Transcript Reflections on the rise of transactional data in social research

www.cresc.ac.uk
The politics of new data and the
challenge of archiving
Mike Savage
CRESC & Sociology
University of Manchester
Issues
•
•
The coming crisis of empirical
sociology…
Implications for the social science
archive
–
–
–
–
•
Commodification
Visualisations and networks
Mapping
Temporality
Conclusions: the scope of archiving
The ‘coming crisis of empirical sociology’
The two main social sciences research repertoires - the
national sample survey and the in-depth interview gained (sudden) precedence in the 1950s, but are now
rather old (and tired?).
The intervening years have seen huge innovation in the
generation of new data and methods of digital analysis,
yet sociologists have not been centrally involved in
these.
In the 1950s, a special effort had to be made to collect
‘social’ data – which could then be archived - now such
data is routinely produced as part of normal transactional
– and non-archived - processes, making the role of
specially commissioned social research less clear.
Social scientists have not adequately responded to the
challenge of these new data sources, preferring to stick
to ‘tried and trusted’ data sources and methods…..
Implications for the social science archive?
• Recent years have seen increasing interests in the
archiving of social science data involving the deposit of
originally collected survey and interview data
– ESDS; Qualidata; Timescapes; .
• Over the past two decades these archival resources
have been increasingly mobilised by social scientists
– The Surrey School’ of secondary data analysis from the early
1980s
– Digitalisation of survey sources and their increasingly routine
deployment by researchers.
– Increasing – though still contested - interest in using archived
qualitative data, partly as a result of the ‘re-use’ debate (Moore,
CRESC)
• Yet the ‘coming crisis’ poses the issue of how we archive
new data sources, in a situation when
informationalisation challenges the distinction between
data and analysis.
Commodification
• Traditional market research relies heavily on
surveys (e.g TGI), but is now challenged by new
digital data
• E.g. Tesco ‘loyalty club’ data, analysed by
Dunhumby which is
– Molecular (we all have our own consumer DNA)
– (Nearly) instantaneous
– Incorporates the consumer into the research process
through reward processes
– Privately owned, not publicly archived, and rapidly
outdated
Archiving visual networks
• The usual archiving of social science data focuses on
depositing data on individuals, with ethical concerns
about ‘informed consent’.
• Transactional data, however, focuses on the links
between transactions. The attributes of the individual
‘transactor’ are not a central research issue.
• The data becomes less significant than the procedures
used to analyse the data. It is these which become
socially significant, and which should be archived so that
they can be evaluated by others (for instance, the role of
neighbourhood classifications).
• Consider the following example, developed by computer
scientists and physicists.
Mapping….
• Transactional data works through surfaces, and
deploys primarily spatial and visual operators.
• Digitised maps are highly manipulable, and offer
very specific details. Should these procedures
also be archived…..
• Following example is from Culture, Class,
Distinction where we (Tony Bennett, Elizabeth
Silva, Alan Warde, Modesto Gayo-Cal and Dave
Wright) have used mapping methods (in the
form of geometric data analysis) to lay out the
‘patterning of culture’ derived from a national
survey
Axis 1(λ1=0.1626): Cultural Engagement: involvement and
disengagement
Black: participation modalities
Red: taste modalities
Axis 2 - 3.86 %
0.8
Cloud of Individuals – axes 1 and
2
Hilda
Joe
0.4
Seren
Rachel
Vasudev
Caroline
Robert
Majid
Maria
Stafford
0
Jim
Margaret
James
Axis 1 - 5.33 %
Poppy
Molly
Jenny
Rita
Cherie
Ruth
Janet
-0.4
Cecilia
-0.8
Sally-Ann
-0.8
-0.4
0
0.4
0.8
Temporality and Sequencing
• Archiving has historically been assumed to
involve depositing data collected at specific
moments in time
• How does archiving deal with the routine ‘real
time’ collection of data in which there is no
central time point which justifies depositing at
any one specific moment?
• Are there lessons from computer science? How
do we retain information permanently? Do we
need to?
Conclusions
• New forms of data challenge assumed notions of
archiving which depend on differentiate the
collecting from the deposition of data.
• Informationalism might be said to conflate these
processes and replace them with a politics of
‘data traces’. In this case, archiving loses its
specificity
• Archiving has traditionally not covered
procedures of data analysis, yet it is these which
are increasingly central to data processes.
• Do we need to radically extend the scope of
archiving as a means of allowing public
contestation of modes of social research