Skip to main content

Quantitative data collection and analysis

What is quantitative research?

Quantitative research is concerned with gathering and interpreting numerical data. This data can be ranked (ordered), measured or categorised through statistical analysis. This analysis assists with uncovering patterns (or relationships) and for making wider generalisations to a wider population. 

This type of research is useful for finding out:

  • how many?
  • how much?
  • how often?
  • to what extent something occurs?

Data collection

 

For a more detailed list with further explanations - see the Statistical terminology document below

Hypothesis

Your proposed relationship between two variables that you are going to test.
Population

The entire set of people or events to which your hypothesis could apply.
Sample A subset of a whole population that you have looked at.
Variable Anything that varies and can be measured. It is the characteristic (attribute) which you would expect to differ from at least one of the others in your sample e.g. hair colour, height, age.

(Note: Most textbooks on research will have a Glossary that defines key statistical terms)

SAGE Research Methods (see link below) has lots of information on designing your research.

To access:

  • Click on Sage Research Methods (or link at the bottom of this page)
  • Click on Browse 
  • Click on Research Design (under topics)

Methods map 

Image taken from Sage Research Methods.

 

Quantitative techniques or tools

Surveys or questionnaires These ask the same questions to large numbers of participants. Can use Likert scales which measures opinions as well as providing numerical data.
Observations Can involve counting the number of times a specific phenomenon occurs, or more detailed coding of observational data in order to translate it into numbers for further analysis.
Document screening This involves sourcing numerical data from such things as financial reports or counting word occurrences in textual analysis.
Experiments Testing hypothesis to establish cause and effect relationships. These could be laboratory tests, field experiments or quasi experiments.

 

Sampling is the process of selecting a given number of units of analysis from a population. Although in social research these units are often individuals, they do not need to be but could be households, firms, countries etc. 

Sampling is often based on randomness to ensure representativeness, although non-random sampling might be used in some circumstances.

(Definition from: Sage Research Methods)

 

Sampling error is an estimate of how a sample statistic is expected to differ from a population parameter in a random sample drawn from the population i.e. how does the sample you are looking at reflect / is different from the population you want to expand your hypothesis to.

e.g. An estimate of the Mean from the sample will not be be the same as the population value. There will be a difference (an error). This error will be less if we thought about the sample carefully, ensuring representativeness and avoiding any biases.

The higher the error the less certain you can be that the value reflects what you would expect in the population.

 

Sampling variability concerns the situation where if you took repeated samples from the same population and then work out the means and standard deviations of these different samples they will not be the same.

 

Different types of sampling:

Random sample - every member of a population has an equal chance of being selected e.g drawing lots out of a hat.

Systematic sampling - a random sample where a system has been applied e.g. choosing every Xth member of the population.

Stratified sampling  - use if you are aware that specific groups in a population will be different from each other. The number from each group should be proportionate to the number in the general population  e.g. sex or age.

Cluster sampling - divide the population into clusters and randomly sample a number from these clusters.

Convenience sample - this is a sample that is conveniently available to you.

Snowballing - a variant of convenience sampling. You start with one in your sample who then recommend others. Useful for hard to reach populations.

Quota sampling - Participants are selected in a non-random way according to a fixed quota e.g using a quota such as 40% female and 60% male

Purposive sampling - sampling with a specific purpose in mind e.g females between 50 and 60.

  • Always consider how you will analyse your data when you are designing your research instrument, NOT after you have collected your data. 
  • You need to think about the kind of data you will be collecting and the sample size. This will have implications for the type of analysis you can do later.
  • Carefully and clearly define your population.
  • Carefully and clearly define your variable of interest.
  • Choose your sample carefully to avoid sampling error or sampling variability.
  • Become familiar with computer software (e.g. SPSS or R) and how you can manipulate and present your data in these systems.
  • Do not confuse statistical significance with substantive significance.
  • Does your analysis provide an answer to your research question?
  • Consider the nature of the variables being analysed e.g. nominal, ordinal interval/ratio or dichotomous.
  • Make sure each table you use has a heading and is clearly labelled.The table should  be fully understandable on its own. Tables are normally numbered.
  • If your are using a survey - think carefully about how your questions will be interpreted by respondents. Are the questions specific and clear?
  • If you are using interviews be aware of any unconscious bias, e.g by tone of your voice and body language.
  • Take care when coding your responses.

 

Developed from Bryman, A. (2016) Social resarch methods. 5th edn. Oxford: Oxford University Press

SAGE Research Methods