Orientation Observation In-depth interviews Document analysis and semiology Conversation and discourse analysis Secondary Data Surveys Experiments Ethics Research outcomes



Social Research Glossary

About Researching the Real World



© Lee Harvey 2012–2020

Page updated 29 April, 2020

Citation reference: Harvey, L., 2012–2020, Researching the Real World, available at
All rights belong to author.


A Guide to Methodology

8. Surveys

8.1 Introduction to surveys

8.1.1 Longitudinal studies

8.2 Methodological approaches
8.3 Doing survey research
8.4 Statistical analysis
8.5 Summary and conclusion

8.1 Introduction to surveys
Surveys of one sort or another are not only an important aspect of sociological enquiry, they are part of everyday life. 'Eight out of ten cat owners prefer Whiskas'; 'Labour has a 4% lead'; 'More women than men now smoke cigarettes'. These are all examples of surveying of one kind or another. They all involve asking people questions and summarising the answers in quantitative terms.

Market researchers spend enormous amounts of time and money finding out who buys what products. Opinion pollsters try to keep track of which political party is most popular and which issues cause most concern to electors. Surveys are used to find out who reads which newspapers and what people think of different television programmes. In general, surveys involve selecting a group of people and asking them a set of questions that have been designed in advance. The information is collected either by using interviewers, who may ask the questions face-to-face or via telephone or internet, or by asking the respondents to write down their answers on a form (or questionnaire) in hard copy or virtually through electronic medium.

8.1.1 Longitudinal studies
Most surveys tend to be one-off events. Some are designed to have a follow-up at a later date to explore changes and some, longitudinal surveys are designed to get regular responses from the same group of people over a longer period of time, often over often over the course of many years. Medical and pschological research, for example, use longitudinal studies to assess the impact of various treatments and therapies.

A classic longitudinal survey is the Study of Adult Development undertaken by Harvard Medical School. It has been running for 75 years and seeks to identify predictors of healthy aging. It has collected data on 268 year-two Harvard college students from 1939–1944 and is paralleled by a second cohort of 456 disadvantaged non-delinquent Boston inner-city youths from the same period. The subjects were all American nationality, male and white. The men's physical and mental health, perception of their career, marital quality and retirement experience are evaluated at least every two years by questionnaires, information from their physicians, or personal interviews.

The National Child Development Study is another cohort study, in this case of 17,000 children born during the same week in March 1958. They have subsequently been surveyed at 7, 14 and 21 years of age (Fogleman, 1983).

A rather more modest longitudinal cohort study was conducted by Wass et al. (2011, p. 20) examining how students developed critical thinking skills. twenty six zoology students volunteered to take part in the three-year study. They were interviewed in each year using a semi-structured technique that aimed to discover 'what had scaffolded their learning over the previous 12 months and how this had supported the development of critical thinking'.

Another form of longitudinal survey is the panel study, where a group of people is monitored over time (usually no more than a year) to assess changes in attitude, opinion or a limited range of behaviours. The panel members may be questioned several times, often with the same questions. Panel studies are often reported in the press at the time of a general election and broadcasting companies have audience panels who provide information on listening or viewing habits and opinions on programmes.

The key differences between cohort and panel studies are, first, that panel studies usually take place over a shorter period than cohort studies. Second, that panels are representative, as far as possible, of the population from which they are drawn, whereas a cohort is not necessarily a 'representative sample'. Third, panels are mostly used to monitor attitude changes over time while cohort studies are more intensive studies of the cohort themselves. Thus members who drop out of a panel study are replaced by new members with similar characteristics, whereas replacement does not normally take place in a cohort study.

For individual non-governmental research teams, longitudinal studies are difficult and sociological research of this nature unusual. The Teesside studies of Robert MacDonald et al. (2005), which examined transition to adulthood, are a rare example of an in-depth longitudinal study of hard-to-reach subjects. Two studies of 15 to 25-year-olds from Teesside,were undertaken at the end of the 20th Century (Johnston et al., 2000; MacDonald and Marsh, 2001, 2002, 2005) and a follow-up of a subsample was carried out in 2003. The results suggested that it is difficult to escape social exclusion, especially in a contest of de-industrialisation.

The use of continuous surveys like those carried out by the government such as the Census, the General Household Survey and the Labour Force Survey, or opinion polls carried out by market research organisations provide another way by which surveys can keep a 'running record' of changes. Continuous surveys, unlike longitudinal surveys, do not reinterview the same sample or maintain a panel. However, they allow researchers to study patterns of social change over time. Continuous surveying is very expensive and requires a substantial organisation working full-time.

Overall, the key advantage to longitudinal studies is the ability to show the patterns of a variable over time.

The disadvantages are that they are expensive and it is difficult to get funding for them. It is also difficult to keep all the initial respondents in the study as people drop out for various reasons or data gets lost. For example, the National Child Development Study, in 1981, 'mislaid' the names and addresses of 6,000 of their sample. There is also the problem that the subjects of the study become too familiar with it and unwittingly adapt to whatever they think the study is trying to show. More importantly, longitudinal studies cannot control for all the other things that may impact on the variables under study, such as other social events and changes in lifestyle.

Next 8.2 Methological approaches