Chapter 10: Research planning and design
A research design relates directly to the answering of a question. In
quantitative research, this is a detailed outline for the testing of a hypothesis,
in clear and definite terms. It specifies the operations which need to be
performed to test hypotheses under given conditions. Don’t confuse this with
research management, which is a plan to guide the researcher through the
research process. In qualitative research, the research design is more flexible
and repetitive.
The research design ensures a high internal validity, which examines the
extent to which a particular research design has excluded all other possible
hypotheses which could explain the same phenomenon. It asks, “Do the
changes in the dependent variable definitely relate to the independent
variable?” This eliminates other explanations for the relationship between two
factors or variables. In quantitative research, the research will have a high
internal validity if it controls as many extraneous variables as possible. In
qualitative research, it is also known as credibility and refers to whether the
researchers’ method of data collection and analysis is related to and answers
the research question.
The most common research topics are:
1. Conditions: These are studied when the researcher wants to understand the
current state of the participants. For example, a researcher that is interested in
the health of elderly people in rural areas might measure their heart rates after
light exercise. This kind of research is usually more quantitative
2. Orientations: These are concerned with participants’ attitudes, beliefs and
lifestyles. This tends to be more qualitative as it is tough to measure views in
terms of numbers and statistics.
3. Actions: These may be observed directly or reported by the person
themselves, or others who observed the individual’s actions. An example of this
would be observing which method of transport people use to get to work every
day.
Even though researchers might focus on one of the above topics, they are all
interconnected.
The unit of analysis is the person or object that the researcher collects data
from. It may be on three different levels – a micro level (an individual or group
of people), a mezzo level (organisations and communities) or a macro level
(national). The way that the unit of analysis is chosen is very different between
qualitative and quantitative research. In quantitative studies, a group is chosen
that represents the general population (e.g. diverse ethnic, cultural, religious
and socio-economic backgrounds). Participants are selected through probability
samples. In qualitative studies, the unit of analysis (participants) are chosen
selectively according to what is being studied. If a researcher wants to know
the effect that different religious views have on ones attitudes, he might only
choose people from different religions and not irreligious people. The main
units of analysis include:
Individuals: This is the most commonly used one. Researchers investigate
the conditions, orientations, actions, attitudes and beliefs of a group of
individuals who are usually selected quite specifically.
, Groups of people: Examples of this include studies done on twins,
married couples, families, etc. Groups of people are studied and
compared.
Organisations: This might include businesses, non-profit organisations,
etc. Questions of interest might relate to working conditions, employees
health, etc.
Periods of time: This is a strange concept, but an example of this would
be when researchers study the change in infant mortality over a 20-year
period. Each year is a unit, but the 2—year span is the unit of analysis. It
is what’s being analysed.
Social artefacts: These are the products of human beings, and can be
anything from songs and letters to cars and farm tools. These artefacts
can often tell you valuable information about the people that used them.
If a researcher draws a conclusion about one unit of analysis while the research
is focused on a different unit of analysis, it is called the ecological fallacy. For
example, if the researcher does a test on women's perceptions of birth control,
and then he cannot claim findings about how people feel about oral
contraceptives because there were no men in his study. He was studying one
group of people and making claims about another.
On the other hand, sometimes researchers focus solely on one unit of analysis
and ignore others entirely – this is known as reductionism. We need to
remember that everything is interconnected.
The way a research design works with regards to time is also important. There
are two types of
1. Cross-sectional research designs: This is when all the data is collected at the
same time.
2. Longitudinal designs: This is when the data is collected over a period of time.
Two different types of longitudinal design include:
Cohort studies: This uses a type of longitudinal design that tracks
particular age groups over time. An example of this is the Birth-to-Twenty
project, where researchers are tracking the development of a certain
group of children from birth until they turn twenty.
Tracer studies: This traces people or follows their lives over a period of
time. Often data is collected at only one point (which is mostly cross-
sectional). Therefore the tracer study is not just longitudinal, but
produces data that is similar to a longitudinal design.
Research designs must be specifically designed to suit the problem or research
question. Research designs have two main components. The first component is
observation – when the researcher observes the variables and units of analysis.
The second component is the analysis of the relationships between variables.
In quantitative research you would manipulate variables to see what effect
they have on each other; in qualitative research you would compare
information, and possibly do more research to clarify the results. There are
three categories of research design for qualitative research:
1. Pre-experimental (exploratory and descriptive) designs are the methods of
qualitative research. They try to describe and understand a phenomenon.
These designs have little requirements and are the last choice, as they are the
least scientifically accurate. They also often use small, non-probability samples.