Lectures research methods for Health Sciences
THEME 1: introduction and research question
Lecture 2: Introduction to research questions (research perspectives, questions, priorities, agendas) by Dirk
Essink (wed 2 sep 2020)
1. Understand positivism (objectivist) – part 1
2. Understand interpretivism (constructivist) - part 1
3. Inductive and deductive research – part 2
4. get acquainted with different research methodologies – part 3
a. experimental designs
b. analytical survey designs
c. phenomenological research
d. participatory action research
Literature: Chapter 1 and 2, Gray, Doing Research in the Real World
Research perspectives:
Part 1: What can we know? This is a critical question which concerns each researcher. In this lecture of three
parts I start with touching upon two perspectives that are quite common within health research: positivism and
interpretivism. Research from a positivist perspective aims to find generalizable truths, this is the dominant
perspective within clinical research and epidemiology. Whereas interpretivist approaches to research
acknowledge that any truth is socially constructed and this socially constructed truth is largely based on the
interpretations of the researcher and his/her participants. This perspective is more common in the social
sciences. I describe these as two separate perspectives, however they can be blended as is often done within
health sciences at the VU Amsterdam. I hope that this brief introduction will inspire you to always think about
the boundaries of the knowledge that we can produce as researchers.
What is the purpose of research? The purpose of health science/research is finding an objective, generalizable
truth. When you want to discover a new vaccine, we want it to work in every situation. However, when we
want to know what is the house seeking behaviour it is very difficult to come up with generalizable law and
even if we can it is probably context and time dependent. In relation to this we shall
discuss positivism (objectivism) and interpretivism (constructivism).
Within positivism there are people who think that they can come up with a law about society that we can
generalize. Researchers like Einstein came up with laws how the natural world around us functions. But then
we have the social sciences and that is about how people function together and how we interact; People
constantly interpret the world around them. Based on all their opinions, norms and values they construct an
idea about the world around us. Because we live in groups, a lot of the constructs are similar. For example,
money; the only reason money has value because we see it that way. Much of the things around us is based on
our shared interpretations and ideas, but that is not similar across cultures, nor time.
Objectivism-positivism
- Reality can be observed
- Presenting facts as truth
- Knowledge can be formulated into laws
- Single reality, external, waiting to be found
We have this researcher who looks at society, who has a good protocol and discovers facts and truths about
society and tries to formulate that in to a law. It is not bound to contexts. It is generalizable.
Constructivism-interpretivism
- Truth and meaning are constructed by the person/researcher (who is a subject)
- Interpretations of the world (the object)
- Researchers inherently view the world through their frame of reference
- Multiple realities (are experienced), and meaning is not stable.
1
,Because of different values, experiences and cultural backgrounds, everyone interprets the world differently.
For example, when I see someone does not follow corona rules (shake hands) I interpret that very differently
than someone else does. The world around us is changing and therefore, I am changing and people around us
are changing (opinions), so when we want to research society, are we finding the truth? Or are we because we
view the world through our own frame of references and construct a reality which is plausible to us? And if I
can create a reality, then maybe someone else can create a reality that is slightly different than mine.
Researcher A could construct reality A and researcher B could construct reality B.
> The objects we study, we study them through the eyes of other subjects, persons. When we do an
interview, we invite people who will give their own interpretation within their own frame of reference. So,
we construct groups and construct realities. This is not a bad thing but it tells something about the
certainty.
> We cannot always generalize. Individuals construct their own truth regarding phenomena. Take a waiting
list for example, does it really matter whether it is 5 or 12 days or does it actually matter how people
experience that time? One may experience 10 days completely different than one other.
> Most of the researchers still come from a positivist background where they are looking at causality, law
and prediction, doing a lot of quantitative research, but if you also do qualitative research, you will have to
interpret your findings. You have to construct reality.
Different perspectives, different types of knowledge we need to integrate them:
o Let’s take covid for example. You need people who investigate what covid does in our body: what is an
effective treatment? How many people are infected? What behaviour did infected persons perform?
o On the other hand, you will need people: why did they perform this behaviour? (motives, values, belief,
etc). they all perceive a different kind of reality.
o Another important question could be: what barriers in sociocultural systems (including the health system)
inhibit rapid outbreak control in this community? This person looks at a certain community/context and
tries to understand why something is the
case in the specific context. He will look at
a specific community to get an
understanding why this is the case. This
person probably needs the constructs of
the other people mentioned above. You
can see that you need different kind of
knowledge. We need to share glasses
together.
Positivism-objectivism:
- observe facts of one single reality
- value- free
- measure and predict
- generalizable
Interpretivism- constructivism:
- construct reality
- observations are value-bound
- to deeply understand
- time and context bound
hypothesis/generalization
In sum and so what:
- positive perspective to find truths about a
single reality
- interpretive approach to construct reality
-> in the positivism perspective it is all about
facts. In the interpretivism it is all about meaning.
2
,Part 2: In the second part of the lecture I will discuss inductive and deductive approaches to research. Deductive
approaches are often very structured – researchers collect and analyse data using clearly defined properties
(variables). Thereby aiming at verifying or falsifying theory / hypothesis.
Whereas inductive approaches aim at collecting empirical data far less structured and let meaning arise from
this data. They intend to contribute to the development of new theory, although these theories are often bound
to the context where they emerged.
Inductive and deductive research and reasoning:
Approach 1: All health sciences students are smart (theory). Marie is a health science student. So, she is smart.
Approach 2: I asses the intelligence of all health science students, all appear to be smart. Therefore, health
sciences students are smart (theory).
The first approach is more deductive; we start from a theory, look at our reality and we confirm something. For
the second approach we are more inductive; we have a lot of data and together it could become evidence. We
start with data, see a pattern and end with a theory.
> Approach 1/ deductive/ starts with a theory
> approach 2/ inductive/ ends with a theory
Deductive:
We manipulate and control our data to test this theory and we use very formal instruments.
- begins with hypotheses and theories
- manipulation and control
- uses formal instruments
- experimentation/survey/structured interviews
- seeks for confirmation/ rejection
For example, the theory we start with is the theory of planned behaviour (combines attitude/social norms,
whether you will or will not perform a certain behaviour). The hypothesis using this theory is that exposure to
fast food leads to increased consumption, we do the test and expose one group more than the other so that
we can confirm or reject our hypothesis.
Inductive:
Has a slightly different approach. We look at patterns. Thick descriptions of data. We want to end with a
hypotheses or theory. We look for the emergence of patterns and the researcher is an instrument. Like a
deductive theory, you know specifically what you look for. When you are more open you do not really know
where you look for. You are open for whatever comes in. When you analyse you start to look for patterns and
from that you get a hypothesis.
- thick description (not only describe behaviour but also context)
- ends with hypotheses and grounded theory
- emergence and portrayal
- researcher as instrument
- seeks for (contextual) theory
For example, we do an observation on the view of fast food junkies. In our observation and analysis, we see
that most of them have an absence of life goals so we come up with a new hypothesis when you have no
lifegoals you will eat more fast food and we develop the grounded theory: lifegoal theory. A theory is an idea
how the world is working in a relation to a specific topic.
3
,We can check a new theory with a deductive method; We could do some more deductive research to test this
theory. In reality you use deductive and inductive within one study. You will be open minded: inductive. On the
other hand, you will be more structured because you have ideas and theory: deductive/ In most cases you do a
bit of both; if you have a dataset and you are trying to confirm a hypothesis and you find other interesting
things in your data, you will look at that and report that, so it is partly inductive. When you do interviews you
can be driven by the ideas you already have so you are partly deductive but still very open.
Deductive and inductive:
In deductive research we have empirical data ‘reality’, but when we do deductive research we have a strong
theory (conceptual framework, a structure which the researcher believes can best explain the natural
progression of the phenomenon to be studied) and we look at our data with very clear units of analysis; we
know what we looking for and we try to find it in our data to either confirm or reject our theory.
➔ look at your data with clear units of analyses -> structured
With inductive research we start with the data and look for patterns. We analyse those patterns and a new
theory may emerge. The way we look at this data is less structured, because we are open to new things.
➔ Analyse data for emerging patterns -> theory unstructured.
In sum and so what:
Deductive: confirm/reject theory
Look at your data with clear units of analysis -> structured
Inductive: open and aim to develop theory
Analyse data for emerging patterns -> you try to develop a theory and you are unstructured.
A child has the idea that pie is nice. When she was told there was pie, deductively she would know; this is
great! One day, she was given a pie she did not like. So the theory of pie is nice was rejected. Based on this new
input of data a new theory was developed. Now the new theory was most pies are nice but not those which
have a lot of whipped cream.
Part 3: In this third session I will discuss some of the dominant methodologies applied within public health
sciences: experimental designs, analytical survey designs, phenomenological designs and participatory action
research. In this lecture I use the terminology of Gray and largely the descriptions used in chapter 2. Most of
these approaches are discussed in lectures. Some explicit (experimental designs – participatory action research)
some less explicit (a lot of the measurement lectures cover analytical surveys – inductive analysis focuses on
phenomenology). Note that in this lecture we only focus on methodologies that emphasis primary data
collection, we do not discuss critical approaches that rely on secondary data analysis such as systematic
literature reviews or the analysis of big data.
(quasi) experimental
- Determine causality
- Manipulate the independent (determinant) variable to determine effect on the dependent (outcome)
variable: control over variables
• Randomly assign participants to groups (RCT)
• Use existing groups (Quasi)
o Clear indicators to determine outcome
o Aim to generalize from experiment
o Associated with positivism/deductive approaches
> We are looking for causality. What if we tweak one independent variable, what will the effect be on our
outcome variable? We want to be sure we can control other variables. A classic example is a randomized
controlled trial (RCT). We randomly assign patients to treatment groups. One group will get the treatment,
the other will get a placebo (control group).
> Quasi: Experimental research in which either there is no control group or where the assignment to
experimental and control groups has not been made randomly.
> When you randomly assign people to groups we could call it an experiment but when you use existing
groups you could call it quasi experiments.
> We know what we want to measure; deductive. We want to see whether an intervention works or not and
if we can generalize from this experiment; positivism.
4
,Analytical survey:
- To explore and test proportions/associations/predictors between variables
- Observational studies
- Structured questions/units and limited options for respondents
- Generalization from sample
- Highly deductive, associated with positivism
How many people have unsafe sex? What factors are associated with the act of unsafe sex behaviour? What
factors predict unsafe sex behaviour? It focuses on causality. These are often observational studies. When we
do analytical studies, we have a very clear idea what we want to look for. Questions are structured for
example: highly agree/disagree. Determining sample size and calculation of sample size is critical because we
want to generalize from our sample.
Phenomenological studies:
- Aims for contextual description and analysis of ‘phenomena’
- Phenomenology holds that any attempt to understand social reality has to be grounded in people’s
experiences of that social reality
• Emphasizes inductive logic
• Seeks the opinions and subjective accounts and interpretations of participants
• Relies on qualitative analysis of data
o Is not so much concerned with generalizations to larger populations
> Phenomenological studies are totally different. They do not aim for generalization and are not deductive.
When you assume that every person has its own personal background it has a different perspective of
reality, it becomes very difficult to say that I as a researcher with a very structured questionnaire to
unravel how they give meaning to their own reality. So phenomenological studies are relatively open and
uses qualitative data so that meaning can arise: inductive.
> We want to describe the phenomena within its context. For example, a PhD student want to understand
how woman perceive empowerment. This will be totally different how women in Africa perceive it that
how women in the Netherlands perceive is.
(participatory) action research:
- Research that aims to change practice in real life
• Collaborations between researchers and practitioners and users (e.g. patients, community members)
• Iterative designs
• Mixed methods
• Understanding of perspectives in order to determine change and (often) measuring the change
• Deductive and inductive reasoning.
> It is participatory when it really emphasizes the inclusion of practitioners, patients, etc in all phases of the
study. It is not a participatory research when the researcher makes up the research question and
methodology. Participants contribute to what is relevant to study and how to steady it, interpretation of
the results and conclusions. It is focused on a change in for example a community.
> Iterative design: Data collection and data analysis alternate. You always make decisions about your design.
Action research can be seen as a cycle because change is not shaped in one day. We call the cycle
processes an iterative design.
Research agenda’s and priority setting:
Research questions (RQ) and Research Objectives (RO) emerge from a knowledge gap that is either more
practice or more theory focused. But also, policy issues play a (major) role. Understanding the knowledge gap
and determining the RO/RQ requires insight from scientific literature and contextual insights. Therefore, we
highlight the importance of research agenda and stakeholder engagement in defining a research agenda and
some relevant examples will be presented
5
,Priority setting & research agenda’s in the world of evidence-based
medicine door Raymond Ostelo Part 1:
Evidence based medicine triad: at the center of the triangle we try to
improve patient outcomes. We do that by, finding the best clinical
evidence, we look at patient’s values and expectations, and the clinical
expertise: are health care providers able to provide such a treatment?
There is more than just the evidence in implementing research. Our course
focuses on the bottom.
What is research priority setting?
• Organizations conducting or funding public health
research have to select research priorities while often
facing competing demands and scare resources.
• Therefore, a collective activity (policy makers, health
care systems, etc) for deciding which uncertainties are
most worth trying to resolve through research is
warranted.
• Uncertainties considered may be problems to be
understood or solutions to be developed or tested;
across broad or narrow areas.
Wheel of research priority setting:
We start at the top: defining your scope and identifying
stakeholders. You should have a clear understanding in what
field we are going to look for uncertainties and who should
be involved. At the right: you have to think about methods to
rank; which uncertainty has more priority? Bottom: identify questions and topics. Left: you can submit your
research agenda. What is the most important research we should do = priority setting.
Examples priority setting:
> “setting research priorities in tobacco control: a stakeholder engagement project” Methods: two surveys
and a workshop. A range of stakeholders participated, including: members of the public (smokers and ex-
smokers), Clinicians, Researchers, research funders, health-care commissioners, public health
organizations. The stakeholders identified 183 unanswered research questions (uncertainties). These were
categorized into 15 research categories. Per category they came up with the top 3 questions based on
voting (see slide 8).
> “research priority setting in organ transplantation: a systematic review”
They did a systematic review, in which they summarized the priority setting. Methods: They did not involve
in asking stakeholders. The studies they identified in terms of research priority in the field of kidney
transplantation. They identified 9 studies of which 7 are focused on organ donation as a top research
priority, based on which topic is most prominent in literature. In this way, they summarized the top
research priority (see slide 10).
Carol grand-pearce (1998):
• Important question: do all stakeholder understand this process and has everybody a voice in such a
process?
• Mismatches in priorities for health research between professionals and consumers: a report to the
Standing Advisory Group on Consumer Involvement in NHS R&D (research and development) Program (1
Jan. 1998): They were interested if there is a difference in priority setting if you only focus on
professionals/researchers in contract to the priority’s patients/consumers have.
• These results confirm that consumers have little understanding of the priority-setting process, little
knowledge of health R&D priorities, & relatively poorly formulated priorities of their own: they think it is
important that their problem is studied but they find it difficult to formulate what they would like to study.
• These views may bolster the opinion that consumers have little to contribute to R&D priority-setting.
• Alternatively, they may be seen as evidence that closer involvement of consumers with priority-setting
processes could lead to useful inputs of views and experience
6
, Part 2: In the first part we talked about priority setting in general. In this part we focus more the details on
priority setting, research agenda and funding of research.
How to develop research priorities?
Cochrane methods priority setting: They aim to prepare documents for guidance on how to develop priority
setting in the first place for Cochrane reviews: what reviews should be done? But it can be used by external
organizations as well. The first two are examples on funding institutes. We will highlight how the priority
setting is related to the funding of research.
• The National Health Care Institute, Zorginstituut Nederland
• ZonMW, Doelmatigheidstudies (Health Technology Assessment)
National Health Care Institute, Zorginstituut Nederland:
• Priority setting departs from the question which types of health care are reimbursed (basic health
insurance package) or not.
• Evidence gap: Systematic reviews
• Other considerations: rare diseases etc
Main task of the national health care institute is to advise the minister on what interventions should or should
not be reimbursed in the patient health insurance package. In terms of priority, they focus on those kinds of
questions. They look at the literature first. They fund this type of research in order to get an answer whether or
not an intervention is effective and only then, the intervention can be included in the patient insurance
package. It is not just the evidence from systematic reviews. In rare diseases other considerations are also
important. Systematic reviews are prioritized then they invite researchers to submit a proposal (see slide 5).
The national health care institute. Final decision based on:
• Evidence
• Other considerations
o Is it ‘necessary care’ (e.g. cosmetic surgery?)
o Own responsibility (e.g. walking aids?): when people start to get problems in their walking they
get a working stick themselves, but when the problems increase, they often need a rollator. Why
should we reimburse that if the walking stick is also paid by the patient themselves?
The above was an example of priority setting in a structured way. Sometimes the process is less structured.
Priority setting by megaphone: There was an intervention and only a few surgeons performed a surgery. They
made a lot of fuss that this technique was not included in the package and was not reimbursed but it was a
successful intervention and it should be reimbursed. They wrote letters to the national health care institute. It
was decided that the effectiveness was not supported by good research and could not be reimbursed in the
package, therefore, there should be a study in this technique-> research priority setting.You see that some
priority settings are well structured, others, for example public opinion sets the priorities.
Another example by the national health care institute in close collaboration with ZonMW, called veel
belovende zorg: promising in the sense that it could be admitted to the patient’s insurance package. The
priority is set by the question, is it indeed a new and promising treatment or is it a well-established treatment
which is done a lot in practice but doesn’t have strong evidence yet? If you submit a research proposal that
focuses on these kinds of questions, then it gets a high priority.
Research agendas: that It is an outcome of the whole process of the previous phases.
• Mostly made by professional organizations
• Patient involvement is a key element
Example research agendas (see slide 25-31):
• Aim: What research should be done?
• Involve the Dutch Society, an open process in society
• (research) institutes
• Public
7
