Skip to Main Content

Skills for Learning: Research Skills


Research design is the framework of research methods and techniques chosen by a researcher. Good research design enables you to obtain appropriate evidence to address the research problem effectively. However, some students begin their data collection too early, without thinking carefully about the best research strategies and approaches. Poor research design can affect the validity of your results and conclusions.

We run interactive workshops to help you develop skills related to doing research, such as data analysis, writing literature reviews and preparing for dissertations. Find out more on the Skills for Learning Workshops page.

We have online academic skills modules within MyBeckett for all levels of university study. These modules will help your academic development and support your success at LBU. You can work through the modules at your own pace, revisiting them as required. Find out more from our FAQ What academic skills modules are available?

Research approaches

There is considerable debate regarding alternative strategies to research methods (quantitative versus qualitative). However, it is important to remember that different problems require different methods.

You can use:

  • a mono-method approach – either quantitative OR qualitative
  • a mixed-methods approach – a mixture of both

The table below illustrates the differences between quantitative and qualitative approaches.



Data collected is numerical.

Data collected is more descriptive (words).

Data is analysed and converted into statistics.

Data is sorted and categorised in order to understand and explain opinions/attitudes/beliefs.

Sample sizes can be quite large (to ensure they are statistically significant).

Sample sizes are generally smaller. Obtaining and analysing the data can be time-consuming.

Researcher is more likely to start from stated theories and hypotheses.

Researcher is more likely to start with a research question. Theories and hypotheses emerge or are generated from the data.

Comparison adapted from Cook and Campbell (1979) and Guba and Lincoln (1989).

There are major differences between quantitative and qualitative research methods, as shown above. However, both perspectives can be equally valid and useful. You should choose the method which is most appropriate to your research question(s).

Many researchers combine both in what is referred to as a 'mixed-methods' approach. Below are three ways that quantitative and qualitative methods can complement each other:

  1. You can conduct qualitative research as a preliminary to quantitative research. For instance, observations, interviews and focus groups can be used to find out more about a situation or topic. This information can help you to identify a research problem.
  2. Qualitative methods can be used to supplement or validate quantitative work, e.g. as part of the triangulation process.
  3. You can use qualitative research to explore complex ideas that cannot be explained using quantitative methods. For instance, qualitative methods can allow you to investigate complex behaviours.

Qualitative research is more useful for examining interactions and participants' perceptions. Meanwhile, quantitative research is more suitable for identifying frequencies and patterns. For example, you may choose quantitative analysis to reveal previously unsuspected relationships or correlations between variables. With qualitative research, you might gain an understanding of how the interactions take place in reality.

Research methods

You need to choose the most suitable research method or technique to meet your research objectives. Check Discover for resources on research methods used in your subject area.

Download the Research Methods Checklist to help you.

Common research methods

Observation refers to the process of observing and recording events or situations. The technique is particularly useful for discovering whether individuals or groups do what they say they do, or behave in the way they say they do.

There are two main types of observation: participant and non-participant.

  • In participant observation, the researcher becomes part of the group studied and joins in with activities, observing everyday situations and behaviour. People's own interpretations of events are uncovered through conversations.
  • In non-participant observation, the researcher simply observes without taking part. Whilst this prevents researchers from becoming over-involved, they might be less likely to understand the meanings and motivations behind what they observe.

In observation studies, the observer can remain covert (hiding his/her true identity as a researcher) or be overt (where their identity is revealed to those studied). It might be argued that the subjects of covert research are less likely to modify their behaviour. This is because they don't know they are being researched. However, there are ethical issues with covert research. For more information on this, see our topic Research Ethics.

Advantages of observation

Disadvantages of observation

Requires little training or familiarisation.


Can understand meanings behind actions.

Problems with recording data.

Behaviour can be observed in its natural environment.

For studying small groups only.

Can study deviant or illegal groups.

Cannot make generalisations. No way of judging whether the group is typical.

Flexibility: researcher may come across conditions and events not previously known.

There are ethical issues with covert research.


Moral, legal and injury risks are also associated with this method.

Data collection

It is impossible to keep a record of everything, and you must decide at the outset where your interests lie. You may decide to film or record events. Some researchers write notes in private after the event or set up systems for recording categories of behaviour using graphs, charts and plans. Your decisions will depend on the specific issues you're investigating.

Prepare carefully before observation begins. Remember, the aim of observation is to be unobtrusive. This allows behaviour to be as normal as possible.

The most common criticism of observation is that it is highly subjective, depending largely on the researcher's own focus and interpretations of what they have observed.

General ideas for carrying out observation

This is not an easy option for a research project. It takes meticulous planning.

Examples of use of participant observation

Studies of the social structure and functioning of small communities or 'deviant' groups, such as criminals.

ACCORDION: Converted tabs - this tab opens by default
ACCORDION OPTIONS: enable collapse / start fully collapsed

A questionnaire is a type of survey where respondents write answers to questions on a form, which might be on paper or online. All respondents are asked identical questions, so that information can be analysed for patterns and comparisons.

Questionnaires can be used to gather information on almost any topic involving large or small numbers of people. The commonest type of questionnaire involves closed choice or fixed questions where the respondent is required to answer by choosing an option from a number of given answers, usually by ticking a box or circling an answer. These types of questionnaires gather straightforward, uncomplicated information. Open-ended questionnaires allow the respondents to record answers in their own words. These are more qualitative and can produce detailed answers to complex problems.


Closed choice question:

For which of the following reasons do you most often go to the town centre?

  • to meet friends
  • to have a meal
  • to do some shopping
  • to visit the cinema or theatre

Open-ended question:

People go to the town centre for different reasons, Why do you most often go?

There are advantages and disadvantages associated with each type of questionnaire. Open-ended questions give greater insight into the topic researched but the answers can be difficult to classify and interpret.

Advantages of closed questionnaires

Disadvantages of closed questionnaires


Limited answers.


Lack of qualitative depth can result in superficiality.

Easier to analyse data.

No way of probing for more information.

Easier to survey a large number of people.

Not possible to verify answers and questions may mean different things to different people.

Consistent and less scope for bias introduced by different researchers.

Predetermined boxes may not be appropriate.


Low response rate.


Instructions must be clear and unambiguous and questions carefully worded.

Data collection

Printed questionnaires can be given to respondents personally, or posted which increases the cost and decreases the response rate. Web based questionnaires are easy to construct - and easy for participants to complete, though there can be problems with getting responses.

General ideas for constructing questionnaires:

  • Get the beginning right to encourage respondents to read on. State what the survey is about and how long it will take to complete.
  • Make the questionnaire look clear and attractive.
  • Use a large enough type size and avoid block capitals.
  • Keep sentences short and sentence construction simple.
  • Avoid leading questions.
  • Avoid jargon and technical terms and make sure questions are not ambiguous.
  • Watch out for questions which ask two separate questions at once.

Interviews are a type of survey where questions are delivered face-to-face. An interview is a conversation with a purpose - in this case, obtaining information relevant to the research topic.

Interviews can be either quantitative or qualitative and there are many variations. Purely quantitative interviews are like a closed ended questionnaire which the interviewer fills in.

An unstructured, purely qualitative interview is rather like an informal conversation. Here questions are asked in the natural course of interaction.

A large number of interviews will fall somewhere in between and are known as semi-structured interviews. These have predetermined questions, asked in a particular order, or a list of issues to be covered.

Structured interviews maximise reliability and are easier to classify and quantify. By contrast unstructured interviews can give greater insight and understanding, but need more expertise to control and more time for analysis.

Advantages of interviews

Disadvantages of interviews

High response rate.

Limited sample size.

Can collect complex information.

Can be difficult to analyse.

High degree of researcher control achieved.

May be a hostile reaction.

Can be made more responsive to early results.

Whole process is time consuming.

Relaxed environment.

Recording techniques may cause problems.


There is room for interviewer bias - this should be acknowledged.

Data collection

Conducting a structured interview involves ticking boxes on a form. A less structured format needs some way of collecting data, such as taking notes or audio recording. With audio, details cannot be missed, and the interviewer can give their full attention to the respondent. However, interviewees might be uncomfortable about the device and transcribing audio is very time consuming.

General ideas for carrying out interviews:

  • Explain who you are, what the survey is about, and discuss confidentiality.
  • Establish rapport: be friendly and look interested in the respondent.
  • Think about your body language.

Interviewees can feel more at ease with someone who seems similar to them. Whilst the majority of personal features can not be changed, you can dress appropriately - for example, a company director at their place of work is unlikely to be wearing jeans and a t-shirt.

Be familiar with your questions and ask them in a neutral manner. Your role as an interviewer is to listen, not to speak.

The focus group is a type of interview which involves carefully selected individuals who usually do not know each other. There are generally 7-10 members plus the researcher. Individuals are selected because of characteristics that are of interest to the researcher. A group discussion is held in a relaxed environment in order to extract opinions and share ideas. It is not necessary to reach a consensus.

Focus groups are especially useful for providing qualitative data which gives an insight into attitudes and perceptions.

Document analysis refers to the process of using any kind of printed document or resource, such as film, books, papers or letters, for analysis in relation to a particular research question. It can be used as the sole method of research or to supplement other methods.

Document analysis (also referred to as content analysis) differs from the majority of research methods in two major ways:

  1. It is an indirect form of research; you are investigating something which has been produced by others.
  2. It is an 'unobtrusive', or 'non-reactive' method. The document will not be affected in any way by your research - it cannot react as a human can.

Reliability and validity are central concerns in document analysis. Documents generally exist for a purpose and knowledge of this purpose is important in understanding and interpreting the results of the analysis.

Advantages of document analysis

Disadvantages of document analysis

The data never alters.

Sources might be subject to bias and subjectivity.


Evidence may be out of date.

Events can be compared over time and cultures.

May not be accurately recorded.

Gives an expert understanding.

Documents available may be limited.


Can be laborious and time consuming.

Data collection

A recording unit must be defined. For example, a study of newspaper content may concentrate on the number of stories on a particular topic or the column inches devoted to a particular subject. There is a vast range of research possible, but categories for analysis have to be decided upon.

General ideas for carrying out document analysis:

  • Decide initially on categories for research.
  • Keep focused and do not let your research become too wide.

Examples of use:

  • For studying racial or sexual bias in newspapers.
  • To obtain a historical understanding of a particular institution or group.

Put simply, experimental research measures the changes that occur when something new is tried out (Robson, 1978). It is associated particularly with the physical sciences, although experimentation is also used in social science.

To conduct experimental research, the researcher must alter at least one element of the study. The researcher then assess the effects of changing this variable. Measurement is required before, during and after the experiment. The experiment must be replicable, producing broadly similar results each time, to be considered significant.

There are two different types of experiment: the laboratory experiment and the field experiment.

  1. In laboratory experiments the researcher will conduct a small scale study where subjects can be manipulated, observed and tested in a highly controlled environment. Statistical data is often obtained from these types of study. Such research creates an artificial situation where events normally linked are separated.
  2. A field experiment is an experiment which takes place outside the laboratory. This leads to a decrease in researcher control which may hide the effects of changes made. The results gained can be generalised to the real world.

Research into human behaviour tested by experimentation is subject to much criticism for ethical reasons. Until the experiment is complete, it is unclear whether this method will be beneficial or disadvantageous.

Advantages of experiments

Disadvantages of experiments

Ideas can be tested in a controlled way.

Where human subjects are involved it is generally viewed as unethical.

Ideal for investigating causal relationships.

Results may be different in the real world to those discovered in a controlled environment.

Can generalise effects.

The influence of all variables can never be eliminated.

Scientifically validated findings give greater value to research.

Restricted range.


Large amount of preparation is required.


Humans may respond to expectations of the experiment not to the experiment itself.

General ideas for carrying out experiments:

  • Careful preparation is essential and experienced researchers should be consulted before experimentation begins.
  • Project design, sample selection and measurement of dependent variables are crucial to the success of the research.

Mathematical modelling can be used to analyse relationships between different variables and to predict possible outcomes, or causal effects.

Experiments can be designed from models of systems, which aim to define links between variables and outcomes.

Advantages of modelling

Disadvantages of modelling

Can extend powers of deductive reasoning.

Does not explain why variables are linked to particular outcomes - cannot explain why particular variables are important.

Attempts to be objective - maths is 'neutral'.

Model produced is limited to one situation and therefore may not apply to others.

Is an aid to causal explanation and can therefore help calculate the effects of actions.

Inability to distinguish causal from accidental relations.


Could be built on preconceptions.

Example of use:

  • To explore why there are gender differences in the student intake of science and technology courses.

Using mathematical modelling it is possible to isolate variables that may have an effect on the choices that men and women make.


Before you choose a sampling technique, it’s useful to identify your sampling unit. Ask yourself the following questions:

  • What problems might you encounter when attempting to gain access to your sample population?
  • Who should be included in the sample? The population of a town? A doctor's list of patients? A particular group of people (identified by race, gender, ability or ethnic background, perhaps)?
  • Considering your original research aims, why is it important to sample this group?

Gaining access to your chosen sample population is not always straightforward. You may encounter problems with the following:

  • Some organisations are unlikely to allow you access for your research, e.g. army barracks.
  • Your own personal details (age, ethnic group, gender, etc.) may make it more difficult for you to access some groups.
  • You may be faced with ‘gatekeepers’: individuals within an organisation, such as managers, who control access to information.

When you have thought carefully about these issues, you can begin to design your sample.

There are two ways of approaching sampling: probability sampling and non-probability sampling. The main types of probability sampling are stratified random sampling, multi-stage cluster sampling and systematic sampling. The main types of non-probability sampling are quota sampling, purposive sampling and snowball sampling.

Download the Sampling Checklist to help you.

Sampling method

Probability sampling can be split up into the following:

  • Stratified random sampling
  • Multi-stage cluster sampling
  • Systematic sampling

For probability sampling, you must ensure that each person has an equal chance of being included in the sample. A complete (as near as possible) list of the population must exist, if everyone is to have an equal chance. This list is called a sampling frame. Examples might include the electoral register, the Postcode Address File and telephone directories. It is possible to measure the degree of sampling error when using probability sampling. However, it's important to remember that all of these sources can have drawbacks. For example, some people may not be included or the database might be out of date.

When a sampling frame has been defined, the next step is to select a random sample from the frame. A random sample is precise and statistical; it is not a haphazard selection of names, addresses etc. In order to create a random sample, give each person or address a unique number, starting from one. Then, you must make a mathematically random selection. Probability sampling techniques include stratified random sampling, multi-stage cluster sampling and systematic sampling.

Sampling techniques associated with probability sampling:

Stratified random sampling

Involves dividing the population you're researching into separate strata or groups (e.g. age, gender, class, ethnicity).

Example: if you were researching the career plans of LBU students, you would want a representative sample of the population. To achieve this, your sampling frame should be stratified into:

  • Subject studied 
  • Year of study

A percentage, perhaps 10%, would then be drawn from each subject of study, or stratum.

Stratified sampling enables the researcher to avoid the sample containing a higher proportion from one particular group. This problem, which is common with simple random sampling, makes the sample biased. The researcher may need to stratify the sample further – for example, by gender or age.

Multi-stage cluster sampling

This technique simply involves selecting one sample from another. This is a good way of obtaining a random sample if it's too difficult or expensive to gain access to a complete list of your chosen population. First, decide which population you want to sample from (e.g. schools, universities, health centres etc.). Next, draw a sample from this population. Proceed in this way until you have obtained the number of people you require for your final sample.

Example: If you wanted to obtain a national sample of midwives, you could:

  1. Draw up a list of health authorities and randomly sample from this list, then
  2. Draw up a randomly chosen sample of health centres, and
  3. Lastly, take a randomly chosen sample of midwives from each centre.

As can be seen from this example, multi-stage cluster sampling involves many sampling stages. At each stage, the sampling units or 'clusters' are subdivided.

This technique, however, may produce a sample that does not represent different groups of the population. For example, 'clusters' from certain geographical areas may be biased towards one class or ethnic group. It is a good idea to select as many clusters as possible and to stratify your final sample.

Systematic sampling

This type of sampling involves selecting a random number and then systematically sampling every nth individual, school etc. The main problem with this technique is that bias can result.

Example: If you select every 5th household from an electoral list, you may create a sample of households that are very similar. The sample may not represent the population of households within the area.

ACCORDION: Converted tabs - this tab opens by default
ACCORDION OPTIONS: enable collapse / start fully collapsed

Non-probability sampling can be split up into the following:

  • Quota sampling
  • Purposive sampling
  • Snowball sampling

In order to carry out probability sampling, you need to have a sampling frame. Non-probability sampling is used when a sampling frame is not available. It's also used if the research does not generalise from the sample to the wider population. If the research aims to build theory rather than generalise, non-probability sampling is appropriate. However, the sampling error cannot be measured using this approach.

This type of sampling is often used in market research. It's useful when you're seeking to gain opinions, rather than obtain a representative sample. Examples of non-probability sampling techniques include: quota sampling, purposive sampling and snowball sampling.

Quota sampling

As the title suggests, this type of sampling involves selecting a quota of the research population. Here the sample is not selected at random from a sampling frame. There are some similarities with stratified sampling: the researcher decides which categories of people will be included and how many. However, instead of selecting a random sample, the researcher looks for the right number of people for each category.

Quota sampling is less likely to produce a representative sample of the research population. This can be problematic. However, it's an easy, inexpensive and fast method. For these reasons, quota sampling is often used for opinion polls and market research.

Example: First, you need to decide what participant characteristics are important. They might include age and sex – e.g. women aged 20-35. If the sample was 200 people and the quota of women in this group was 70, you would find 70 women of this age group.

Purposive sampling

This involves selecting a particular group or place for the research to take place because of a known characteristic.

Example: You may decide you want your research to take place at a particular organisation. This may be because the people that work there have knowledge of the issues you're researching.

Criticisms of this technique include failure to be representative. You may gain access to a group of individuals who are similar in terms of values, beliefs and opinion.

Snowball sampling

This type of sampling technique is so called because it's similar to rolling a snowball. The size of a snowball increases as it come into contact with snow. This sampling method is generated from one person who meets the research criteria, and then introduces the researcher to members of the same population. Snowball sampling is often used when a population is difficult to access (for example, intravenous drug users, homeless people and former prisoners).

Example: For a sample of homeless people, initial contact would be made with one person. This person would then introduce the researcher to another homeless person, and so on. In this way, you would build up a network of homeless people and generate a sample.

This technique has been criticised for being self-selecting. These samples may contain 'like minded' people who, because of their shared experience, have similar values and opinions. However, snowballing creates a sample that may have been difficult to access otherwise. This is a major advantage.


{{You can add more boxes below for links specific to this page [this note will not appear on user pages] }}

Skills for Learning FAQs