Earlier this month I attended the National Centre for Research Methods’ 6th ESRC Research Methods Festival, at St Catherine’s College in Oxford. You can see the full programme from that link – including keynotes and paper presentations, but there was also a poster competition and a number of trade and exhibition stands (I flew like a pigeon and played with Lego).
I was interested from the perspective of my own ethnographic research in hyperlocal media, as I come to the end of an MPhil pilot study, and will be looking to develop and rethink method in time for my transfer to PhD ethnography work. I’ve already developed ideas in terms of this shift, but the festival was helpful in informing my ideas.
I started with a session of five presentations on Challenges/Opportunities of Using Social Media for Social Science Research. The first presentation offered ‘findings from a network of researchers using social media’, covering issues such as researcher identity and wellbeing, attitudes to data collection and actions that should be taken even when public data is being used and informed consent is not strictly necessary. There was some discussion over whether data could be used from tweets that were later deleted,as this might imply withdrawal from the public realm?
Next came fascinating insights into how social media has been mapped with a view to crime prediction, in the COSMOS project. More work exploring the “naturally occurring data” of social media was presented by Demos, looking at what we can understand about “attitudes” from Twitter. Natural Language Processing is a “mathematical appreciation of language”, although there was some concern about the validity of longitudinal studies in this respect as “language on Twitter is so unbelievably event specific”. Instead they used these methods to look at a volume of discussion on one subject, offering the case study of the Mark Duggan shooting that contributed to the riots of 2011. They also advised against dangers of extrapolating national sentiment from Twitter because, of course, not everyone tweets. Their next step is to explore whether language analysis can be used to make judgements about people re: ethnicity, class, demographic profile.
Farida Vis was last to present in this panel, on “The challenge of analysing images on social media”. She started by describing the new, mobile and everyday use of cameras, as opposed to tools for special occasions. The significance of Twitter images was discussed: tweets including an embedded image (as opposed to one you have to click away to view) get a 35% bump in RTs. She also presented some interesting thoughts about the role and identity of photographed images. Is someone manipulating Google Streetview to ‘capture’ an image of a street ‘doing photography’, given that image is part of a data bank taken by a bot at some point in the past?
On Thursday Professor Louise Ryan presented her use of the ‘sociogram’ as research method (I would actually liked to have seen more discussion of specific methods over the course of the festival). She used them in qualitative interviews as part of ‘ego network studies’ as opposed to quantitative whole network studies. These are drawn during interview to construct a visualisation of their networks – without necessarily being ‘accurate’, it is a method of self-reflexive representation. They can be used as part of interview technique, for informing and counterbalancing against discussion, often surprising the participants, and identifying where connections are not always positive even if they are close. They may represent conflict. They may represent offline or online relationships.
This was followed by two more presentations of longitudinal studies: ‘Insights from the National Child Development Study 1958: designing research instruments, data triangulation and participation in longitudinal studies’ and ‘Reconciling numbers and qualitative data in Young Lives, a 15 year study of children growing up in poverty in Ethiopia, Andhra Pradesh India, Peru and Vietnam’. Ethics had been a major discussion point across the course of the festival, and the trend followed here. Whilst these studies often deal with secondary research of existing data, there were still questions for the panel about how children give consent at such an early age, dropout rates art later stages of such studies, and participants’ level of comprehension.
All in all, an informative event… which means that annoyingly, I’ve picked up more reading for my literature review.