How to...
Conduct empirical research

Empirical research is research that is based on observation and measurement of phenomena, as directly experienced by the researcher. The data thus gathered may be compared against a theory or hypothesis, but the results are still based on real life experience. The data gathered is all primary data, although secondary data from a literature review may form the theoretical background.

What is empirical research?

Typically, empirical research embodies the following elements:

  • 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:


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.


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.


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.


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.

Image: warning signNote 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.


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!)


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.)


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.


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 and analysis

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

Structured data is highly organised, and can easily be coded and analysed using statistics. Unstructured data is disorganised, often generated by open questions, observation etc. and requires work before coding.
Collected by structured techniques. Collected by semi-structured techniques.

Structured techniques

These present a series of questions with a choice of pre-set alternative answers.

Questionnaires – are used frequently in survey research, and can be printed and sent out by post, done over the telephone, or increasingly over the Internet. They represent an economical way of contacting a large number of people. Very often, they will be used to collect highly structured data and use closed questions (yes/no, choice of boxes e.g. for age, or Likert scale). (To achieve a higher response rate, make the questionnaire short and easy to fill in.)

A questionnaire with a series of statements next to a Likert scale, strongly agree = 4, strongly disagree = 1:

  • Navigating the CD is easy.
  • The design of the CD is attractive.
  • There is sufficient technical help.
  • The exercises are well laid out.
  • The exercises have sufficient instructions.
  • The exercises have sufficient feedback.
  • The exercises helped me learn etc.

The returned questionnaires were read into a machine and the responses recorded as percentages.

Jiao and Onwuegbuzie used the Library Anxiety Scale, a 43-item, 5-point Likert-format instrument which assesses levels of library anxiety, with five subscales – barriers with staff, affective barriers, comfort with the library, knowledge of the library, and mechanical barriers, along with the Social Independence Scale, a 23-item, 5-point Likert format instrument measuring individuals' cooperative, competitive, and individualistic perceptions, in  Dimensions of library anxiety and social interdependence: implications for library services (Library Review, Volume 51 Number 2).

Repertory grid – This is similar to the questionnaire using a Likert rating, but the person/institution interviewed is involved in the design of the questionnaire.

Peters in Repertory Grid as a Tool for Training Needs Analysis (The Learning Organization, Volume 1 Number 2) describes the use of this tool, and also gives a definition of the term. 

Structured interview – this will use a series of identical questions for each respondent. This will proceed according to a written guide but will also contain directions as to sequence, and possibly explanations of certain terms.

Structured observation – this is usually either effected by a machine (for instance, EPoS which tracks product sales) as opposed to a human, or by means of time-sampling, i.e. the observation of phenomenon over a particular period of time.

Analysis of structured data

Because the data conforms to a series of pre-set categories, it can be coded using symbols (usually numbers but occasionally letters). Analysis is usually done by statistics. The two main forms of statistical analysis are:

  • Descriptive – the statistics will either be used to reveal patterns and show differences among variables.
  • Inferential – the statistics will be used to draw conclusions, explain cause and effect, and make predictions. Here, the researcher is interested in the relationship between dependent variables (whose score depends on or is affected by another variable) and independent variables (who determines the score of a dependent variable).

It is best to enter the data onto a spreadsheet such as Excel, and then use a statistical package such as SPSS or Minitab to carry out the analysis.

Jiao and Onwuegbuzie in the article quoted above use statistical measures to analyse the results of their surveys. In particular they looked at how the various subscales of the two questionnaires might correlate with one another.

Semi-structured techniques

These all allow for responses which do not conform to a set pattern.

Interviews – these may be:

  • semi-structured – this is where you have a "theme" for your interview, and some carefully defined questions, but leave open the possibility of discussion of other cognate areas.
  • key informant – there may be several people who are key to your research, whom it will be important to interview to gain key information about the project. May be used at an exploratory stage of the research, when the issues are being identified.

Hansen and Willcox use semi-structured interviews in Cultural assumptions in career management: practice implications from Germany (Career Development International, Volume 2 Number 4).

Ove C. Hansemark, Marie Albinsson also use semi-structured interviews in Customer satisfaction and retention: the experiences of individual employees (Managing Service Quality, Volume 14 Number 1) analysing the data according to differences and patterns.

Focus groups – in this technique, a group of people are deliberately assembled to discuss an issue, and the discussion recorded and then analysed.

Von Seggern and Young look at the focus group's use in librarianship research in The focus group method in libraries: issues relating to process and data analysis (Reference Services Review, Volume 31 Number 3), providing a description of the methodology of the focus group, as well as an interesting account of the process of data analysis using a piece of software.

Montoya-Weiss et al. describe online focus groups in On-line focus groups: conceptual issues and a research tool (European Journal of Marketing, Volume 32 Number 7/8).

Unstructured observation – while some software packages can be used, more traditional forms of data capture include notes, reports, and diaries (from the participants).

Conversations – This refers to situations when you talk informally to people about your research topic, for example during visits to an organisation you are investigating you may find yourself in conversation with people whom you didn't plan to interview. You may be able to steer them informally towards topics relevant to your research.

Analysis of unstructured data

This will be more difficult, though not impossible, to code, and a key skill involves exploring the data for patterns, concepts and themes. A common technique for analysing unstructured data is content analysis . This involves analysing the content according to particular categories. The stages are to first determine the categories, then define them and provide key indicators, i.e. phrases in the material which would indicate the category. It is also important to determine the unit of analysis, which may be the word, the sentence or the paragraph.

Guthrie et al. in Using content analysis as a research method to inquire into intellectual capital reporting (Journal of Intellectual Capital, Volume 5 Number 2) looks at content analysis in this area and also provides comments on the technique.

Perry and Bodkin in Content analysis of Fortune 100 company Web sites (Corporate Communications: An International Journal, Volume 5 Number 2) apply content analysis to websites.

Content analysis is often used for printed material and websites, which we have not discussed in these pages, but which can provide important secondary sources.

Image: warning signNote it is possible, and indeed desirable, to combine qualitative and quantitative data techniques: this strengthens the analysis and makes for a richer picture.

Fan, in Strategic outsourcing: evidence from British companies (Marketing Intelligence & Planning, Volume 18 Number 4) describes the use of a postal survey followed by the use of interviews to obtain further data.

Reporting the findings of empirical research

There are obviously a number of possible ways of reporting research findings but we are here assuming that you intend to submit a paper to an Emerald journal. Given the amount of attention devoted to preparing and presenting academic articles elsewhere in this site, it is not proposed to go into too much detail here.

The "Notes for Contributors" to Emerald journals state that you should describe the methodology under a separate heading. The methodology should describe the process of achieving the research objectives and answering the research question, as well as the methods used.

Kuo et al. provide a good example of research methodology in A case study assessment of performance measurement in distribution centers (Industrial Management & Data Systems, Volume 99 Number 2). Note how they describe the type of research (i.e. exploratory and descriptive), state the unit of analysis (the distribution center measurement system), the theoretical focus (inductive), the research approach (case study), and types of data (both quantitative and qualitative).

The findings should encompass the main conclusions as developed from an analysis of the data. Note that you do NOT need to describe either the research instruments, or the data they generate, in great detail.

Most editors and peer reviewers in selecting articles for publication put the "So what?" factor high on their list of criteria. In other words, they want to see what the research adds to the whole body of knowledge, and you need to bring this out. Most Emerald journals also aim to appeal to practitioners, so you should also bring out implications for practice.

Overall, the best advice is to read carefully the "Notes for contributors" of the journal you have targeted for your paper to see if they say anything about the research approach and description of methodology. You should of course look at articles published in that and other journals to see how other researchers present their research.

Further information


There are a number of well-written books on business research methods, which are generally aimed at an undergraduates and MBA students writing projects. If you want something more detailed, you will need to go to a book on a particular methodology.

Research Methods for Business Students, Third Edition
Mark Saunders, Philip Lewis, and Adrian Thornhill
(FT Prentice Hall, 2004, ISBN 1-4058-1397-0)

Particularly good on statistical analysis.

The Management of a Student Research Project
John Sharp, John Peters and Keith Howard
(Ashgate, 2003, ISBN 0-566-08490-2)

Also covers doctoral research.

Business Research Projects
Devi Jankowicz
(Thomson Learning, 1997, ISBN 1-844-80082-2) 

Good on qualitative methods.

Research Methods for Business: A Skill-building Approach
Uma Sekaram
(Wiley, 2002, ISBN 0-471-38448-8)

Is concerned with a scientific approach to business, and quantitative methods.

Surviving Your Thesis
edited by Suzan Burton and Peter Steane
(Routledge, 2004, ISBN 0-415-32222-7)

Mainly concerned with a whole range of issues concerning the PhD thesis, including finding a supervisor, funding etc., but does include a couple of good chapters on qualitative and quantitative research respectively.


There are also some good websites, again, mainly written with students in mind.

Empirical research
A good introduction to the subject developed by Andrew Roberts of the University of Middlesex.

Writing guide on empirical research
Note that these writing guides are written by English PhD students so inevitably biassed towards English, but still has some good information.

Online textbook about research methods
by William M. Trochim of Cornell University.