Data and Sampling (AQA A-Level Sociology): Revision Notes
Samples and Pilot Studies
Understanding basic concepts
When conducting sociological research, understanding different sampling methods becomes essential for producing reliable and valid results. Researchers must make informed choices about how they select participants for their studies.
The target population refers to the entire group that researchers want to investigate. This could be extremely large - for instance, when conducting a census, the target population includes everyone in the UK. However, practical limitations such as time and budget constraints make it impossible to study entire populations.
Practical limitations like time, money, and accessibility often make studying entire populations impossible, which is why sampling becomes crucial for sociological research.
Instead, researchers work with a sample - a smaller group of people who share similar social characteristics to the target population. The quality of research findings depends heavily on how well this sample represents the broader group being studied.
A sampling frame provides the list from which researchers select their sample participants. For example, a school study might use the student register as its sampling frame, or researchers might use specific community groups. One challenge arises when working with hidden or secretive populations where complete sampling frames may not exist.
Sample size matters considerably - if too few people participate, the data may not accurately reflect the target population's characteristics. Researchers must ensure their sample size is adequate or consider alternative sampling approaches.
Representative samples
The fundamental principle of good research suggests that samples should be representative - meaning they accurately reflect the social characteristics of the larger population being examined.
Random sampling
This represents the most straightforward sampling approach. Every person within the sampling frame has an equal opportunity to be selected, similar to drawing names from a hat. This method eliminates researcher bias in participant selection.
Stratified random sampling
This method involves dividing the population into specific categories that reflect important characteristics such as social class, gender, ethnicity, and age. Researchers then randomly select participants from each category in proportions that match the target population.
If black and minority ethnic (BME) groups represent 10% of the target population, then 10% of the sample should come from these communities. This approach improves representativeness compared to simple random sampling.
Systematic sampling
Researchers select every nth person from the sampling frame - perhaps every third, fifth, or tenth name. The interval between selections depends on both the total sampling frame size and the desired sample size. This method provides a structured approach to selection while maintaining randomness.
Cluster sampling
When obvious sampling frames are unavailable, researchers may randomly select groups or clusters of people who are naturally grouped together. However, if the chosen clusters are not representative, the entire sample may become biassed.
Multi-stage sampling
This approach involves selecting samples from other samples. Researchers might first choose typical geographical areas, then select representative streets within those areas, and finally randomly choose individuals from those streets. When done correctly with representative areas and streets, this method can produce good results.
Quota sampling
More common in market research than sociology, this method gives interviewers specific quotas to fill from predetermined groups based on characteristics like social class, gender, ethnicity, and age. While useful, one limitation is that people's characteristics may overlap across several categories, potentially affecting data quality.
Non-representative samples
These sampling methods do not aim to represent the broader population but serve specific research purposes.
Snowball sampling
Like a snowball growing as it rolls downhill, this method relies on existing participants recommending others to join the research. It often begins with a single person who acts as a gatekeeper, providing access to their network. Researchers traditionally used this approach when studying hard-to-reach groups such as criminal organisations, people involved in illegal activities, or exclusive religious communities.
The main limitation is that researchers cannot control who gets recommended, and samples tend to be small, creating risks for representativeness.
Purposive sampling
Researchers deliberately choose specific groups to test particular hypotheses or explore unique situations. Wright et al (2006) used purposive sampling in their street violence research, selecting participants from six prisons and correctional institutions. They specifically chose offenders serving sentences for crimes involving street violence, including robbery, grievous bodily harm (GBH), actual bodily harm (ABH), and firearms offences.
Sociologists sometimes study unusual groups or societies for specific purposes. Feminist researchers might examine gender-equal societies to challenge assumptions about patriarchal dominance being natural or inevitable.
Pilot studies
Many researchers conduct pilot studies before beginning their main research. These small-scale preliminary investigations test whether the proposed research methods will work effectively.
Pilot studies serve several important purposes. They help researchers trial questionnaires and interview schedules to ensure questions produce the required quality and quantity of data. If people ignore questions or misunderstand what responses are needed, researchers can modify their approach. Interviewers may use pilot studies as ice-breakers to build rapport with participants and establish whether respondents understand the research process properly.
By testing the research process with a small pilot sample, problems can be identified and corrected at minimal cost. This prevents expensive mistakes and can even prevent research projects from being abandoned if problems appear insurmountable.
Pilot studies typically involve people with similar characteristics to the target population but not those who will participate in the final research.
Contemporary applications
Modern sampling often uses computer software to reduce selection bias. However, automated sampling only works once researchers have identified sampling lists and target populations. This creates challenges when trying to identify members of secretive or hidden groups.
Key research examples
Research Example: Sharp and Atherton (2007)
Sharp and Atherton investigated young people's experiences of community policing in black and minority ethnic (BME) groups. They used snowball sampling to access their target population, beginning with contacts from local youth groups and organisations. Initial participants then suggested others to join the study. The researchers found snowballing worked well because participants explained the research purpose to others, making recruitment more effective. Without this approach, finding enough willing young participants would have been extremely difficult.
Research Example: The English Longitudinal Study of Ageing (ELSA)
This ongoing study collects health information from people over 50, involving collaboration between University College London, the Institute of Fiscal Studies, and the National Centre for Social Research.
Originally, ELSA followed a nationally representative random sample of approximately 8,500 people every two years. Researchers selected postcodes randomly from the Postcode Address File, then sent letters to households explaining the research and requesting participation in interviews and questionnaires.
The study has since been updated using stratified random sampling, incorporating respondents from Health and Safety Executive surveys between 2001-2011. Through combining questionnaires and interviews, researchers collect both quantitative and qualitative data, providing comprehensive information about health, economic circumstances, and quality of life among ageing people.
Key Points to Remember:
- Target populations are usually too large to study completely, so researchers use samples that share the same social characteristics
- Representative samples aim to mirror the target population through methods like random, stratified, systematic, cluster, multi-stage, and quota sampling
- Non-representative samples (snowball and purposive) serve specific research purposes but cannot be generalised to wider populations
- Pilot studies test research methods before main studies begin, identifying problems early and saving costs
- Sample size and representativeness are crucial - unrepresentative samples cannot support generalisations about the broader population