HOME | CURRENT | ARCHIVES | FORUM

Research World, Volume 6, 2009
Online Version


Article S6.10

Conducting an Empirical Research Project

Seminar Leader: Malay Kumar Dey
Cotsakos School of Business, William Paterson Universiy, New Jersey, USA
mdey77[at]yahoo.com

Empirical Research is based on experimentation or observation, which provide evidence. Such research is often conducted to answer a specific question or to test a hypothesis (an educated guess). Empirical research uses data, either from secondary sources or from primary sources. The process of empirical research includes the following activities:

(a) Picking up a topic and framing the research question
(b) Reviewing the relevant literature
(c) Framing an empirical strategy
(d) Collecting data and formulating a model
(e) Interpretating the results
(f) Preparing a report

1. Topic and Research Question

Empirical research normally starts with some prior theory to explain what happens in the real world and predict what might happen in future. Alternatively, empirical research can also start with a question about the nature of some real-world phenomenon. The purpose of empirical research is to test the theory or find an answer to the question. Of course, usually the answer raises new questions. In some cases, research is conducted to build new theory (the so-called grounded theory approach). The research topic can come from many sources:

(a) current news, for example, the recent financial crisis, its causes and effects
(b) relevant literature, for example, abstracts of articles in scholarly or professional journals in one’s field
(c) in-house resources, for example, seminars and academic/professional meetings

While choosing a research topic, the researcher should have a justification in mind regarding the importance of the topic and the potential beneficiaries of the research. Moreover, every important question cannot become a suitable research topic. It should be doable, relevant, and somewhat broad. A research topic/question can be considered important if has some of the following features:

(a) it is of contemporary interest
(b) it relates to an area where the available understanding seems inadequate
(c) it may call for new analytical tools and techniques

2. Literature Review

A literature review is an account of what has been published on the chosen topic by accredited scholars and researchers. Occasionally, the literature review comes as a separate assignment, but more often it is part of the introduction to the research report or thesis. In writing the literature review, the researcher's purpose is to convey to the reader the knowledge and ideas on the chosen topic which are already available and considered well established. As a piece of writing, the literature review must be structured carefully sp as to clarify the research issue, the research objectives, and the research direction to be pursued. It is not a mere description of the material available. Besides enlarging the researcher's knowledge about the topic, writing a literature review helps a researcher develop skills in the following two areas:

(a) information seeking: the ability to scan the literature efficiently, using manual or computerised methods, to identify a set of relevant material
(b) critical appraisal: the ability to choose the higher-quality material from those identified (for example, those which are unbiased and valid)

A literature review must be organised around and related directly to the research topic or question, synthesise the material into a reasonable coherent narrative, identify areas of controversy in the literature, and formulate questions that need further research. Apart from books and academic journals, Web-based sources such as EconLit, ABI Inform, and SSRN may be used to track useful literature on the chosen topic. Working papers of different leading academic institutions, home pages of contemporary researchers and scholars, academic and professional meetings, journal databases, and dissertation indices of academic institutes may also be referred.

3. Empirical Strategy

Defining the empirical strategy is the most difficult step. At this stage, further research questions need to be derived from the principal research topic or question. For example, consider the principal research question: “Has a training programme for low-skilled workers worked?” The relevant questions to be asked at this stage could be the following:

(a) Who are "low-skilled workers"?
(b) What does “worked” mean? Is it about extending their skills, increasing their wages, reducing their levels of unemployment, or increasing their job satisfaction?
(c) Why is the question important? Is it because of cost or is it because similar programmes have mixed reviews, or is it because of labor market equilibrium issues (i.e., pertaining to the labour supply and labour demand that determines wages)?

A suitable empirical strategy is to be devised to test each of the above subquestions. For example, to test whether the programme has resulted in higher wages, one can compare the means of wages before and after the programme was implemented. Similarly, the chi square test may be required to compare the distribution of wages before and after the programme. The literature review often provides the necessary guidance in this regard.

4. Data and Model

There are different forms of quantitative data, such as cross-sectional data (e.g., sales of various firms during a particular quarter) time series (e.g., stock prices for a single company across various points in time). A panel data set includes both time series and cross sections. An empirical researcher has to be very careful about collecting the appropriate kind of data in the appropriate manner. There should be enough justification for using particular data sources. Besides, the researcher needs to be clear about any specific features of the data that may affect the results, for example, the data may over-represent a certain demographic group or the data may be plagued by what is known as self-selection bias.

The empirical strategy leads to a statistical model which is appropriate for the data available. Statistical models depict the algebraic relationship among a set of random variables. For example, consider the following model:

Yit = α + µi + ηt

Here µ and η are both random variables with some statistical properties (or parameters), say mean and variance. The subscripts i and t are cross-sectional and time-series indicators respectively. Accordingly, Y is also a random variable and its parameters depend on the means and variances of µ and η. Certain statistical/econometric adjustments need to be made for using certain kinds of model, for example, in time-series models adjustments need to be made for nonstationarity. Like in case of data, the researcher needs to be clear about any specific features of the model that may affect the results, for example, the model may be plagued by what is known as specification bias.

5. Research Report

The research report should be clear and concise and should indicate the following:

(a) Question asked and its importance: The way a question is framed and its importance relating to specific objectives should be explained here.
(b) Empirical methodology: In empirical methodology, the nature of research design, the sampling plan, and data collection and analysis procedures should be described.
(c) Statistical hypotheses: A hypothesis is an educated assumption on which the researcher bases the inquiry. The report should clarify whay those hypotheses were considered important enough to be studied.
(d) Data source and collection: The data sources and the methods of data collections are mentioned here. In case of primary data, the method of collection should be specified very clearly. Any specific features of the data (including any biases) are clarified.
(e) Descriptive statistics: The descriptive statistics such as mean, median, mode, minimum, maximum, standard deviation, skewness, kurtosis, and so forth, are given.
(f) Modelling: The statistical model used to analyse the data is presented, with any justification for its selection. Any specific features of the model (including any biases) are clarified.
(g) Conclusions: The conclusions should be brief and specific. These should be about the results that are found out from the particular empirical research undertaken. These should indicate how far the results can answer the research questions.

Reported by Rohita Kumar Mishra, with inputs from Soumya Guha Deb and D. P. Dash. [March 10, 2009]


Copyleft The article may be used freely, for a noncommercial purpose, as long as the original source is properly acknowledged.

Xavier Institute of Management, Xavier Square, Bhubaneswar 751013, India
Research World (ISSN 0974-2379) http://www1.ximb.ac.in/RW.nsf/pages/Home