How to start a research process?
Before starting a research process, there are some questions I need to answer before.
Firstly, I have to have some kind of motivation (interesting observation, gap in knowledge base, helping a business etc.) for the research.
Then I need to define the research question (this is the most important part).
- if the question is more like “what/who/where”, the research would be more exploratory
- if the question is like “how/why”, the research would be more explanatory
The action plan for the research includes:
- what reasoning will be used? → Three ways of thinking
- what is the goal of the research? → Reasoning modes in the research (what balance of exploration, rationalization, and validation?)
- what I want to focus on? position the research on those scales → Research decisions
After answering the questions in the action plan, I can choose the right Research methodology for my research.
I also need to be aware of my “research philosophy”, which influences all my decisions: MIS - Philosophy and IS research
Three ways of thinking
Deduction
- direction: general → specific
- I start with some existing theory, deduct a hypothesis from it and test it out on new data
- key point is Falsifiability, trying to (dis)prove the theory
- another definition: Given a rule and a cause → deduce the effect
Induction
- direction: specific → general
- this is the “discovery” way: I observe cases/phenomenons and the I generalize from them (look at data, find patterns and induce/derive some general rules from it)
- warning: beware of “weak induction”
- another definition: Given a cause and an effect → induce a rule
Abduction
- direction: observation → simplest (Occam’s razor) or most likely explanation
- essentially an educated guess, trial-and-error trying the explain the observations
- another definition: Given a rule and an effect → abduce a cause
Reasoning modes in the research
- Exploration - discovering new phenomena or patterns (abduction, induction)
- Rationalization - building theoretical explanations (deduction, abduction)
- Validation - testing if the explanations/hypotheses hold up against new data (induction, deduction)
- a good research should combine all three modes (each one requires a different mode of thinking)
Research decisions
- this is highly dependent on the defined research question, it’s choice influences the decisions
| Spectrum | One end of continuum | vs. | Other end of continuum |
|---|---|---|---|
| Aim | Exploratory | vs. | Explanatory |
| Method | Qualitative | vs. | Quantitative |
| Boundary | Case | vs. | Statistical |
| Setting | Field | vs. | Laboratory |
| Timing | Cross-sectional | vs. | Longitudinal |
| Outcome | Descriptive | vs. | Causal |
| Ambition | Analysing | vs. | Designing |
- on each spectrum, I need to decide, where I want to be with my research (the research could have more parts with different spectrum positioning)
Research methodology
- the specific strategy, how do I do the research
- each methodology has different strengths, so it’s important to take them into account when choosing the main methodology
- Controllability
- how much control has the researcher over events during the study (Design science)
- Deductibility
- how much the strategy allows for deductive reasoning (Computational and Quantitative)
- Repeatability
- how much the research could be repeated with the same/similar results (Design science and Computational)
- Generalisability
- how much the research results could be generalized beyond the study (Quantitative)
- Explorability
- the extent to which previously unknown findings might be discovered (Qualitative)
- Complexity
- how much the research could lead to exhaustive knowledge contributions (Qualitative)
- Controllability
- methodologies:
- Quantitative strategies
- Qualitative strategies
- Mixed methods - a mix between quantitative and qualitative strategies
- Design science methods - designing and building artifacts (see The Science of Design)
- Computational methods - automated collection and analysis of digital data
Systematic literature review (SLR)
There are actually 2 ways how to get information from existing literature.
- ad-hoc literature review = just going through papers and books for the first research proposal, to see, what is available, where are the gaps etc.
- that is good, but it is not systematic and often it does not lead you anywhere in the actual research
- SLR = a whole theory (research strategy) about doing the literature review “the right way”, so I can formulate my research question properly and build the theoretical foundations needed to position my research correctly
- it should be unbiased and rigorous
- it has to be properly documented, so it is clear, what was done and how (which books were included, which not and why)
- it has 3 main phases
- planning the SLR
- conducting the SLR
- reporting the SLR
Systematic literature review
- it is a specialized research method that offers a predefined search strategy for answering a specific research question
- a literature review is a prerequisite in almost every research project
- why do I want to do a literature review?
- to come up with a good research question (identify the gaps in the current research)
- summarize the existing evidence
- theory building (providing a background to position new research activities)
- ⇒ to get context and put my research into it
Ad hoc literature review vs. systematic literature review (SLR)
- ad hoc = does not answer a specific research question
- usable when drafting my research proposal, first exploration of the research area, formulating problem, formulating the research question
- SLR
- a definition from (Kitchenham & Charters, 2007): a means of identifying, evaluating and interpreting all available research relevant to a particular research question, or topic area, or phenomenon of interest […] in an unbiased and rigorous way
- it’s a predefined research strategy, it’s a full overview of the research area, synthesis of all relevant information regarding a topic
- it answers a specific research question
Three phases of the SLR
- Planning the review
- I can build a literature review protocol, which consists of:
- research questions that should be answered by the literature review (so we know what we are looking for)
- search strategy (where to search, which terms/strings to use)
- study selection criteria (all studies should be relevant to our research by given metrics)
- what studies to include and what studies to exclude
- I can build a literature review protocol, which consists of:
- Conducting the review
- identify potential sources
- automated search: WoS (a digital library), Google Scholar etc. using a special notation (*, ?, OR, AND etc.)
- or using manual search
- backwards search (browse references cited in the article of interest) - find older articles
- forward search (browse articles that cite my article of interest) - find newer articles
- apply inclusion/exclusion criteria (to keep only the relevant ones)
- obtain the full versions of the articles (to extract data)
- extract data and do a synthesis of relevant information
- identify potential sources
- Reporting the review
- report the protocol (from phase 1) and the deviations from it (with rationales)
- report information about search results (all potential sources) and results of inclusion/exclusion
- report table of included studies with relevant information