Archival and Secondary Data

Tarani Chandola,Cara Booker

• Verlag: SAGE Publications Ltd
• Jahr: 2022
• ISBN:
  9781529710960
• Auflage: 1
• Thema: Ciencias sociales
• Thema: Investigación y metodología de las ciencias sociales

Data archives provide rich and expansive sources of information for researchers. This book highlights the utility of secondary data analyses whilst showing you how to select the right datasets for your study, and in turn get the most out of your research. Topics include:

· Generating your research question

· Selecting appropriate datasets and variables

· Examining univariate, bivariate and multivariate associations

· Visualisng your data with tables and graphs

Part of The SAGE Quantitative Research Kit, this book boosts students with know-how and confidence, to help them succeed on their quantitative research journey.

1 What Is Archival and Secondary Data Analysis?

Tarani Chandola

Cara Booker

Chapter Overview

• Background and history 2
• Secondary data analysis: what does it mean? 3
• Why should I do research on second-hand data? 3
• The ontological and epistemological issues of secondary analysis 4
• The social construction of data 5
• Useful concepts and terms in secondary analysis 6
• Further Reading 8

Background and history

The image of a researcher dusting books off a shelf of a library is one that applies to many disciplines still today, but in the social sciences, much of the data today is digital and held in online digital archives. These digital data archives contain a wide range of data on different aspects of social life from many countries. These primary data have been collected by other researchers for their own research purposes and deposited at a data archive for other researchers to use in secondary data analysis. This book is a step-by-step illustration for people new to secondary data analysis to use the data sets stored in such digital data archives.

While in some disciplines there may have been some reluctance to share data by those who collect data, there is now a growing acceptance among people who collect data of the scientific value for data sharing. This acceptance has grown partly due to the remarkable advances in electronic infrastructure for archiving and sharing of such data. There has also been a drive by research funders towards promoting and facilitating the reuse and secondary analysis of data. Furthermore, primary data collectors now see considerable value in having other researchers not related to their study use the data in other contexts, for other types of research questions and also to validate and test their original analyses in the spirit of open scientific enquiry.

The UK social sciences have a long tradition of secondary data analysis which has been championed by the Economic and Social Research Council (ESRC) with funding for digital data archives such as the UK Data Service (UKDS; www.ukdataservice.ac.uk) and its previous incarnations – the UK Data Archive and the SSRC (Social Sciences Research Council) Survey Archive. In other countries, there are data archives which hold large social science data sets like the GESIS – Leibniz-Institute for the Social Sciences, which is the largest infrastructure institution for the social sciences in Germany. To find out more about other European digital data archives, you can visit the CESSDA (Consortium of European Social Science Data Archives) Consortium (www.cessda.eu/Consortium). This is a consortium of social science data archives from several European countries. In the USA, the Roper Center for Public Opinion Research (https://ropercenter.cornell.edu) holds vast amounts of public opinion data, while the Inter-University Consortium for Political and Social Research (ICPSR; www.icpsr.umich.edu/icpsrweb) is another rich source of secondary data.

In addition to these digital data archives, there are a number of high-quality cross-national social science data sets which are also available for secondary data analysis, such as the European Social Survey (www.europeansocialsurvey.org) and the Survey of Health and Retirement in Europe (www.share-project.org). Researchers based in academic institutions can often freely access these data archives and data sets through open and internet-based application procedures.

Secondary data analysis: what does it mean?

A definition of secondary analysis that is widely used is ‘any further analysis of an existing data set which presents interpretations, conclusions or knowledge additional to, or different from, those presented in the first report on the inquiry as a whole and its main results’ (Hakim, 1982, p. 2). Secondary analysis thus implies some sort of reanalysis of data that have already been collected and analysed.

The term secondary analysis is most often used in relation to survey data sets, although many other types of data are available for secondary analysis, including administrative data (which are not typically collected for research purposes) and qualitative data. Data sets that describe regions and countries such as census data sets are also available for secondary analysis. However, much of secondary data analysis is of surveys of individuals, so the secondary analysis of survey data is the main focus of this book.

A primary data source is an original document that contains first-hand information about a topic or an event. There are many different types of primary data sources, but in the social sciences, we tend to rely on data from a wide range of sources, including questionnaire and interview transcripts, video footage and photographs, historical records, blogs and social media entries, eyewitness accounts and newspaper articles.

Analysis of a secondary data source is an interpretation, discussion or evaluation of an event or issue that is based on primary source evidence. Existing survey data can be a primary data source, but they could also include reviews of other research, journal articles, abstracts and bibliographies. The differences between a primary and a secondary source can be ambiguous. A source may be primary in one context and secondary in another. Some key elements of difference are around whether the data creator/collector was part of the data analysis. If that is the case, then that is likely to be a primary data analysis.

Why should I do research on second-hand data?

You may feel that all the interesting research questions have already been answered by the researchers who collected the primary data. And it is certainly important to read their papers and reports using the data to see if they have already addressed some, if not all, of the aspects of your particular research question. But even if that were the case, there would still be some utility in doing secondary analysis because it is important to reproduce results from previous studies. It may be that you could be using different methods to address the same research question, and you may need to find out if your results differ from the primary data collector’s results. Many statistical methods come with a variety of assumptions, and it is useful to analyse the same research question on the same data using a different statistical method that may not have the same set of assumptions to see if the results or inferences are robust to the assumptions underlying the statistical method.

Another big advantage in doing secondary data analysis is saving the cost of and time taken to do primary data collection. Conducting a high-quality survey on a large number of people could involve quite a lot of money, even if you use internet surveys to lower the cost of data collection. Then, there is the time spent in waiting for survey respondents to fill in their answers to questions, and then the time to get all the survey data into a manageable data file ready for analysis. Reusing data from a data archive, on the other hand, is often free for academic researchers, and most data can be downloaded instantly from data archives after filling in some details about your proposed secondary research. Much of the secondary data in digital data archives contain larger sample sizes of high-quality data than what most researchers could realistically produce themselves, saving time and resources.

While the data may be ‘second hand’, with a previous owner, this should not put researchers off. In fact, unlike second-hand cars, the more frequently the data are used, the greater the likelihood of the data being of high quality. Most secondary researchers will tend to stay away from data that has been badly collected or with little relevance beyond a specific research project.

The ontological and epistemological issues of secondary analysis

Secondary data analysis tends to follow a scientific paradigm which consists of the following components: ontology, epistemology and methodology. The scientific paradigm or positivism originated from studies of the natural world and has then been applied to understanding the social world (Scotland, 2012). The ontological position of data analysis is one of realism, one of the central tenets of which is that the objects of research (e.g. survey participants) have an existence independent of the researcher (e.g. the survey interviewer or a secondary data researcher). A positivist epistemology is often associated with this realist ontology. Most positivists assume that reality is not completely dependent on the subjective perceptions of the data collector, and thus, data collected by somebody else can reveal an objective reality that is independent of the person collecting the data. The positivist epistemology is one of objectivism. Data collectors can go about discovering an objective reality that is impartial to the senses and perceptions of the data collector. The researcher, data collector and the researched are independent entities. The data that is collected can be objective, valid and factual and does not solely reside in the perceptions of the data collector.

Positivists view their methodology as value neutral and deduce true relationships from the objective data. Methods such as correlation and experiments are used to reduce complex interactions between data in the discovery of new patterns or truths. Methods such as inferential statistics from secondary data sets allow the analysis of data from specific samples to be generalised to wider populations. Positivist research is valid if the results from researchers can be generalised/transferred to other populations or situations (external validity) and different researchers can arrive at the same conclusions (replicable and reliable).

The positivist paradigm of secondary data analysis comes with some limitations. Methods developed to understand the natural world may not always be directly transferable to the social world. Positivism attempts to reduce the complexity of the social world to a few variables and cannot completely capture all the different aspects and contexts of social life. Some key data or variables may not be known to the data collector at the time of data collection. For example, in some studies of health conducted in the first half of the 20th century, data on smoking was not collected because the relationship between smoking and ill health was not known at that time. This meant that secondary data analysis looking at health outcomes from those particular data sets were limited, because many people were smokers at that time, and this information was not collected in the data.

Positivists can also self-delude themselves into thinking that their research is ‘objective’ and ‘value free’. However, researchers often make value-laden judgements – for example, through the selection of variables to be analysed, the data to be observed and collected and the interpretation of findings. Someone conducting research using secondary data analysis is not necessarily a positivist as they may be well aware of the limitations of the data and try to take account of these limitations when analysing the data or making inferences from the data.

The social construction of data

Any analysis of secondary data needs to explore who defined the original research topics, what methods and definitions were used in that research and what the assumptions were behind those research questions. As all data are subjective to some degree, it is important that these assumptions underpinning the original primary research are made clear and explicit to secondary researchers.

For example, while some people may view official data such as a population census as ‘objective’, the type of data that are collected in such official data is a result of some value-laden judgements of a committee or researchers about what are the important questions that need to be asked in that population. Furthermore, the categories of data that are collected are subject to some degree of subjectivity. For some data like ethnicity, the degree to which data on potentially small ethnic groups are collected needs to be balanced against the cost of asking such detail in the population. If you have a questionnaire with close to 100 categories of ethnicity to choose from, then that makes it difficult for people to choose the group they identify with. So there is another level of subjectivity that results in specific data being collected even in official statistics.

Another source of subjectivity is the process of data collection. Although a lot of data collection exercises use standardised methods such as an interview schedule or questionnaire and standardised methods of interacting with participants in a study, there is still room for some element of subjectivity. For example, we know that there are effects of the interviewer’s race and gender on interviewees and how they respond to a survey. Moreover, the interviewer may (consciously or unconsciously) bias a participant’s response to a set of questions. An additional source of subjectivity can come from the data analysts themselves, who select particular sets of analyses to conduct and present a selection of their research for publication or review.

The subjectivity in any survey data collection and analyses does not mean all survey data analyses are biased to the point that there is little value in doing such research. Rather, it is to point out that all research is subjective to some degree and that an awareness of this subjectivity is a necessary first step in making the assumptions underlying the data and analysis explicit.

Useful concepts and terms in secondary analysis

Data and theory

Social data and theory are intertwined. Data are empirical observations, and theories are the ideas and concepts that organise existing data. Theories are also used to make predictions about new data and what we should expect to find (or hypothesise) with new primary or secondary data that have not yet been analysed for a specific research question. Good theories and hypotheses must be strongly supported by the data. Secondary data is often very useful in testing middle-range theories and hypotheses under specific conditions. For example, a theory about how social mobility occurs may have been generated by observations (or data) from a specific country (or context). Secondary data on social mobility and the related factors from a different country could test whether the theory holds in other contexts.

Secondary data analysis versus big data analysis

Some of the secondary data sets held in data archives contain very large amounts of data, although they differ from what many people call ‘big data’. Big data, like secondary data in data archives, are digital data, r...

Inhaltsverzeichnis

• Cover
• Half Title
• Acknowledgements
• Title Page
• Copyright Page
• Contents
• Illustration List
• Table List
• About the Authors
• 1 What Is Archival and Secondary Data Analysis?
• 2 From Ideas to Research Questions
• 3 Finding Secondary Data
• 4 Getting to Know the Data
• 5 Basic Data Management
• 6 Manipulating Data and Basic Statistical Analysis
• 7 Writing Up Your Analyses
• 8 Complexities of Working With Secondary Data: Limitations of Archival and Secondary Data Analysis
• 9 Conclusions
• Appendix 1 Case Study Example of Secondary Data Analysis Using Biomarker Data From Understanding Society, the UK Household Longitudinal Study
• Appendix 2 Software and Data Used in This Book and Related User Guides
• Glossary
• References
• Index