What makes good research?

SOCIAL SCIENCE FOR SCHOOLS
Methodologies
What makes good research?
To be scientific, social research follows definite rules that have been developed over time with the aim of ensuring
that the results are not just the product of the researcher’s own beliefs. Social scientists debate the details of these
rules, but there is general agreement about some key principles:
The research must be valid
The research should be reliable
This does not mean ‘is it worthwhile?’ Rather it means
whether the research actually captures or measures
what we set out to examine in the first place. For
example, we might be interested in measuring and
comparing how much work men and women do.
We would need to define what we meant by ‘work’
carefully. If we only counted employment, or work done
for payment, we would get different results than if we
included unpaid work such as housework or childcare.
And what about other activities? Would we include
gardening, or is that a hobby? What about shopping?
We might think of a visit to the supermarket for
groceries as work. But what about a Saturday afternoon
spent buying a new computer game?
This means that if our methods are good, we ought to
get the same results when we apply them to different
examples of the same situation. A simple example is a
ruler. If we use it to measure two things with the same
length, it will give the same result. A good example in
social research are questions about voting behaviour.
Simply asking adults whether they voted in the last
election does not give reliable results, as some people
who did not vote will tell an interviewer that they
voted, because it is a more socially acceptable answer;
some people can feel embarrassment in reporting to a
stranger that they did not vote. Researchers correct for
this bias by asking questions worded to make it easier
to report non-voting.
It is sometimes easier to ensure validity in qualitative
research where all the relevant details of a respondent’s
situation can be explored. In survey based quantitative
research, where we usually have to rely on a small
number of briefly answered questions, we need to
design and test these questions carefully to makes sure
that the respondents understand them in the same way
as the researcher expects them to. the Understanding
Society survey defined bullying as:
o
Hit kick or push you
o
Take your belongings
o
Call you nasty names
o
Make fun of you
If research is reliable it should be possible for different
researchers to replicate or repeat it and obtain the
same results. In the natural sciences this can be
straightforward since the way the natural world
behaves does not change over relatively short time
spans . In the social sciences this is more difficult for
two reasons. First societies do change over time or
space. If we obtain different results it may be because
we are studying something different, not because our
survey is unreliable. Second, unlike inanimate objects,
people are usually aware of being the subjects of
research. They may change their beliefs or behaviour,
or what they tell researchers, because of knowledge of
previous research.
Do you think this is a valid definition?
Note that research can be reliable without being valid! If we had a badly calibrated ruler with inaccurate markings, it would
not give us valid results, but they would still be ‘reliable’ because we would still obtain the same results each time we
measured the same thing. Similarly, a poorly designed questionnaire might give us consistent, but misleading results.
Research can also be valid without being reliable. We might produce a wonderfully detailed case study that gives us insight
into some aspect of society, but it might be impossible to repeat the methods used, or show that the study could not have
been interpreted differently.
6
SOCIAL SCIENCE FOR SCHOOLS
Methodologies
What makes good research?
Research should be representative
Usually we want the people or organisations we study
to represent a larger group, so that any results we find
can be applied to them. For example, if we just asked
our friends about bullying we would get results mostly
for children from a particular age group, area of the
country, type of school, circle of friends and so on - they
would not represent all children.
Usually we get representativeness through sampling:
strict rules about how we choose who we study. In
simple random sampling, every member of the whole
group we want to study (e.g. children aged 10 to 15 in
the UK) has an equal chance of being included in our
study.
Research that is valid, reliable and
representative is generalizable
This simply means that we can generalize the results
from our surveys, case studies or experiments on our
subjects to the wider group they represent. For example
rather than talk about the children who took part in
the Understanding Society study, we can talk about all
children in the UK.
Usually there are compromises to be made between
validity, reliability and representativeness in the design
of research, and these compromises get bigger as the
resources we have to do the research get smaller!
Random sampling is a very powerful technique
because it allows us to calculate how accurate our
measurements are likely to be and because it enables
us to make very exact descriptions of very large groups
of people (such as the whole population of the UK) by
surveying only a few hundred. Often exact random
sampling is not possible, and we rely on approximations.
7
SOCIAL SCIENCE FOR SCHOOLS
Methodologies
Joined up thinking
How quantitative and qualitative methods fit together
While it is useful to make a broad distinction between
quantitative and qualitative approaches to research,
in practice most good research uses a mix of both
approaches. In-depth interviewing or participant
observation can give us very detailed accounts of the
attitudes or behaviour of a small number of people,
but we need to turn to survey research to get a good
idea of how widespread they might be in different
social groups. On the other hand, most survey research
depends on careful qualitative investigation of the way
people understand and interpret different questions.
For example demographers are interested in how
people decide about having children. Surveys can reveal
such things as how many children women have or
plan to have, the ages at which women from different
backgrounds or in different social circumstances tend to
have children, and so on. However qualitative research
is used to explore such factors as the reasons for having
different numbers of children or the very different
things planning to have children can mean for different
women. For some it might be a very definite goal, which
is a high priority in their lives. For others it might be
only a vague desire that might or might not be fulfilled
depending on other circumstances.
There is much debate about the strengths and
weaknesses of both kinds of research and about how
best to fit them together. However there is also a
growing interest in combining both kinds of research.
A programme called Timescapes is pioneering
“longitudinal qualitative research”. Its researchers are
working with several major longitudinal studies centres
to bring greater breadth to the quantitative data that
have been collected. For example, researchers used
essay-writing exercises to capture young people’s future
aspirations. Timescapes expanded this activity through
in-depth interviews on why young people held certain
aspirations, how they imagined “getting there” and
what influenced them. The data will act as a baseline
for tracking and comparing these future projections
with the actual course of their lives. In this way, the
qualitative work will guide the questions posed in later
quantitative research.
8
SOCIAL SCIENCE FOR SCHOOLS
Methodologies
Elements of a research project
In some ways, conducting a research project is similar to undertaking a school project or essay. However, a
research study needs to be more methodical and scientific. It also needs:
A research question
Pick a topic or issue you are interested in or a problem you wish to solve (for instance, fashion), and narrow the
focus to something specific (for instance, how does teenagers’ interest in clothing fashion change between age
13 and age 17?) -- and possibly testable hypotheses (e.g. 17 year olds spend more money on clothing than 13
year olds; or ‘17 years olds know the names of more fashion labels than 13 year old’s; or ‘17 year olds are more
likely to be able to describe their style of dress; or when asked open ended questions about their interests and
how they spend their leisure time, older children are more likely to mention fashion).
A review of the existing research on the topic
Researchers conduct a literature review to find out what research has already be done on their topic. This
may help in formulating the research question. The review might reveal that the topic has already been
comprehensively covered. Or it might show that there are gaps in the evidence: perhaps we know little about the
fashion interests of older boys, or children from particular ethnic groups, or in rural areas without easy access to
fashion stores. There might be evidence that interest in fashion changes with age but not why this change occurs.
A methodological design
This is your plan, like an architect’s blueprint for building your project. It will set out the methods you are going
to use, and how the evidence these methods will produce can be used to answer your question.
The collection of raw data
This is called the ‘fieldwork’ because this is where you gather the evidence, for example through interviews,
scrutinising records or documents, observing behaviour – the interviews, surveys, observation and so on.
Scrutiny of its validity - Are your methods and samples appropriate for your research question? Do your findings
make sense?
A review of ethical issues and objectivity
Is anyone in danger of being embarrassed, distressed or harmed? Have you made sure people cannot be
identified? Are the questions neutral and unbiased? (‘How do you rate X’s appearance?’ Not ‘Is X ugly?’ Have you
left your own prejudices out of your analysis?
To generate new knowledge
Analyse your raw data, looking for common themes or striking figures. What have you found out?
(adapted from How to Develop Children as Researchers by Mary Kellett, which has a clear and detailed description of this process)
9