How to...
Conduct empirical research

What is empirical research?

Typically, empirical research embodies the following elements:

• A research question, which will determine research objectives.
• A particular and planned design for the research, which will depend on the question and which will find ways of answering it with appropriate use of resources.
• The gathering of primary data, which is then analysed.
• A particular methodology for collecting and analysing the data, such as an experiment or survey.
• The limitation of the data to a particular group, area or time scale, known as a sample: for example, a specific number of employees of a particular company type, or all users of a library over a given time scale. The sample should be somehow representative of a wider population.
• The ability to recreate the study and test the results. This is known as reliability.
• The ability to generalise from the findings to a larger sample and to other situations.

The research question

The starting point for your research should be your research question. This should be a formulation of the issue which is at the heart of the area which you are researching, which has the right degree of breadth and depth to make the research feasible within your resources. The following points are useful to remember when coming up with your research question, or RQ:

1. The RQ should arise from your research stream, or topic of interest. This may come from:
   • your doctoral thesis;
   • reading the relevant literature in journals, especially literature reviews which are good at giving an overview, and spotting interesting conceptual developments;
   • looking at research priorities of funding bodies, professional institutes etc.;
   • going to conferences;
   • looking out for calls for papers;
   • developing a dialogue with other researchers in your area.
2. To narrow down your research topic, brainstorm ideas around it, possibly with your colleagues if you have decided to collaborate, noting all the questions down.
3. Come up with a "general focus" question; then develop some other more specific ones.
4. Having come up with your RQs, check that:
   • they are not too broad;
   • they are not so narrow as to yield uninteresting results;
   • will the research entailed be covered by your resources, i.e. will you have sufficient time and money;
   • there is sufficient background literature on the topic;
   • you can carry out appropriate field research;
   • you have stated your question in the simplest possible way.

Let's look at some examples:

Bisking et al. examine whether or not gender has an influence on disciplinary action in their article Does the sex of the leader and subordinate influence a leader's disciplinary decisions? (Management Decision, Volume 41 Number 10) and come up with the following series of inter-related questions:

1. Given the same infraction, would a male leader impose the same disciplinary action on male and female subordinates?
2. Given the same infraction, would a female leader impose the same disciplinary action on male and female subordinates?
3. Given the same infraction, would a female leader impose the same disciplinary action on female subordinates as a male leader would on male subordinates?
4. Given the same infraction, would a female leader impose the same disciplinary action on male subordinates as a male leader would on female subordinates?
5. Given the same infraction, would a male and female leader impose the same disciplinary action on male subordinates?
6. Given the same infraction, would a male and female leader impose the same disciplinary action on female subordinates?
7. Do female and male leaders impose the same discipline on subordinates regardless of the type of infraction?
8. Is it possible to predict how female and male leaders will impose disciplinary actions based on their respective BSRI femininity and masculinity scores?

Motion et al. examined co-branding in Equity in Corporate Co-branding (European Journal of Marketing, Volume 37 Number 7/8) and came up with the following RQs:

RQ1: What objectives underpinned the corporate brand?

RQ2: How were brand values deployed to establish the corporate co-brand within particular discourse contexts?

RQ3: How was the desired rearticulation promoted to shareholders?

RQ4: What are the sources of corporate co-brand equity?

Note, the above two examples state the RQs very explicitly; sometimes the RQ is implicit:

Qun G. Jiao, Anthony J. Onwuegbuzie are library researchers who examined the question: "What is the relationship between library anxiety and social interdependence?" in a number of articles, see Dimensions of library anxiety and social interdependence: implications for library services (Library Review, Volume 51 Number 2).

Or sometimes the RQ is stated as a general objective:

Ying Fan describes outsourcing in British companies in Strategic outsourcing: evidence from British companies (Marketing Intelligence & Planning, Volume 18 Number 4) and states his research question as an objective:

The main objective of the research was to explore the two key areas in the outsourcing process, namely:

1. pre-outsourcing decision process; and
2. post-outsourcing supplier management.

or as a proposition:

Karin Klenke explores issues of gender in management decisions in Gender influences in decision-making processes in top management teams (Management Decision, Volume 41 Number 10).

Given the exploratory nature of this research, no specific hypotheses were formulated. Instead, the following general propositions are postulated:

P1. Female and male members of TMTs exercise different types of power in the strategic decision making process.

P2. Female and male members of TMTs differ in the extent in which they employ political savvy in the strategic decision making process.

P3. Male and female members of TMTs manage conflict in strategic decision making situations differently.

P4. Female and male members of TMTs utilise different types of trust in the decision making process.

Sometimes, the theoretical underpinning (see next section) of the research leads you to formulate a hypothesis rather than a question:

Martin et al. explored the effect of fast-forwarding of ads (called zipping) in Remote control marketing: how ad fast-forwarding and ad repetition affect consumers (Marketing Intelligence & Planning, Volume 20 Number 1) and his research explores the following hypotheses:

The influence of zipping
H1. Individuals viewing advertisements played at normal speed will exhibit higher ad recall and recognition than those who view zipped advertisements.

Ad repetition effects
H2. Individuals viewing a repeated advertisement will exhibit higher ad recall and recognition than those who see an advertisement once.

Zipping and ad repetition
H3. Individuals viewing zipped, repeated advertisements will exhibit higher ad recall and recognition than those who see a normal speed advertisement that is played once.

The theoretical framework

Empirical research is not divorced from theoretical considerations; and a consideration of theory should form one of the starting points of your research. This applies particularly in the case of management research which by its very nature is practical and applied to the real world. The link between research and theory is symbiotic: theory should inform research, and the findings of research should inform theory.

There are a number of different theoretical perspectives; if you are unfamiliar with them, we suggest that you look at any good research methods textbook for a full account (see Further information), but this page will contain notes on the following:

Positivism

This is the approach of the natural sciences, emphasising total objectivity and independence on the part of the researcher, a highly scientific methodology, with data being collected in a value-free manner and using quantitative techniques with some statistical measures of analysis. Assumes that there are 'independent facts' in the social world as in the natural world. The object is to generalise from what has been observed and hence add to the body of theory.

Empiricism

Very similar to positivism in that it has a strong reliance on objectivity and quantitative methods of data collection, but with less of a reliance on theory. There is emphasis on data and facts in their own right; they do not need to be linked to theory.

Interpretivism

This view criticises positivism as being inappropriate for the social world of business and management which is dominated by people rather than the laws of nature and hence has an inevitable subjective element as people will have different interpretations of situations and events. The business world can only be understood through people's interpretation. This view is more likely to emphasise qualitative methods such as participant observation, focus groups and semi-structured interviewing.

Quantitative methods:	Qualitative methods:
typically use numbers.	typically use words.
are deductive.	are inductive.
involve the researcher as ideally an objective, impartial observer.	require more participation and involvement on the part of the researcher.
may focus on cause and effect.	focuses on understanding of phenomena in their social, institutional, political and economic context.
require a hypothesis.	do not require a hypothesis.
have the drawback that they may force people into categories, also it cannot go into much depth about subjects and issues.	have the drawback that they focus on a few individuals, and may therefore be difficult to generalise.

Realism

While reality exists independently of human experience, people are not like objects in the natural world but are subject to social influences and processes. Like empiricism and positivism, this emphasises the importance of explanation, but is also concerned with the social world and with its underlying structures.

Inductive and deductive approaches

At what point in your research you bring in a theoretical perspective will depend on whether you choose an:

• Inductive approach – collect the data, then develop the theory.
• Deductive approach – assume a theoretical position then test it against the data.

The inductive approach:	The deductive approach:
is more usually linked with an interpretive approach.	is more usually linked with the positivist approach.
is more likely to use qualitative methods, such as interviewing, observation etc., with a more flexible structure.	is more likely to use quantitative methods, such as experiments, questionnaires etc., and a highly structured methodology with controls.
does not simply look at cause and effect, but at people's perceptions of events, and at the context of the research.	is the more scientific method, concerned with cause and effect, and the relationship between variables.
builds theory after collection of the data.	starts from a theoretical perspective, and develops a hypothesis which is tested against the data.
is more likely to use an in-depth study of a smaller sample.	is more likely to use a larger sample.
is less likely to be concerned with generalisation (a danger is that no patterns emerge).	is concerned with generalisation.
tresses the researcher involvement.	stresses the independence of the researcher.

It should be emphasised that none of the above approaches are mutually exclusive and can be used in combination.

Sampling techniques

Sampling may be done either:

• On a probability basis – that is, each member of a given population has an equal chance of being selected, as when your population is the workforce of an organisation, and you select members from it:
  • On a random basis – a given number is selected completely at random.
  • On a systematic basis – every nth element of the population is selected.
  • On a stratified random basis – the population is divided into segments, for example, in a University, you could divide the population into academic, administrators, and academic related. A random number of each group is then selected.
  • On a cluster basis – a particular subgroup is chosen at random.
• On a non-probability basis – the population does not have an equal chance of being selected; instead, selection happens according to some factor such as:
  • Convenience – being present at a particular time e.g. at lunch in the canteen.
  • Purposive – people can be selected deliberately because their views are relevant to the issue concerned.
  • Quota – the assumption is made that there are subgroups in the population, and a quota of respondents is chosen to reflect this diversity.

Useful articles

Richard Laughlin in Empirical research in accounting: alternative approaches and a case for "middle-range" thinking provides an interesting general overview of the different perspectives on theory and methodology as applied to accounting. (Accounting, Auditing & Accountability Journal, Volume 8 Number 1).

D. Tranfield and K. Starkey in The Nature, Social Organization and Promotion of Management Research: Towards Policy look at the relationship between theory and practice in management research, and develop a number of analytical frameworks, including looking at Becher's conceptual schema for disciplines and Gibbons et al.'s taxonomy of knowledge production systems. (British Journal of Management, vol. 9, no. 4 – abstract only).

Design of the research

Research design is about how you go about answering your question: what strategy you adopt, and what methods do you use to achieve your results. In particular you should ask yourself:

Where will your study be conducted, and what type of study?

What is the operational setting of your study, i.e. are you locating it within a particular context such as an organisation?

Are you conducting an exploratory study, obtaining an initial grasp of a phenomenon, a descriptive study, providing a profile of a topic or institution:

Karin Klenke provides an exploratory study of issues of gender in management decisions in Gender influences in decision-making processes in top management teams (Management Decision, Volume 41 Number 10).

Damien McLoughlin provides a descriptive study of action learning as a case study in There can be no learning without action and no action without learning (European Journal of Marketing, Volume 38 Number 3/4).

Or it can be explanatory , examining the causal relationship between variables: this can include the testing of hypotheses or examination of causes:

Martin et al. examined ad zipping and repetition in Remote control marketing: how ad fast-forwarding and ad repetition affect consumers (Marketing Intelligence & Planning, Volume 20 Number 1) with a number of hypotheses e.g. that people are more likely to remember an ad that they have seen repeatedly.

What research methods will you be using?

Methods are are "a systematic and orderly approach taken towards the collection and analysis of data so that information can be obtained from those data" (Jankowicz, 2000: 209), whereas techniques are "particular, step-by-step procedures which you can follow in order to gather data, and analyse them for the information they contain" (Jankowicz, 2000: p. 211). The main research methods will be discussed in the next section and are:

• Experiment.
• Survey.
• Case study.
• Grounded theory.
• Ethnographic and observation.
• Action research.

Note it is possible, and indeed desirable, to use more than one method: this is called triangulation and has the benefit of being able to enhance the validity of the results.

Over what time period will your research take place?

Should the research be a "snapshot", examining a particular phenomenon at a particular time, or should it be longitutinal, examining an issue over a time period? If the latter, the object will be to explore changes over the period.

A longitudinal study of corporate social reporting in Singapore (Eric W K Tsang, Accounting, Auditing & Accountability Journal, Volume 11 Number 5) examines social reporting in that country from 1986 to 1995.

How large will your sample be? What will your unit of analysis be?

The sample refers to the subset of your population (the total group you wish to investigate). The sample should be sufficiently large to be representative of the population as a whole.

The unit of analysis is the level at which the data is aggregated: for example, it could be a study of individuals as in the women manager studies quoted above, of dyads, as in a study of mentor/mentee relationships, of groups (as in studies of departments in an organisation), of organisations, or of industries.

What techniques will you use to collect and analyse the data?

This refers to techniques for the capture and analysis of data, such as:

• Interviews (structured and semi-structured).
• Structured questionnaires.
• Observation.

How will you ensure the reliability and generalisability of your research?

Finally, it's important to be aware of four things at all stages of your research; without them, your research will fall flat on its face.

1. reliability.
2. validity.
3. generalisability.
4. transferability.

Reliability

This is about the replicability of your reseach and the accuracy of the procedures and research techniques. Will the same results be repeated if the research is repeated? Are the measurements of the research methods accurate and consistent? Could they be used in other similar contexts with equivalent results? Would the same results be achieved by another researcher using the same instruments? Is the research free from error or bias on the part of the researcher, or the participants? (E.g. do the participants say what they believe the management, or the researcher, wants? For example, in a survey done on some course material, that on a mathematical module received glowing reports – which led the researcher to wonder whether this was anything to do with the author being the Head of Department!)

Validity

How successfully has the research actually achieved what it set out to achieve? Can the results of the study be transferred to other situations? Does x really cause y, in other words is the researcher correct in maintaining a causal link between these two variables? Is the research design sufficiently rigorous, have alternative explanations been considered? Have the findings really be accurately interpreted? Have other events intervened which might impact on the study, e.g. a large scale redundancy programme? (For example, in an evaluation of the use of CDs for self study with a world-wide group of students, it was established that some groups had not had sufficient explanation from the tutors as to how to use the CD. This could have affected their rather negative views.)

Generalisability

Are the findings applicable in other research settings? Can a theory be developed that can apply to other populations? For example, can a particular study about dissatisfaction amongst lecturers in a particular university be applied generally? This is particularly applicable to research which has a relatively wide sample, as in a questionnaire, or which adopts a scientific technique, as with the experiment.

Transferability

Can the research be applied to other situations? Particularly relevant when applied to case studies.

Methods of empirical research

The last page discussed general design issues; now we look at systematic ways of approaching data collection before describing some procedures for collecting data.

Quantitative vs. qualitative

First, however, it is important to distinguish quantitative and qualitative approaches to data collection:

Quantitative methods:	Qualitative methods:
typically use numbers.	typically use words.
are deductive.	are inductive.
involve the researcher as ideally an objective, impartial observer.	require more participation and involvement on the part of the researcher.
may focus on cause and effect.	focuses on understanding of phenomena in their social, institutional, political and economic context.
require a hypothesis.	do not require a hypothesis.
have the drawback that they may force people into categories, also it cannot go into much depth about subjects and issues.	have the drawback that they focus on a few individuals, and may therefore be difficult to generalise.

Main methods used in empirical research

Experiment – an experiment involves deliberately testing a hypothesis and reaching a conclusion, by creating a situation where one of the variables is manipulated: what happens to one variable (usually called the independent variable) when another variable (usually called dependent) is removed or altered. It starts with a hypothesis, then tests it, analysing the resultant data and reporting the findings.

Pike et al. describe a student experiment in recycling in Science education and sustainability initiatives (International Journal of Sustainability in Higher Education, Volume 4 Number 3).

Survey – this method involves collecting a large amount of data from a large population, most usually by questionnaires or structured interviews. Most usually it is a quantitative method, involving "closed" questions with a predetermined number of answers. These are in fact much easier to fill in, and therefore more likely to get a high response rate, as does keeping the questionnaire short. It's a good idea to trial the survey to ensure ease of completion and lack of ambiguity.

Case study – these are much used in business research, and involve looking at a particular set of issues in a particular context in a particular organisation or part of an organisation. There are many case studies published in Emerald journals, and to access some examples and read more about this method, go to How to write a case study.

Ethnographic and observational methods – as the term suggests, this has its roots in anthropology and requires involvement in the setting of the research. Various forms of observation are much used in management research, although they can be time-consuming. It is most usually a qualitative method, although it can be used quantitatively if highly structured. Often done at exploratory stages of research. It is particularly useful when watching people interacting with something, for example students interacting with learning material, people interacting with their environment in a shopping precinct or leisure centre.

Mathews and Boote, in Saying is one thing; doing is another: the role of observation in marketing research (Qualitative Market Research: An International Journal, Volume 2 Number 1) provide an excellent description of observation techniques as applied to marketing research, with a classification and a case study involving the siting of a restaurant.

Vinten describes use of participant observation (when the researcher is directly involved) in Participant Observation: A Model for Organizational Investigation? (Journal of Managerial Psychology, Volume 9 Number 2).

Grounded theory – this is a research approach where there is an initial observation with minimal preconceptions followed by the generation of a hypothesis, theory or prediction, which is then further tested. Its use of data is therefore iterative, with theory being grounded and refined as further data is sought. As a method it is initially inductive, but can become deductive at a later stage.

Leonard and McAdam explore grounded theory in Grounded theory methodology and practitioner reflexivity in TQM research (International Journal of Quality & Reliability Management, Volume 18 Number 2).

Action research – This occurs in situations where people are ostensibly reflecting on their own work and self-consciously trying to improve practice and performance. There will here be close collaboration between the practitioner and the researcher, and a strong focus on change.

Vinten provides a definition of action research in Participant Observation: A Model for Organizational Investigation? (Journal of Managerial Psychology, Volume 9 Number 2).

Techniques of data collection & analysis

First, it is necessary to differentiate between two different types of data:

There's a lot more to this article; just fill in the form below to instantly see the complete content.