by QSR

Supports robust qualitative and mixed methods research for virtually any data source and any research method.

Learn more

by QSR

Intuitive data analysis software designed for public policy experts analysing surveys.

Learn more

Creating software to help you discover the rich insights from humanised data.

Learn more

Gendered parenthood in everyday ‘talk’ – using NVivo to analyze social media data

28 May 2015 - IN microlinguistics, mumsnet, NVivo, qualitative, research, sociolinguistics

My research combines my fascination with the rapid grown of social media and my interest in the gendering of parental roles.

The popular parenting website ‘Mumsnet’ brings together these interests and so offers fruitful grounds for my research. Its ‘Talk’ forum provides Mumsnet users with a platform on which to discuss topics ranging from gardening to conception, politics and everything between.

I see myself as a sociolinguist and a qualitative researcher. I’m interested in the rich detail of particular moments of digital interaction, in the way Mumsnet users negotiate a complex range of social forces as they construct a particular version of themselves at a given moment in time. NVivo has helped me along on my journey to pull apart my ‘small data’ and examine it from the inside out.

Capturing Mumsnet 'Talk'

Though I describe my data as ‘small’, I began my study by collecting a not insignificant number of 50 threads from Mumsnet Talk. I collected these threads over a period of several months, using a semi-ethnographic approach to immerse myself in my research context and select relevant data.

By using NVivo’s ‘NCapture’ tool, I was able to capture threads in their entirety without losing any of the idiosyncracies of interaction in this online context, such as emoticons, strikethrough text and ‘sidebar’ advertisements. My captured threads looked exactly the same in NVivo as they first did on the Mumsnet site, as you can see here:

Using the NCapture add-on was a fairly straightforward process. Once I downloaded it from the NVivo website, an icon appeared in the top right hand corner of my internet browser. Clicking on this icon automatically saved whatever page I had open at the time.

Once saved from my ‘downloads’ file, I was able to upload threads to my NVivo workspace using the ‘external data’ tab to select the files I needed from my computer.

Having access to threads in my NVivo workspace meant that I could utilise the full range of NVivo tools to familiarise myself with and analyse my data.

Getting to grips with my data: coding in NVivo

Throughout the process of observation and data collection, I continually organised my data by theme using the ‘node’ system.

[Tweet "The flexibility of NVivo’s nodes encouraged me to engage with coding as an analytical process."]

As I added new threads to my collection, I continually refined my nodes according to new insights, for example by subdividing them into two more nodes, changing their titles or merging similar nodes as one.

As I collected more and more threads, the coding process became more complex, so I began to use additional NVivo functions to explore the data in different ways.  For example, I used the ‘text search’ function in a way I can only describe as ‘mopping up’; after a long and constantly evolving process, there were bound to be key references I’d missed and text searches helped me to find them.

This screenshot shows how I used a text search to quickly find references for the node ‘enacting stereotypically ‘feminine’ qualities and language’:

Once I had collected and coded all of my threads, I continued to refine my nodes. At this stage, I also began to group nodes into ‘trees’ by dragging free nodes into a parent node.

In this way, I developed a much clearer sense of the overarching themes and relationships in my collection. The screenshot below shows my final five parent nodes, and some of the child nodes.

Using NVivo to support microlinguistic analysis

But what about the small data?

From my bank of 50, I analysed two threads in microlinguistic detail.

At this stage, I coded the data afresh, paying particular attention to linguistic patterns as well as content themes.

Using node trees, I was able to sort isolated linguistic features into groups and see at a glance how they could combine to create particular effects in the text. For example, in one thread there were a number of linguistic features that mimicked the style of personal advertisements:

Some final points

I’ve been using NVivo for about a year now and I could say a lot more about how it has supported my research.

For example, I haven’t mentioned the ‘memo’ function, which has helped me to organise my reflections in a systematic way and integrate them with my analyses.

I haven’t explained, either, that I used NVivo alongside a more traditional paper-and-pen approach, or how this worked in practice.

I also haven’t explored some of the problems I encountered, particularly with transferring social media data to NVivo. I do discuss some of these issues in my blog post ‘Using NVivo to analyse forum data’, if you’d like to read more. Some of these issues have since been resolved; for example, I can now view unconventional text features such as images, smileys and strikethrough text within nodes as well as within the original source document.

If you’d like to hear more about my experiences with NVivo, I’d love to hear from you - tweet me at @macksocioling

jai_profile_larger.gif Jai Mackenzie is a doctoral researcher in the School of Languages and Social Sciences at Aston University in Birmingham, England. She is currently using qualitative methods to explore the discursive construction of parenthood in Mumsnet ‘talk’.