Professional Documents
Culture Documents
Lecture 4 Research Methodology
Lecture 4 Research Methodology
of the
Research Methodology
The Research Process
The Research Process
Research Idea
Literature Review
Formulate The Research Problem/Question
Research Design
Collecting Data
Data Analysis
Answering The Empirical Research Questions
Results Interpretation
Comparing with Earlier Research
Conclusion
Evaluate/Reflect
Overview
Steps of the Scientific Method
Formulate the Research Problem
Collecting Data
Developing The Research
Hypothesis
Conclusion
Evaluate/Reflect
Steps of the Scientific Method - Recap
Does X cause Y?
Does X prevent Y?
Does X cause more Y than does Z?
Is X better at preventing Y than is Z?
Does X cause more Y than does Z under one condition
but not others?
Design
What is an effective way to achieve X?
How can we improve X?
When Looking for Research Problems:
Consider if the research will
Educate you?
Lead to even better research projects?
Be an enjoyable way to spend your time? A PhD takes 3 or more
years…
Serve a goal that will still seem worthy 6, 12, or 48 months from
now?
Be likely to "succeed" in some sense? (guideline: will it make an
interesting conference/journal article?)
Get the academic research community interested in you and
your work?
Prove to an industrial employer that you have what they want?
Make you a so-called "famous grad student"?
Overview
Steps of the Scientific Method
Formulate The Research Question
Collecting Data
Developing The Research Hypothesis
Conclusion
Overview
Collect Data
Preparation: Make Hypothesis Testable
(Operationalization)
Preparation: Design the Study
Conduct the Experiment or Observation
Organize the Data
Organize the data and analyze it to see if it supports or
rejects your hypothesis.
The exact type of test used depends upon many
things, including
the field
the type of data and
sample size among other things.
The vast majority of scientific research is ultimately
tested by statistical methods, all giving a degree of
confidence in the results.
Operationalize
Operationalization is used to give some indication of the
exact definitions of the variables, and the type of
measurements used.
This will lead to the proposal of a viable hypothesis.
The hypothesis should be both testable and falsfiable.
Then design a study and construct a test/experiment to
collect data.
Be aware of validity when choosing variables, especially
when studying people.
You might not be measuring what you think you are
measuring.
Validity
1. Construct validity
Do the operational measures reflect what the researcher had in mind ?
2. Internal validity
Are there any other factors that may affect the results ?
Mainly when investigating causality !
3. External validity
To what extent can the findings be generalized ?
Precise research question & units of analysis required
4. Reliability
To what extent is the data and the analysis dependent on the researcher
(the instruments, …)
Other categories have been proposed as well
credibility, transferability, dependability, confirmability
Qualitative vs. Quantitative
Qualitative tends to have more open questions and
hypotheses.
Quantitative research have an experimental
approach focusing more on counting and classifying
observations.
Data Collection for Qualitative Studies
Interviews
Field Notes
Observations
Documents
Log Files/Content Analysis
Biographies
Oral histories and story telling
Newsletters, rules, principles and pictures
Data Collection for Quantitative Research
Quantitative, Experimental
objective
data collection, logging
numerical measurements
performance metrics
questionnaires on large groups
comparison with a control
group or “baseline” system
Types of Evaluation Data
Qualitative, Rich
Subjective
in-depth
Qualitative
Interviews
Descriptive
Naturalistic
Types of quantitative data
7. Simulation
Feasibility Study
Is it possible to solve a specific kind of problem … effectively ?
+ computer science perspective (P = NP, Turing test, …)
+ engineering perspective (build efficiently; fast — small)
+ economic perspective (cost effective; profitable)
• Is the technique new / novel / innovative ?
+ compare against alternatives
➡ See literature survey; comparative study
• Proof by construction
+ build a prototype
+ often by applying on a “CASE”
• Conclusions
+ primarily qualitative; "lessons learned"
+ quantitative
- economic perspective: cost - benefit
- engineering perspective: speed - memory footprint
Pilot Case (a.k.a. Demonstrator)
As an example, begin with the null hypothesis: Sleep does not affect reflexes.
If we can disprove this, we find that sleep does have an effect, then go to the next null
hypothesis: Different amounts of sleep have the same effect on reflexes.
If we can disprove this, go to: Maximum reflex efficiency is not achieved after eight hours
of sleep.
And so on. At each stage in the investigation, we conduct experiments designed to try to
disprove the null hypothesis
Hypothesis: Significance Test
Significance tests determine the probability that the null
hypothesis is true.
E.g. we use a significance test and find that the probability
that the null hypothesis is true is less than 5 in 100, i.e.
stated as p <.05.
If the chances that something is true are less than 5 in
100, It’s a good bet that it’s not true.
If it’s probably not true, we reject the null hypothesis.
Hypothesis: Significance Test
There is no rule of nature that dictates at what probability
level the null hypothesis should be rejected.
However, conventional wisdom suggests that .05 or less
(such as .01 or .001) is reasonable.
Researchers should state in their reports the probability level
they used to determine whether to reject the null hypothesis.
Note that when we fail to reject the null hypothesis
because the probability is > .05, we “fail to reject” the null
hypothesis and it stays on our list of possible explanations
i.e. we never “accept” the null hypothesis as the only
explanation.
Conclusion
Measure performance of your approach
allow replication of results by other researchers –
repeatability
Allow objective comparisons
e.g. your new method vs. state-of-the-art method, vanilla
method,“straw man”, etc.
lesion experiments – measure contribution of system
components
Statistical testing allows us to express the level of
confidence in a result
Provide benchmarks for future research
Conclusion