What kind of research design will best suit your purposes?


To answer this question, I need to consider the two main approaches to research—quantitative and qualitative. They will be discussed again separately, not because I wish to promote a separatist way of thinking but for ease of understanding and because the two approaches arise from different research traditions and so have developed different research designs. the first and second figures below give an overview of quantitative and qualitative research designs.

phd30 Quantitative research design is more linear and sequential than qualitative. One step determines the next, and each is dependent on what has gone before. The logic is deductive in that it requires researchers to work from a theory or hypothesis and then gather data to describe it or test it.

phd30 Qualitative designs are more evolving and often circular. The logic is inductive—from data to theory.

Once you have narrowed down your research-question problem or issue, you need to work through a further series of questions to help you select the most appropriate research design:

  1. What will be the most suitable methodology, approach, or research style?
  2. What kind of data do you anticipate gathering?
  3. How might you gather this data?
  4. From whom will you gather this data?
  5. How might you analyse this data?
  6. How might you display this data?

The third figure below sets out these questions along with selected answers. These are not the only questions that you could ask at this stage, nor the only possible answers, but they are a useful place to start if you are less familiar with the research process. Although the third figure displays your choices as either quantitative or qualitative, you could design research that combines the two—they are not necessarily mutually exclusive.

How do you choose who or what to study?

When you have considered a suitable research design, LeCompte and Preissle (1993) suggest you next consider your data sources. You might choose to get your information from human sources (e.g., through a quantitative survey or a qualitative case study), or you might use non-human sources (e.g., existing statistics or documents), or you might want to employ a combination of all three. This section gives guidance on choosing your sources and samples.


This figure shows an overview of quantitative research designs.


this figure shows a typical qualitative research design

human data sources

One of the decisions you need to make when using human subjects in research relates to the sample selected to represent your chosen population or the case(s) to illuminate your chosen phenomenon. There are two main sampling techniques:

dota24probability sampling, used mainly in quantitative research; and
dota24non-probability sampling, used mainly in qualitative research.

The aim, when employing descriptive quantitative research, is to make statements with confidence. To get the most accurate description of a situation or picture of a trend, you could ask every member of the relevant population, but this is not possible (the exception being the national census, and we can onlyguess at the amount of organisation and co-ordination that this entails). Instead, you ask a sample from that population that you expect will fairly represent it.

Quantitative research uses probability sampling because it allows us to estimate to a certain level of probability that our sample will be representative. It also ensures that our sample is free from researcher bias. One of the key traditions in this research approach is that of researcher objectivity.

Probability sampling

Probability sampling means that it is possible to specify the likelihood of any element that meets the criteria for the unit of analysis being included in the sample.

There are three main kinds of probability sampling techniques: random, systematic, and stratified.

dota24Random sampling means that any element has the same chance as any other of being included in the sample. The subjects are chosen by random, using a method such as a table of random numbers or a computer-generated random sample.
dota24A systematic random sample is based on some consistent way of selecting subjects, for example, every fifth name in the phone book or every 10th school from the Ministry of Education database.
dota24A stratified random sample selects subjects at random from a set of categories that represents the profile of the population, for example, so many males/ females, a percentage of each ethnic group, a representative geographic spread.

As stated earlier, in line with the characteristics of quantitative research, probability sampling aims to reduce researcher bias and to extrapolate from the findings to the wider population. Great care therefore is taken to ensure that the sampling procedures are accurate and representative.

Non-probability sampling

In non-probability sampling, it is not possible—or even desirable—to generalise from the sample to the population. The sample is chosen for specific reasons to expand our understanding of the phenomena and not to make broad claims. The sample might, for example, be a sample of only one.



This figure shows the research questions and selected answers

The three main kinds of non-probability techniques are purposive, theoretical, and quota sampling.

1.  Purposive samples are selected because they suit the purpose. They might be a typical example, an atypical example, an exemplar, or a well-rounded example of the case or phenomenon you wish to study.

2.  Theoretical samples are guided by the theoretical framework you are using or the theory arising from the data analysis. If, for example, the theory describes four main categories, the sample might be examples of each.

3. Quota sampling is similar to stratified sampling in quantitative research and is made up of quotas for each of the categories you wish to represent (e.g., age, gender, school decile rating).

4. Within these, you might also use convenience or snowball sampling.

5. Convenience sampling simply means that you compromise your search for the perfect example and choose one that is easier to access but will provide useful data to illuminate your phenomenon of interest.

6. Snowball sampling is often used when access to a particular group is more difficult. Your first subject recommends another subject, who recommends the next, and so on. These subjects still fit within your purpose or theory, but you have less control over the actual choices.

However you select your sample, the expectation is that you will have followed an appropriate procedure and can justify your choices.

Non-human data sources

If you are not using human subjects, or you are but also require further information, you could gather your data from existing sources, but you will still need to consider and justify your selection of these. Within educational research, these sources include:

  • Existing statistics: sets of test scores, databases, yearbooks;
  • Documents: curriculum or policy documents, school plans, timetables, teacher or curriculum plans, textbooks, portfolios of children’s work;
  • Archival sources: old published documents, unpublished personal documents (such as diaries, letters, ledgers, minutes of meetings), photographs, workbooks;
  • Visuals: photographs, paintings or sketches, maps, symbols or logos, computer-generated images;
  • Audio-visuals: tape recordings, video recordings, radio broadcasts, films, computer slideshows, musical items, dramatic representations, dance, performances;
  • The Internet: educational sites, sites for children, sites for parents, sites set up by government departments or organisations, email communications, listservs, other discussion lists; and
  • Artefacts: objects of historical or cultural significance, everyday objects, artworks, models, work samples, portfolios.

In a Nutshell

  1. Successful researchers need knowledge of their discipline or field and topic, knowledge of the craft of research, and an understanding of the ethical respon­sibilities of a researcher.
  2. All researchers make decisions about formulating the problem, selecting the research design, choosing who and/or what to study, determining how to approach participants, selecting a means to collect the data, choosing how to analyse the data, interpreting and applying the analysis, and disseminating the findings.
  3. In choosing research topics, researchers need to consider should-do-ability, do-ability, and want-to-do ability.
  4. Researchers need to consider a range of factors when selecting a research topic, such as size, scope, time, resources, access, skill, previous knowledge, and motivation.
  5. Research questions need to be relevant, concise, and related to the choice of research design.
  6. Quantitative research designs are structured and linear.
  7. Qualitative research designs are more emergent and often recursive.
  8. Quantitative research uses probability sampling methods.
  9. Qualitative research uses non-probability, often purposive, sampling.
  10. Data sources in educational research are often human subjects in everyday settings, but researchers also use non-human sources, such as documents and artefacts.

Comments are closed.