For research into the impact of the economic recession on people’s lives, consequential data from Internet searches for credit advice and locations of cash conversion shops could be of value. Self-published data such as online discussion groups and forums could be used for examining people’s attitudes towards the recession and their coping behaviour. For example, the online network and web resource Mumsnet, which is a self-selected online network of parents, has a large number of postings from its members on the recent economic recession. In addition, the organization itself has conducted a survey of its members on the issue of household spending. Example (anonymised) Mumsnet discussion group comments include:
I have to cook on a very tight budget so when I go shopping I go straight to the reduced selection and always stock up with as much reduced food as possible and either cook it that night and freeze it down, or freeze it straight away. That way my family can eat very healthy for a fraction of the price. (‘TAM’, 2011)45
Here, the social science researcher might code for key information, as they would do in a conventional qualitative textual analysis. This might include coding for: gender, type of food planning, attitude to cooking, healthy eating and family, language use, and comment length. Links to other posts might also yield even richer data. The researcher could also take a direct follow-up approach including posting messages to the online group, collecting information using methods such as purposive surveys, follow-up interviews and online discussions (effectively purposive online focus groups). The key challenge here in terms of social science research lies in the purposive nature of the samples and the limits to what can be claimed about any patterns identified in such self-published data.
Closely linked to data on people’s economic circumstances is evidence on people’s consumption behaviour and we now consider this.
2.3.2 Data on Consumer Behaviour
In the UK, a key survey of consumer behaviour and household spending is the Family Resources Survey (FRS),46 which is a continuous survey with an annual target sample size of 24,000 private households. The survey began in 1992. Households interviewed in the survey are asked a wide range of questions about their key demographics and their circumstances (including receipt of welfare benefits, housing costs, assets and savings). An end-user licence version of the data with reduced detail is available via download from the UK Data Service. A special licence version of the FRS is available to approved researchers via the Secure Data Service as described above. The special licence version includes additional variables and increased detail on some variables, particularly geographic location.47 Access to such data is free or available for a minimal administration fee.
Real time consumption data (a type of consequential data) is also now collated by online companies and supermarkets via loyalty cards. Such data are held on restricted access databases; however, samples have been made available for social science researchers.48 The data are of considerable commercial value to the companies and organizations that collect and warehouse them. Alongside individual records of behaviour, data mining techniques can be used to identify associations and patterns in the data. For example, the company Dunnhumby works closely with supermarkets and other retailers examining purchasing patterns in order to target marketing and product range, and optimize the personalization of consumer experience.49 Consumer profiling organizations such as CACI provide data products that contain hundreds of individual-level variables including income, spending, media consumption and types of leisure activities.50 As outlined above, these data products link multiple sources including: surveys, product warranty forms, public records, administrative records such as the Electoral Register, house sale information and consumption records.
Online search engines such as Google and retailers such Amazon collate search patterns and profile customers by page visits and purchases. Samples of this data (for example, Google Trends and Google Analytics) are made available either freely or for purchase. In terms of administrative records, government departments hold consequential data on benefits claims and payments at the individual level. This could, in principle, be combined with survey data to research patterns in consumer behaviour. Such databases can be so large that the importance of making inferences from a sample to a population is of less concern for certain types of research.
2.3.3 Data on Health and Well-being
The Health Survey of England (HSE)51 and the English Longitudinal Study of Ageing (ELSA)52 are two key surveys for examining health outcomes. The HSE is a representative survey of around 15,000 adults and children in England. It combines data on attitudes towards health, eating and exercise with physiological data. Core topics include: general health, smoking, drinking, fruit and vegetable consumption, height, weight, blood and saliva samples. Special topics include: cardiovascular disease, physical activity, accidents, lung function measurement, blood pressure and certain blood components. The data is geo-coded to Government Office Region (GOR) level.
ELSA is a longitudinal survey of around 11,000 people aged 50+ in England, which began in 2002. It includes information on key demographics, income, health and cognitive function. Both HSE and ELSA data are freely available and more detailed versions including additional variables and more detailed geographic information are available as restricted access via the Secure Data Service under strict terms of use.
Other sources of health data include consequential data such as General Practitioner (GP) prescribing records. The ADLS is facilitating access to such data by building links with data holders, developing standards and good practice for data sharing and providing training for researchers in safe handling, analysis and publication from such data.53 Real time prescription data would be a very powerful tool for mapping changes in health and well-being.
The UK Biobank has collated for research purposes genetic and other physiological and behaviour data donated by over half a million citizens for research purposes.54 Genomic data are a potentially invaluable research tool for the social sciences as well as health sciences. This includes studies where researchers use surveys of twins to try to identify the impacts of both contextual and inherited covariates. For example, research by Sturgis et al. (2010), which involved combining attitude data and physiological information, examined the genetic basis for social trust. In the Millennium Cohort Study,55 the collection of DNA from data subjects linked to the survey data is becoming more common. This resource has great potential as it allows the possibility of tracking genetic and environmental influences across the life course. As well as these intentional research resources, several commercial DNA profiling organizations have been set up, for example, Britain’s DNA, where the public are asked to donate a DNA sample.56
Other sources of health data include data traces of online searches recording patterns of health-related queries. Though there is some debate about the accuracy of such methods, the content of tweets and volumes has been shown to be of value in monitoring the spread of flu outbreaks,57 as have Wikipedia searches (see Ortiz et al., 2011; McIver and
