MIS - course summary

This is a comprehensive summary of the Management Information Systems (MIS) research methods course, covering scientific foundations, research process, theory, qualitative and quantitative methods, mixed methods, design science, and research philosophy.

---

What is science? (Ch. 2)

Science is about obtaining facts through proper methods in a systematic and organized way. Scientific research must be falsifiable, questionable, and conducted through defined and repeatable experiments. By doing scientific research, we contribute to a “body of knowledge” by:

• improving explanations of phenomena (theories)
• collecting scientific evidence (e.g. case studies in unique environments)
• improving the methods of science themselves (better measurements, better processes)

Key properties of the scientific method

• Replicable - research can be repeated
• Independent - free from subjective judgment (beware of paradigm effects)
• Precise - constructs and measurement precision must be properly defined
• Falsifiable - theories must be stated in a way that allows them to be disproven

From observation to hypothesis

The path from the real world to testable science follows a chain:

• we observe phenomena we want to explain
• we create concepts (mental representations) and connect them through propositions
• we turn concepts into constructs - things in the real world that can be measured
• inside constructs, variables define current measurable values
• a hypothesis is a testable relationship between variables

IS research as design science

IS research is a “Design Science” - distinct from classical science (physics, math), natural science (biology, chemistry), and social science (sociology, psychology). It combines:

• science approaches (rigor, replicability, knowledge base, falsification)
• engineering principles (learning by doing, evaluating and testing, prototypes)

The IS research cycle has two main loops:

• Environment loop - meeting real-world needs
• Knowledge base loop - storing new information, principles, methods from experience

A central Design cycle loops through improving the artefact and testing/evaluating it.

---

Starting the research process (Ch. 3)

Motivation and research question

Before starting research, I need:

• a motivation (interesting observation, gap in knowledge, helping a business)
• a research question (the most important part)
  • “what/who/where” questions lead to more exploratory research
  • “how/why” questions lead to more explanatory research

Three ways of thinking

Deduction

• direction: general → specific
• start with existing theory, deduct a hypothesis, test it on new data
• key: falsifiability - trying to (dis)prove the theory
• formal: given a rule and a cause → deduce the effect

Induction

• direction: specific → general
• the “discovery” way - observe cases/phenomena, find patterns, derive general rules
• warning: beware of “weak induction”
• formal: given a cause and an effect → induce a rule

Abduction

• direction: observation → simplest/most likely explanation (Occam’s razor)
• essentially an educated guess, trial-and-error
• formal: given a rule and an effect → abduce a cause

Reasoning modes

• Exploration - discovering new phenomena or patterns (abduction, induction)
• Rationalization - building theoretical explanations (deduction, abduction)
• Validation - testing if explanations hold up against new data (induction, deduction)

Good research should combine all three modes.

Research decisions

Each research project positions itself on several spectrums:

• 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

Research methodologies

Each methodology has different strengths:

• Controllability - researcher’s control over events (strong in Design science)
• Deductibility - support for deductive reasoning (strong in Computational and Quantitative)
• Repeatability - can be repeated with same/similar results (strong in Design science and Computational)
• Generalisability - results can be generalized beyond the study (strong in Quantitative)
• Explorability - potential for unknown findings (strong in Qualitative)
• Complexity - can lead to exhaustive knowledge contributions (strong in Qualitative)

The five main methodologies are:

• Quantitative strategies
• Qualitative strategies
• Mixed methods
• Design science methods
• Computational methods

Systematic literature review (SLR)

Two ways to get information from existing literature:

• Ad hoc literature review - informal exploration, good for initial research proposals, not systematic or repeatable
• SLR - a predefined research strategy for answering a specific research question in an unbiased and rigorous way

Three phases of SLR

1) Planning the review

• define research questions for the literature review
• define search strategy (where to search, which terms/strings)
• define study selection criteria (inclusion/exclusion)

2) Conducting the review

• identify potential sources (automated search via WoS, Google Scholar, or manual backward/forward search)
• apply inclusion/exclusion criteria
• obtain full articles and extract data
• synthesize relevant information

3) Reporting the review

• report the protocol and deviations from it
• report search results and inclusion/exclusion outcomes
• report table of included studies with relevant information

---

Theory (Ch. 4)

Definition

A theory is a system of constructs and relationships between those constructs that collectively present a logical, systematic, and coherent explanation of a phenomenon of interest

What theory is NOT

• it is not universal (it has assumptions and boundary conditions)
• it is not self-perpetuating (it has implications, not an end in itself)
• it is not: pure data, description, prediction, design, a list of hypotheses, or an ideographic explanation of a single specific situation

Key building blocks

• Constructs (“What?“) - can be independent, dependent, mediating, or moderating
• Relationships (“How?“) - may be associative, compositional, unidirectional, bidirectional, conditional, or causal; they are testable
• Justifications (“Why?“) - theoretical grounding explaining why a relationship would exist
• Boundaries (“Who/Where/When?“) - scope conditions under which the theory holds; a theory without stated boundaries is unfalsifiable

Types of theories in IS

• Type I - Analysis - says what is; descriptive, no causal claims, no predictions
• Type II - Explanation - says what is, how, why, when, where; provides explanations but no testable predictions
• Type III - Prediction - says what is and what will be; testable propositions but weak causal grounding (common in ML/data-based research)
• Type IV - Explanation and Prediction (EP) - says what is, how, why, when, where, and what will be; the strongest type with both testable propositions and causal explanations
• Type V - Design and Action - says how to do something; prescriptions for constructing artefacts (Design Science Research)

Research usually moves from Type I upwards.

---

Qualitative research methods (Ch. 5a)

Qualitative research is about understanding phenomena in real-life context - when boundaries are not clear, phenomena are complex, context-dependent, or not well-defined.

Data collection techniques

Interviewing

• formats: descriptive, exploratory, explanatory
• semi-structured (between rigid survey and open conversation)
• benefits: rich, targeted, insightful
• drawbacks: reflexivity (interviewee responds with what the interviewer wants to hear), possible inaccuracy, bias

Observation

• direct observation - researcher is not involved (sitting in meetings, taking notes)
• participant observation - researcher participates (better insight, but influences what happens)

Documentation

• analyzing existing documents (meeting minutes, emails, project reports, system logs)
• valuable because they were not created for the research - no reflexivity
• types: structured, semi-structured, unstructured

Triangulation

Using multiple data sources or methods to study the same phenomenon:

• across sources (interviews + documents)
• across methods (quantitative + qualitative)
• across researchers
• across theories

Data analysis techniques

• Coding - assigning labels to chunks of data in three stages:
  • open coding - uncovering and labelling concepts with higher-level categories
  • axial coding - organizing concepts into causal relationships
  • selective coding - identifying central categories and relating other concepts to them
• Memoing - subjective reflection about what was happening (inside thoughts, connections)
• Critical incidents - examining series of events to explore relationships between constructs
• Content analysis - semantic analysis of text
  • conceptual (presence, frequency of concepts)
  • relational (how concepts relate)
• Discourse analysis - structuring and unfolding communication (“how it is said”)

Rigor in qualitative analysis

• Dependability (reliability) - another researcher with same process should reach similar conclusions
• Credibility (internal validity) - does it really reflect reality?
• Transferability (external validity) - do findings apply in a different setting? (not aiming for generalization, but patterns may be applicable elsewhere)

Case study

• investigating a current phenomenon within its real-life context in depth, using multiple data sources
• benefits: richness, depth, real-world context, new emerging concepts
• drawbacks: problems with controlled deduction, replicability, control mechanisms

Action research

• introducing changes/interventions and studying the effects
• the researcher is the agent of change
• two goals: solving current organizational problems + contributing to science
• iterative, cyclic process (similar to Lean startup)

Grounded theory

• a new theory inductively generated based on (grounded in) qualitative data
• characteristics:
  • focus on theory building, not testing
  • no pre-conceived hypothesis from prior domain knowledge
  • iterative process of collecting data, coding, and analysing

---

Quantitative research methods (Ch. 5b)

A set of techniques to answer research questions with quantitative data.

Typical procedure

• come up with a model, theory, and hypothesis
• develop or select concrete measurement instruments
• collect data
• analyse data (e.g. statistical modelling)
• evaluate results

Variables and confounders

• Constructs are general terms (e.g. “business/IT alignment”); variables are their operationalized, measurable versions
• types of variables:
  • Dependent variables - outcomes
  • Independent variables - predictors
  • Confounders - related to both dependent and independent variables; can create spurious relationships if not accounted for
  • Control variables - used to neutralize confounders, either by holding them constant or including them in the statistical model
  • Mediator variables - sit between independent and dependent variables (full mediation = everything goes through the mediator; partial = both direct and indirect effects exist)
  • Moderator variables - change the strength of the relationship between independent and dependent variables (act as catalysts)
• Conditional process modelling = combination of mediator and moderator influences
• variable types:
  • categorical (nominal, ordinal) - binary or polytomous
  • quantitative - interval and ratio measurement levels

Quality criteria

• Reliability - consistent results on replication
• Validity:
  • construct validity - methods measure what we intend
  • internal validity - observed effect is due to the hypothesized cause (risk: confounders)
  • external validity - relationship holds in other settings

Measurement

• Operationalization - choosing/creating a concrete method to measure a construct
• levels of measurement:
  • Nominal - differentiate between values (male/female)
  • Ordinal - differentiate and order (levels of agreement)
  • Interval - ordering + interpretable differences (temperature)
  • Ratio - meaningful zero point, relative comparisons possible (length, weight)
• measurement quality:
  • Validity - measuring the intended construct (influenced by systematic error)
  • Reliability - consistent results (influenced by random error); necessary but not sufficient for validity

Surveys and questionnaires

• Survey - list of questions covering a variety of topics
• Questionnaire - focuses on one construct or a related set of constructs
• rules: clear instructions, one construct measured by multiple questions, avoid complicated questions/double negations, exhaustive and mutually exclusive response options

Sampling

• Population → Sampling frame (actual subset to choose from) → Sample
• Stratum = a subset where elements share some characteristic
• probability sampling:
  • simple random sampling
  • systematic sampling (first element random, then predetermined pattern)
  • stratified random sampling (divide into strata, then random sample each)
• non-probability sampling:
  • convenience sampling
  • snowball sampling (small group recruits others)
  • purposive sampling (researcher’s judgment)

Statistical tests

• Correlation tests (Pearson, Spearman) - test relationships between variables (correlation ≠ causation)
• Regression analysis - relationship between independent variables and one dependent variable
• Tests of mean differences:
  • independent samples t-test - difference between two groups on a continuous dependent variable
  • independent samples z-test - difference between two groups on a proportion
  • ANOVA - difference between more than two groups (avoids running multiple t-tests)
  • ANCOVA - ANOVA that controls for covariates/confounders
• Tests of repeated measures (longitudinal):
  • paired samples t-test - one group, two time-points
  • repeated measures ANOVA - one group, more than two time-points
  • mixed ANOVA - multiple groups, multiple time-points

Latent constructs and models

Some variables (like happiness) cannot be measured directly - they are inferred through observed variables.

• Exploratory factor analysis - exploring which observed variables cluster under latent dimensions (without strong prior theory)
• Confirmatory factor analysis - formally testing a hypothesized factor structure against new data
• Structural equation modelling - regression with latent variables (measuring causal relationships between latent constructs)
• two types of latent constructs:
  • Reflective - the latent construct causes changes in its indicators
  • Formative - the indicators together form/cause the latent construct

---

Mixed methods and design science (Ch. 5c)

Mixed methods

Combining qualitative and quantitative methods within one study to get a better picture of “what, how much, why, and how.”

• all methods are based in one research question and finish in meta-inference (overall conclusion from integrating all method conclusions - they may complement, contradict, or converge)
• types: sequential, conversion, parallel, integration

Strengths

• qualitative adds meaning to quantitative results (and vice versa - numbers add precision)
• benefits from individual strengths of different methods
• easier to generate and test theory within one study
• broader range of research questions, stronger evidence, increased generalizability

Weaknesses

• difficult for a single researcher
• must learn multiple methods and how to mix them
• more resource-intensive and time-consuming
• some methodological details are still debated
• can be difficult to publish

Design science

A research paradigm in which a designer answers questions relevant to human problems via the creation of innovative artefacts, thereby contributing new scientific evidence to the body of knowledge.

---

Research philosophy (Ch. 1/Foundation)

The researcher’s philosophy is the outermost “layer” - it directly influences research questions, methods, data collection, and everything downstream. The key takeaway: same topic + different paradigm = different research question, methods, and contributions.

Research paradigms

A paradigm is built on these pillars (top to bottom):

• Ontology - how is reality defined?
• Epistemology - how do I know something? What is knowledge?
• Theoretical perspective - what approach to get knowledge?
• Methodology - what procedure to get knowledge?
• Methods - what tools to get knowledge?
• Sources - what data can we collect?

The research onion model shows how these layers nest into each other.

Philosophical identities

1. Positivism - “The natural scientist”

• reality is objective, fixed, measurable
• researcher is detached and value-free
• deductive approach, quantitative data
• typical methods: experiments, surveys
• criticism: strips away social context

2. Interpretivism - “The artist/actor”

• reality is socially constructed through culture and language
• researcher is part of what’s being researched
• inductive approach, qualitative data
• typical methods: case study, grounded theory
• criticism: accuracy - can you capture someone else’s lived experience?

3. Critical realism - “The archaeologist”

• bridges positivism and interpretivism
• reality is objective but layered - surface phenomena are caused by deeper hidden mechanisms
• abductive or deductive+inductive, qualitative or quantitative
• typical methods: archival research, survey
• criticism: lack of step-by-step methodological guidance

4. Postmodernism - “The political activist”

• reality is subjective, fragmented, discursively constructed
• challenges power structures, highlights marginalized views
• abductive or deductive+inductive, mainly qualitative
• criticism: limited practical solutions

5. Pragmatism - “The architect”

• reality is the practical consequences of ideas
• whatever works for solving problems in specific contexts
• abductive or deductive+inductive, any data type
• typical methods: mixed methods, action research
• criticism: researchers sometimes bundle data without truly integrating it