You are on page 1of 23

Formulating a Research

Question / Hypothesis

Dilip R Karnad
MD, FACP, FRCP
Doing Research –
the first steps towards the Nobel Prize

• Getting the idea

• Developing the research hypothesis

• Defining the objective of the research

• Framing a tentative research question

• Appropriate study design

• Framing the final research question


Getting the research idea - Curiosity

Single incident Repeated


observation
Unusual
Unexpected
Paradoxical

Chance? Pattern

Why did it happen? Is it really exceptional?

Case Report Hypothesis


Process of Research
Getting the idea Convert to research study
Problem
Interesting Observation
Hypothesis

Pattern
Experiment

Tentative hypothesis Observation

Confirmation
Is the idea worth pursuing?

• Identify What is the problem?

• Prioritize Why is this important?

• Rationale Can the problem be solved?

• Importance What are the benefits to society?


Lets plan a research study

Hypothesis:
Computer games make children smarter

Important? Can be solved? Benefits to society?


Is an answer already available?

Literature search

• Is the question fully answered?

• Have previous attempts failed? Why?

• If answered, is this being used in clinical


practice? Why not?

• Limitations of previous studies


Converting Hypothesis into Research study

Hypothesis: Computer games make children smarter


• Start with a broad, general question / hypothesis

• Narrow down or focus on a specific problem

• Design a study to address specific hypothesis

• Computer games – What type? Which one? How much


time?

• Children – Age? Gender? Class in school? Language

• Smarter – IQ? Maths? Interview? Problem-solving?


General knowledge? Marks in exam?
Structured, holistic approach for research
planning - SHARP

1 Set up a causal model


2 Establish a fact-hypothesis matrix
3 Develop a variable – indicator – measurement
matrix
4 Select the study design
5 Define sampling method, sample size
6 Define statistical methods
7 Consider ethical issues
8 Set up operational plan
Charlton KE et al. S A J Clin Nutr 2000;13:1
Fact – hypothesis matrix
Hypothesis: Computer games make children smarter
Computer games may improve scholastic performance

Hypothesis Fact (published literature)


Playing with other children better Play – helps in cognitive, social and
adjusted in society emotional development
Negative emotions channelized in Play – allows expression of negative
harmless activities emotions - hate, gaining control of
situation, conflicts
Games - More cerebral stimuli - Computer games – vast range of
Imagination images, sounds, animation,
information
Recreation Break from routine studies
Games help in Concentration - Computer games – teach children to
Important for preparing for board sit in one place, prolonged
examinations concentration
Therefore computer games improve
student‟s scholastic performance?
? ? ? ?
Research question
Characteristics of a good hypothesis

• Must be based on facts (published


studies)

• Should not have already been proven.

• Should be convincing

• Often fills gaps in current knowledge

• Usually explains more than one


situation
Are there any other possibilities that
explain the problem?

Should we have several alternative


explanations for all possible results
even before we start the study?
Hypothesis generation
Alternative hypothesis

• 75% of pts with hypertension (HT) drink coffee


• 20% of healthy controls drink coffee (p < 0.001)
Our conclusion:
Cause–effect relationship Coffee drinking causes HT

Other possible explanations:


1 Chance Coffee drinking not related
2 Bias Coffee drinking not related
3 Effect (not cause) HT causes coffee drinking
4 Effect – effect HT and coffee drinking
caused by a third factor
5 Partial cause – effect Coffee drinking is one of
the causes of HT
Mill‟s canons of inductive reasoning

John Stuart Mill (1806-1873) - British Philosopher

 Method of difference – disease present/absent


risk factor present/absent

 Method of agreement – presence of risk factor and


disease

 Method of concomitant – dose response effect


variation

 Method of analogy – 2 diseases occurring together


may have common cause

J S Mill. System of Logic. 1843


Causal Inference – Koch‟s postulates
Robert Koch (1843-1910) – German microbiologist

1 Microbe found in diseased tissue or animals and


not in normal tissue or animal

2 Can be cultured from diseased tissue

3 Organisms from culture can cause disease if


introduced in animals or man

4 Organism can be re-isolated from experimentally


infected animal or man

5 Treatment or preventive measures can eliminate


disease
Robert Koch, 1884
Criteria for causation (weakest to strongest)
1 A is associated (correlated) with B (Strength of
association – correlation coefficient, odds ratio)

2 A precedes B (Temporality)

3 Plausible biological mechanisms lead from A to B


(Plausibility)

4 A graded dose-response relationship exists between A


and B (Dose-response)

5 Presence of A always leads to B (Longitudinal studies)

6 Removal of A always prevents B (Treatment effect)


Variable – indicator - measurement matrix
Hypothesis: Computer games may improve scholastic performance

Variable Indicator Measurement


Type of game WCCG category - C Name of game
Time Hours/week
Dose Attempts Number of tries/week
Duration Number of months
Highest score every
Expertise Performance in game
week
Concentration on Duration of study Hours/day
studies Efficiency of study Chapters/day

Comprehension, Recall
Quality of study Scores in monthly tests
- Periodic tests

Mark sheet of XIIth


Ultimate benefit Exam marks / scores
board exam
Review the objectives
• Is it possible to answer the initial question?

• Do we have the subjects, technology,


budget, time to complete the study?

• Can the question be broken down into smaller


parts, which can be answered separately?

• Do we need to rephrase the initial question?

• Develop a specific, focused research question


Components of an ideal research question

PICO
• Population: Normal/disease, human/animal
• Intervention: Therapy, risk factor,
diagnostic test, preventive measure
• Comparison or Control: Better / worse,
placebo / active control
• Outcome: Clinical, functional, laboratory,
economic
Schwatrz MD, et al. ACP Journal Club 2007;147:A8,9
Ideally framed research questions
PICO (Population – Intervention – Control – Outcome)

• Is low-dose dopamine infusion superior to noradrenaline in


decreasing need for dialysis and in-hospital mortality in
adults with septic shock?

• Does addition of aspirin to standard treatment decrease


need for hospitalization in diabetic females with coronary
artery disease?

• Is the „quantiferon‟ test superior to the tuberculin test in


detecting presence of active tuberculosis in asymptomatic
healthcare workers?

• Does salt fortified with folic acid and Vit B12 decrease
serum homocysteine levels more than unfortified salt in
vegetarians with previous myocardial infarction?
Ideally framed research question
Hypothesis:
Computer games improve scholastic performance
Population:
Intervention:
Control:
Outcome:

Research Question:
Does playing with “Brain Tickler 2010” daily for 2 years
improve performance of female Xth Grade students in the
XIIth Grade examinations when compared to students
with unrestricted access to all computer games?
Summary
1 Conceptualize the problem
Identify - what is the problem?
Prioritize - why is this important?
Rationale - can the problem be solved?
Importance - are there benefits to society?
2 Define a testable hypothesis
3 Frame a broad research question
3 Break the problem / hypothesis / study into parts
4 Refine question - specific, focused
Can be answered with available resources - time, budget,
facilities
It is worthwhile to put in a lot of thought and effort
before starting even a simple study!

You might also like