Using existing data

The ecological fallacy is the mistake some researchers make of assuming that they can infer the nature and causes of individual people’s behaviour by studying more general, aggregated data about the social groups to which they belong). Often, secondary analysis is used to study a sub-group contained within a data-set and it can seem natural to impute something to the sub-group which has been found to apply to the larger set. This is the logical error of confusing ‘some’ people with ‘all’ people.
Reference: Bryman: Social Research Methods: 5th Edition Page(s) 322, Key concept 14.2

The disadvantages of secondary analysis stem from the researcher’s lack of direct involvement in the process of data collection, leading to unfamiliarity with the data. The process of getting to know the range of variables in the study and the ways in which they were coded is time-consuming. Sometimes the data-sets are quite complex, involving responses given at different ‘hierarchical’ levels – data may have been gathered at individual as well as at household level, for example. A further disadvantage lies in the quality of the original data. An examination of disclaimers given about the research should be taken into account. It should be clear by now that secondary analysis really does take a lot of time. Relatively speaking, however, it takes less time than gathering primary data and it is definitely kinder to your bank balance.
Reference: Bryman: Social Research Methods: 5th Edition Page(s) 312,313

The term “unobtrusive method” stems from the work of Webb et al (1966, cited on page 324), who pointed to the value of methods that do not involve the researcher being immersed in the field or interacting with participants. This can be said to reduce the “reactive effect”, whereby people change their behaviour because they know they are being studied. Key concept 14.3 (pages 324) shows four main types of unobtrusive measures, including “archive materials”, which, as Bryman points out, perfectly includes “official statistics”.
Reference: Bryman: Social Research Methods: 5th Edition Page(s) 325,326

Apart from the advantages of reduced cost and time discussed in the previous question, secondary analysis also offers advantages of access to high-quality data, opportunities of studying social trends over extended time periods and unobtrusiveness, among others. However, of its very nature, it does not allow the researcher to witness events at first hand. The researcher is cast in a more reflective mode because they are not confronting their field of enquiry directly.
Reference: Bryman: Social Research Methods: 5th Edition Page(s) 310-312

The first advantage is that the data has already been collected, so the researcher does not have to conduct experiments to get at the data. We cannot be sure that the studies are as suggested in answers (b) and (c). Indeed, we may have a problem with measurement validity unless we locate our research questions very precisely within the frame of the secondary data. However, because the data are compiled over many years, “it is possible to chart trends over time” (Bryman, p304), which cannot be achieved with structured interviewing.
Reference: Bryman: Social Research Methods: 5th Edition Page(s) 319

Official statistics have been regarded with suspicion by critics who point to the social processes involved in constructing these measures. The figures that end up in the official statistics may represent only the final stage of a long process of decision-making by various social actors. Bryman shows a diagram of the crime-statistics recording process in Figure 14.1, on page 320, itemizing seven intermediate steps between the criminal offence and its official recording. Something may happen, or not happen, at any one of these steps, which may lead to the non-recording of the crime. The ‘official crime rate’ reflects crime recorded rather than crime committed.
Reference: Bryman: Social Research Methods: 5th Edition Page(s) 320,321

Since secondary analysis involves the use of data that have already been collected by others, the researcher does not need to spend time and money on data collection. This can make the method attractive to those with limited resources, such as students. This does not imply that this is the only method that can be used by students – far from it. It may be the case that some students will see it as an “easy way out” of doing their own research. Done properly, this requires a lot of time and statistical knowledge. The point is that it may not be possible for students to compile such an elaborate data-set as they may find in secondary sources.
Reference: Bryman: Social Research Methods: 5th Edition Page(s) 309

Large amounts of data are collected by researchers and published regularly. Government departments and agencies are obliged to collect and publish statistics relevant to their areas of responsibility. Bryman asks, (p309) would it not make sense to analyse this data instead of gathering new material? “Secondary analysis” is the term we give to this kind of activity, “coming second” to the data that someone else gathered first. The most important data for quantitative research strategies is, fairly obviously, reported statistics. Secondary analysis does not involve going over the same ground but, rather, developing new insights into the data previously gathered.
Reference: Bryman: Social Research Methods: 5th Edition Page(s) 309

Table 14.1, on pages 316 and 317, shows a list of reliable data sets with details on each, including the Expenditure and Food Survey (EFS). This is a relatively new survey, which combined (and replaced) the Family Expenditure Survey (FES) and the National Food Survey (NFS) in 2001. It provides quantitative data about household income and expenditure, gathered through the use of “structured diaries” (see chapter 10) and “structured interviews” using CAPI (see chapter 9).
Reference: Bryman: Social Research Methods: 5th Edition Page(s) 316,317

Secondary analysts can selectively study a subgroup of the original sample. For example, within a random sample of the general population, you might decide to focus on married women under the age of 30. The ‘General Lifestyle Survey’ (GLF), one of many large data sets shown in Table 14.1 on pages 316 and 317, is one such national survey. Arber and Gilbert (1989, cited on page 310) isolated a group of elderly people from the 1980 GHS (the forerunner of the GLF) for a study on disability. Their suggestion is to use secondary analysis as an important adjunct to other quantitative methods.
Reference: Bryman: Social Research Methods: 5th Edition Page(s) 310