitter a great way to engage with conference participants and presenters in different sessions using the hashtag #GHC14.
At the conference, the CEO of Microsoft presented and caused controversy when he said that women should wait for karma to get a pay rise/promotion in their organisation. At this point, Twitter erupted via #GHC14 and there was a visible shift in the online conversations, giving participants an outlet for their comments.
As I have a keen interest in this conversation, I decided to do some analysis.
1. Capture the tweets
The first task is to capture the data from Twitter. In either Chrome or IE go to https://twitter.com/hashtag/ghc14 to bring up all posts that include that topic. Then choose to NCapture as a dataset (via the blue button at the top of your browser).
2. Import the tweets
Then import this via the From Other Sources tab in NVivo.
Once imported, the data will display as a table with a bunch of meta-data attached (you probably didn’t know it was there!).
I collected 18 000 tweets from a single click. All fields except the comments field are imported as essentially quantitative fields to be used for comparison, and the comments field is like open-ended text in a survey which can be searched and coded qualitatively.
3. Explore patterns
The beauty of using NVivo to analyse social media data is that without being an expert in the software, you can start to explore patterns in the big data immediately.
To start this, click on the Chart tab on the right of the open window and a series of charts showing trends in the data can be found. By default you see which usernames have been used the most in posts. There are 922 posts that mention @GHC organisation, and the same number that mention @satyanadella. To see the posts using @ signs, double click on a bar in the chart.
Then by either going to the newly displayed Chart ribbon or via the right click menu, choose Select Data to change the view of the data to see the number of references (tweets/retweets) over time.
It is easy to move between various views (by day, by week, by month, by year).
It is helpful to see (albeit not surprising) that 8854 of the 18 000 tweets collected were posted on the same day as Satya Nadella gave his speech. The conversation continued strongly on the next day with 6370 tweets (this was the day after the conference finished). Then the tweets begin to peter out. The data clearly shows much less Twitter activity on the day before the controversial speech (the 9th only has 465 posts)
We can then also look at the number of tweets by week, with the overwhelming majority being posted during the conference (16618 compared to 1382 the following week).
It is also possible to view this data over time by username, essentially creating a chart that visualises a matrix (similar to the output of a matrix coding query). This breakdown shows an alternative view of the data with @satyanadella mentioned 641 times on the 10th Oct and 641 times on the 11th Oct. @GHC was mentioned 489 times on the 10th and 347 on the 11th.
There are more charts we can do but let’s look at the map tab on the right.
4. See where tweets are coming from
This shows the location of the person tweeting, indicated by where they said they live when they set up their account. There are 75 tweets from Brisbane - this is me!
There are also 130 tweets from Sydney, and this is a colleague I ran into from the Uni of Sydney at the conference. Note that those who do not list their location are counted as “worldwide” which is sitting behind the 1 in the Arabian Peninsula in the map (106 have this label).
You can also drill down to see the tweets from a particular location, which is how I found my colleague Nicky from Sydney.
5. Dig deeper
This is only just scratching the surface of what we can do with social media data in NVivo.
If you like this post and want to hear more on the next stage of analysis, tweet @adroitresearch and @QSRInt to let us know! Happy researching :)
Bruns, A. and Liang, Y.E. (2012) Tools and methods for capturing Twitter data during natural disasters. First Monday 17.4.
Procter, R. Vis, F. and Voss, A. (2013) Reading the riots on Twitter: methodological innovation for the analysis of big data. International journal of social research methodology 16.3 (2013): 197-214.
Silverman, D. (2013) A Very Short, Fairly Interesting and Reasonably Cheap Book about Qualitative Research, 2nd edition. London: Sage.
Vis, F. (2013) Twitter as a reporting tool for breaking news: Journalists tweeting the 2011 UK riots. Digital Journalism 1.1 pp 27-47.