Quantitative approaches to research design

What is my research question, and what variables am I interested in exploring?
What is the hypothesis?
What are the appropriate measures to use?
How will you analyse the data?
What are the instruments used in quantitative research?
What are the advantages and drawbacks of quantitative research?

The thing that characterizes quantitative research is that it is objective. The assumption is that facts exist totally independently and the researcher is a totally objective observer of situations, and has no power to influence them. At such, it probably starts from a positivist or empiricist position.

The research design is based on one iteration in collection of the data: the categories are isolated prior to the study, and the design is planned out and generally not changed during the study (as it may be in qualitative research).

In planning quantitative studies, it is important to consider the following:

What is my research question, and what variables am I interested in exploring?

It is usual to start your research by carrying out a literature review, which should help you formulate a research question.

Part of the task of the above is to help you determine what variables you are considering. What are the key variables for your research and what is the relationship between them - are you looking to explore issues, to compare two variables or to look at cause and effect?

The Dutch heart health community intervention "Hartslag Limburg": evaluation design and baseline data describes a trial of a cardiovascular prevention programme which indicated the importance of its further implementation. The key variables are the types of health related behaviours which affect a person's chance of heart disease.

The following studies compare variables:

Service failures away from home: benefits in intercultural service encounters compares service encounters (the independent variable) inside and outside Taiwan (the dependent variable) in order to look at certain aspects of 'critical incidents' in intercultural service encounters.

The concept of fit in services flexibility and research: an empirical approach looks at managerial flexibility in relation to different types of business, service and manufacturing.

It can also look at cause and effect:

In 'Remote control marketing: how ad fast-forwarding and ad repetition affect consumers' which looks at two variables associated with advertising, notably zipping and fast forwarding, and in their effect on a third variable, consumer behaviour - i.e. ability to remember ads. Furthermore, it looks at the interaction between the first two variables - i.e. whether they interact on one another to help increase recall.

What is the hypothesis?

It is usual with quantitative research to proceed from a particular hypothesis. The object of research would then be to test the hypothesis.

In the example quoted above, 'Remote control marketing: how ad fast-forwarding and ad repetition affect consumers' , the researchers decided to explore a neglected area of the literature: the interaction between ad zipping and repetition, and came up with three 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.

Martin et al., Marketing Intelligence & Planning, Volume 20 Number 1 2002 pp. 44-48

What are the appropriate measures to use?

It is very important, when designing your research, to understand what you are measuring. This will call for a close examination of the issues involved: is your measure suitable to the hypothesis and research question under consideration? The type of scale you will use will dictate the statistical procedure which you can use to analyse your data, and it is important to have an understanding of the latter at the outset in order to obtain the correct level of analysis, and one that will throw the best light on your research question, and help test your hypothesis.

It is also important to understand what type of data you are trying to collect. Are you wanting to collect data that relates simply to different types of categories, for example, men and women (as in, say, differences in decision-making between men and women managers), or do you want to rank the data in some way? Choices as far as the nature of data are concerned again dictate the type of statistical analysis.

Data can be categorized as follows:

Nominal Representing particular categories, e.g. men or women.
Ordinal Ranked in some way such as order of passing a particular point in a shopping centre.
Interval Ranked according to the interval between the data, which remains the same. Most typical of this type of data is temperature.
Ratio Where it is possible to measure the difference between different types of data - for example applying a measurement.
Scalar This type of data has intervals between it, which are not quantifiable.

Note that some of the above categories, especially 'interval' and 'ratio' are drawn from a scientific model which assumes exact measurement of data (temperature, length etc.). In management research, you are unlikely to want to or be able to apply such a high degree of exactitude, and are more likely to be measuring less exact criteria which do not have an exact interval between them.

Here are some examples of use of data in management research. This one illustrates the use of different categories:

'The concept of fit in services flexibility and research: an empirical approach' uses an approach which itemizes the different aspects which the researchers wished to measure flexibility mix, performance and the form's general data.

This one looks at categories and also at ranked data (ordinal):

For example, in 'Remote control marketing: how ad fast-forwarding and ad repetition affect consumers', the measure involved 2 (speed of ad presentation: normal, fast-forwarded) ×\ 2 (repetition: none, one repetition) between-subjects factorial design.

The following examples look at measures on a scale, which may relate to tangible factors such as frequency, or more intangible ones which relate to attitude or opinion:

How many holidays do you take in a year?

One Between 2 and 5 Between 5 and 10 More than 10

Tick the box which most agrees with your views:

Navigating my way around the CD was:

Very easy Easy Neither easy nor hard Hard Very hard

The later type of data are very common in management research, and are known as scalar data. A very common measure for such data is known as the Likert scale:

Strongly agree Agree Neither agree nor disagree Disagree Strongly disagree

How will you analyse the data?

Quantitative data are invariably analysed by some sort of statistical means, such as a t-test, a chi test, a , regression, cluster analysis etc. It is very important to decide at the planning stage what your method of analysis will be: this will in turn affect your choice of measure. Both your analysis and measure should be suitable to test your hypothesis.

You need also to consider what type of package will you need to analyse your data. It may be sufficient to enter it into an Excel spreadsheet, or you may wish to use a statistical package such as SPSS or Mintab

What are the instruments used in quantitative research?

Or, put more simply, what methods will you use to collect your data?

In scientific research, it is possible to be reasonably precise by generating experiments in laboratory conditions. Whilst the field experiment has a place in management research, as does observation, the most usual instrument for producing quantitative data is the survey, most often carried out by means of a questionnaire.

You will find numerous examples of questionnaires and surveys in research published by Emerald, as you will in any database of management research. Questionnaires will be discussed at a later stage but here are some key issues:

  • It is important to know exactly what questions you want answers to. A common failing is to realise, once you have got the questionnaire back, that you really need answers to a question which you never asked. Thus the questionnaire should be rigorously researched and the questions phrased as precisely as possible.
  • You are more likely to get a response if you give people a reason to respond - commercial companies sometimes offer a prize, which may not be possible or appropriate if you are a researcher in a university, but it is usual in that case to give the reason behind your research, which gives your respondent a context. Even more motivational is the ease with which the questionnaire can be filled in.
  • How many responses will I need? This concerns the eventual size of your dataset and depends upon the degree of complexity of your planned analysis, how you are treating your variables (for example, if you are wanting to show the effect of a variable, you will need a larger response size, likewize if you are showing changes in variables).

Other instruments that are used in quantitative research to generate data are experiments, historical records and documents, and observation.

Note that some authors claim that for a design to be a true experiment, items must be randomly assigned to groups; if there is some sort of control group or multiple measures, then it may be quasi experimental. If your survey fits neither of these descriptions, it may according to these authors be sufficient for descriptive purposes, but not if you seek to establish a causal relationship.

For more information on types of design, see William Trocken's Research Methods Knowledge Base1 section on types of design.

What are the advantages and drawbacks of quantitative research?

The main advantage of quantitative research is that it is easy to determine its rigour: because of the objectivity of quantitative studies, it is easy to replicate them in another situation. For example, a well-constructed questionnaire can be used to analyse job satisfaction in two different companies; likewise, an observation studying consumer behaviour in a shopping centre can take place in two different such centres.

Quantitative methods are also good at obtaining a good deal of reliable data from a large number of sources. Their drawback is that they are heavily dependent on the reliability of the instrument: that is, in the case of the questionnaire, it is vital to ask the right questions in the right way. This in turn is dependent upon having sufficient information about a situation, which is not always possible. In addition, quantitative studies may generate a large amount of data, but the data may lack depth and fail to explain complex human processes such as attitudes to organizational change, or how how learning takes place.

For example, a quantitative study on a piece of educational software may show that on the whole people felt that they had learnt something, but may not necessarily show how they learnt, which an observation could.

For this reason, quantitative methods are often used in conjunction with qualitative methods: for example, qualitative methods of interviewing may be used as a way of finding out more about a situation in order to draw up an informed quantitative instrument; or to explore certain issues which have appeared in the quantitative study in greater depth.

1Trochim, William M. The Research Methods Knowledge Base, 2nd Edition. Internet WWW page, at URL: <http://trochim.human.cornell.edu/kb/index.htm> (version current as of August 16, 2004).

Qualitative approaches to research design