DIMENSIONS OF RESEARCH

Dissecting the Dimensions of Research

• topic: physical–biological–psychological–sociological
• novelty: create new vs review published data or info
• technology: develop new vs use existing methods
• scope: study a single case vs a sample
• mode: observe vs intervene
• methodology: qualitative vs quantitative (info vs
numbers)
• ideology: objective vs subjective (positivist vs
interpretivist)
• politics: neutral vs partisan
• utility: pure vs applied
• reassembling the dimensions
 Topic: what are you researching?
biophysical psychosocial
clinical behavioral psychological economic social
• Examples
– Clinical: the effect of a herb on performance.
– Psychological: factors affecting work-place satisfaction.
– Behavioral: how can we reduce truancy at this school?
– Economic: characterize the productivity of new immigrants.
– Social: develop risk-management procedures at a gym.
• Finding a good question/problem to address can be hard.
– It helps to have a good supervisor, good colleagues, and/or
knowledge or practical experience of and affinity for a topic.
– You must read journal articles to find out what's already
known.
• Authors also often point out topics for future research.
 Novelty: creating new or reviewing published info?
create review

• Most research projects are so-called original

investigations.
– You obtain new data or information about a
phenomenon.
– You reach a conclusion and try to publish it.
• Some research projects are reviews of the
literature.
– You use other researchers' published data or info about
a phenomenon.
• A quantitative statistical review is called a meta-analysis.
Technology: develop new or use existing method(s)?
develop new use existing
• Sometimes a legitimate topic for study is methodological.
• For example, development or novel investigation of…
– a measuring device
– a psychometric instrument (questionnaire or inventory)
– a protocol for a physical performance test
– a diagnostic test
– a method of analysis.
• You usually include or focus on a reliability and/or validity
study of the measure provided by the method.
– Validity = the relationship between observed and true values.
– Reliability = reproducibility of observed values.
 Scope: case or sample?
case sample
• Are you solving a single case of something, or is it a sample
that will allow you to generalize to a population?
• In a case study…
– You are interested in "what happened or will happen here".
– Your finding applies only locally: to the case you studied.
– The quest for an answer can be like that in a court case.
– Qualitative methods are often required.
– You reach an answer by applying logic (= common sense?) and
skepticism to your knowledge and to the information you
gather.
• Be wary of conventional wisdom and your own prejudices.
– It may be possible to estimate probabilities of benefit or truth
of various answers.
• In a study of a sample…
– You are interested in "what happens in general".
– Rarely, "what" is simply descriptive: the frequency, mean
value or other simple statistic of something in the sample.
– Most often, the "what" is the value of an effect statistic:
the relationship between the thing of interest (a
dependent variable, such as health, performance…) and
something else (a predictor variable, such as training,
gender, diet…) in the sample.
• Examples of effect statistics: difference or change in a mean
value; ratio of frequencies (relative risk); correlation coefficient.
– You control for other possible predictor variables either by
holding them constant or measuring and including them in
the analysis.
• Example: the effect of physical activity on health, controlling for
the effect of age on health.
• In controlled trials (interventions), a control group accounts for
any effect of time that would have happened anyway.
• More about studying a sample…
– You study a sample, because it is impractical and wasteful
(and therefore unethical) to study a population.
– “What happens in general" refers to the average person or
situation in a population represented by your sample.
– "Population" is a defined group, not the entire human race
or all possible situations.
– You make inferences about that population; that is, you
generalize from the sample to a population.
• You can make inferences to other populations only if you can
argue that those populations are similar to your sample with
respect to the effect you have studied.
– There are several ways to generalize from sample to
population…
• Old: develop a null hypothesis about a relationship, then
test the hypothesis (that is, try to falsify it) using statistical
significance based on something called the P value.
• New: identify a relationship, measure its magnitude, state
the uncertainty in the true value using confidence limits,
then make a conclusion about its clinical or practical
importance in the population.
– Sample size is a big issue.
• The smaller the sample, the more the uncertainty.
• A stronger relationship needs less certainty.
• So a stronger relationship needs a smaller sample.
• Unfortunately most relationships are weak or trivial,
so you usually need large samples.
 Mode of Enquiry: observational or
interventionist?
observational interventionist
or non-experimental or experimental
or descriptive

• In an observational study…
– The aim is to gather data or information about the world
as it is.
– So you hope the act of studying doesn't substantially
modify the thing you are interested in.
• In an interventionist study…
– You do something to the world and see what happens.
– You gather data or information almost always before and
after the intervention, then look for changes.
• The following comments refer to observational and
interventionist studies with samples.
• The estimate of the magnitude of a relationship is
less likely to be biased (that is, not the same as in a
population) if…
– the sample is selected randomly from the population,
and…
– you have a high compliance (low proportion of dropouts).
• An observational study of a sample…
– usually establishes only an association between variables
rather than a causal relationship;
– needs hundreds or even thousands of subjects for
accurate estimation of trivial or small effects.
• Types of observational study with a sample, weak to strong:

– Case series, e.g. 20 gold medallists.

– Cross-sectional (correlational), e.g. a sample of 1000 athletes.
– Case-control (retrospective), e.g. 200 Olympians and 800 non-
Olympians.
– Cohort (prospective or longitudinal), e.g. measure
characteristics of 1000 athletes then determine incidence of
Olympic medals after 10 years.
• In an intervention with a sample…
– You can establish causality: X really does affect Y.
– You may need only scores of subjects for accurate
generalization about trivial or small effects.
– The outcome is the effect of a treatment on the average subject.
– Researchers usually neglect the important question of individual
responses to the treatment.
• Types of intervention with a sample, weak to strong:
– No control group (time series), e.g. measure performance
in 10 athletes before and after a training intervention.
– Crossover, e.g. give 5 athletes a drug and another 5
athletes a placebo, measure performance; wait a while to
wash out the treatments, then cross over the treatments
and measure again.
• Ethically good, because all subjects get all treatments.
• But can't use if the effect of the treatment takes too long to wash
out.
• Each subject can receive more than two treatments.
– Controlled trial, e.g. measure performance of 20 athletes
before and after a drug and another 20 before and after a
placebo.
• You need up to 4x as many subjects as in a crossover.
• In interventions, bias is less likely if…
– Subjects are randomly assigned to treatments.
– Assignment is balanced in respect of any characteristics
that might affect the outcome.
• In other words, you want treatment groups to be similar.
– Subjects and researchers are blind to the identity of the
active and control (placebo) treatments.
• Single blind = subjects don't know which is which.
• Double blind = the researchers administering the treatments and
doing the measurements and analysis don't know either.
 Methods: quantitative or qualitative?
quantitative qualitative

• With quantitative methods…

– You gather data with an instrument, such as a stopwatch,
a blood test, a video analysis package, or a structured
questionnaire.
– You derive measures or variables from the data, then
investigate relationships among the variables.
• Some people think you have to do it by testing hypotheses.
– Error of measurement is an important issue.
• Almost all measures have noise or other errors.
• Errors affect the relationship between measures.
• You attend to errors via validity and reliability.
• A pilot study to investigate error can be valuable.
• With qualitative methods…
– You gather information or themes from texts, conversations
or loosely structured interviews, then tell a coherent story.
• Software such as NVivo can help.
– The open-ended nature of these methods allows for more
flexibility and serendipity in identifying factors and practical
strategies than the formal structured quantitative approach.
• The direction of the research may change mid-stream.
– Formal procedures enhance trustworthiness of the
information.
• Triangulation–aim for congruence of info from various sources.
• Member checking or respondent validation–the subjects check the
researcher’s analysis.
• Peer debriefing–colleagues or experts check the analysis.
• Hybrid or mixed method: analyze a sample of cases
qualitatively, then code information into values of
variables to make inferences about a population
quantitatively.
 Ideology: objective or subjective?
objective subjective
positivist interpretivist

• Others refer to this dimension as paradigmatic or

philosophical.
– A paradigm sometimes has religious status for its adherents:
thou shalt not question it!
• Positivist or objective
– We make and share observations, identify problems and solve
them without disagreement about the nature of meaning or
reality.
– This so-called dominant paradigm is responsible for our current
understanding of life, the Universe, and almost everything.
• Interpretivist
– Part of the truth of a situation can be found in the
researcher's interpretation of the self-understandings
of participants.
– Truth is discovered partly by thought as well as by
observation.
– Grounded theory of social science is interpretivist:
truth emerges from your observations; you do not test
a hypothesis.
 Politics: neutral or partisan?
neutral partisan
• Most researchers aim to be politically neutral or
impartial by presenting all sides of an argument.
• Sometimes the researcher is overtly partisan or
adversarial.
– In social science such research is known as critical or
radical.
• The researcher attempts to raise understanding about oppression
and to facilitate collective action against it.
• Some commentators regard critical research as a specific paradigm
in social science, but…
– In my experience even biomedical researchers sometimes
adopt an overtly partisan or adversarial stance on an issue.
• Or there are often hidden agendas and biased reporting.
• Maybe that’s OK, because their stance stimulates debate.
 Utility: pure or applied?
pure applied
• In pure, basic, theoretical or academic projects, the aim is
to understand the cause or mechanism of a phenomenon.
• Applied or practical projects impact directly on health,
wealth, or culture (art, recreation…), or on development of
a method.
• Even so, try to include mechanisms in an applied project.
– It will help you publish in a high-impact journal, because their
editors and reviewers can be snooty about pure research.
– Understanding something may give you ideas for more
projects.
– A mechanism variable in an unblinded intervention can help
exclude the possibility of a placebo effect.
• Pure is sometimes lab-based, lacking naturalness.
• Applied is sometimes field-based, lacking control.
 Reassembling the Dimensions
• A given research project is a point in multidimensional space.
• Some regions of this space are popular:
biophysical topic psychosocial
sample scope case
quantitative method qualitative
interventionist mode observational
objective ideology subjective
neutral politics partisan

These often go together as These often go together as

quantitative research. qualitative research.
 This pigeonholing doesn’t apply to the novelty, technology and
utility dimensions.
 Missing dimension………
conceptual empirical

Conceptual: relates to some abstract idea(s)

or theory, to develop new concepts or to
reinterpret existing ones.

Empirical: relies on experience or

observation alone without regard to system
or theory. Data-based research, coming up
with a conclusion verified by
observation/experiment
• Some regions are less popular, but worth visiting. For example:
– Action research is a subjective intervention with a case or sample.
• Dealing with the problems of everyday life is an informal kind of action
research.
– Some researchers identify the extreme subjects in a quantitative
survey, then interview them subjectively/qualitatively as cases.
– Others do a qualitative pilot study of a few cases to identify a
problem and the appropriate measures for a larger quantitative study
of a sample.
• A project based in an unusual region may give new insights…
– But you may struggle to publish in journals devoted to more popular
regions.
• Researchers who mix qualitative methods (such as intensive
interviews) with studying a sample (for generalizing to a
population) can run into a sample-size problem, as follows...
– Qualitative methods applied to a sample often result in a small
sample size because…
• subjects are hard to get, or…
• the interviews are too time consuming, or…
• the researchers dislike the idea of large samples.
– But a study with a small sample can adequately characterize
only strong associations (large effects) in a population.
– So these small-scale qualitative studies are not definitive for a
small or trivial effect.
– Furthermore, open-ended inquiry is equivalent to assaying
many variables, so there is a high risk of finding a spurious
association.
– If the sample is small, the spurious association will be strong.
– Therefore small-scale qualitative studies are not definitive even
for a moderate or large effect.
– Bottom line: when using qualitative methods to generalize to a
population, you need a large sample to characterize small
effects.
This presentation is updated from a paper at:

Hopkins WG (2002). Dimensions of research. Sportscience 6,

sportsci.org/2002