Normative vs. Empirical Analysis

Normative Analysis

• A prescriptive approach that focuses on what should be done
• Makes judgments about right and wrong based on reason and logic, not data or observation
• Arguments are accepted/rejected based on the reasoning and premises
• Often found in political theory and philosophy where arguments are built on value judgments
• Arguments must take a position based on some set of values and cannot be value-neutral

Empirical Analysis

• A descriptive or explanatory approach
• Focuses on understanding "what is" and "why" things happen
• Relies on observation & measurement of the physical/social world
• Arguments are judged based on the quality and existence of evidence
• Evidence-based and objective, unlike normative analysis which is value-based
• It can strengthen normative claims and empirical research can challenge normative assumptions

Intersubjectivity

• The practice of independent researchers studying the same issue
• Helps ensure findings aren't tied to a specific researcher, method, or context
• Increases reliability and credibility through comparison of results
• Shows findings reflect real-world consistency, not one-off occurrences

Scientific Approach (Positivism)

• Relies on shared epistemology (understanding knowledge) and methodology (how we obtain knowledge)
• Epistemology asks, "What is knowledge?" and "How do we know we know something?"
• Methodology refers to how valid knowledge is gathered
• Beliefs:
  • Knowledge comes from real-world observation (empiricism)
  • Everything has a discoverable cause (determinism)
  • Science aims for an unbiased reality representation (objectivity)
  • Research should be repeatable and confirmed (replication)

Scientific Approach Steps

• Propose a Research Question:
  • Descriptive ("What?") or explanatory ("Why?") question
  • Review existing literature to build on knowledge and avoid redundancy
  • Consider intersubjectivity and replication to ensure cross-checking
• Gather Empirical Evidence:
  • Evidence is referred to as data (datum singular)
  • Data is either qualitative (text-based) or quantitative (numerical)
  • Different methods are for small-n vs. large-n studies and for qualitative vs. quantitative approaches
  • Mixed methods combine both data types
• Communicate Your Results:
  • Communication is essential for knowledge advancement
  • Established researchers publish in peer-reviewed outlets
  • Students/junior researchers communicate through supervisors or early publications

Theory in Research

• Provides a simplified explanation of the world
• Explains, predicts, and understands relationships between concepts
• Abstract representation of ideas, not specific examples

Theory Applications

• Explaining past occurrences
• Predicting future outcomes
• Explaining differences between cases/regions
• Explaining change over time
• Concepts represent broader ideas

Hypothesis

• Researchers use concepts to develop testable hypotheses
• Theory example: voters are less likely to re elect a governing party during an economic recession
• Hypothesis example: governing parties are more likely to lose in elections called during a recession
• Tested through historical data

Research Types

• Theory-oriented research answers theory-driven questions
• Basic research is theory-driven
• Applied research uses theory to address real-world problems

Theory Effectiveness

• Parsimonious: simple with the fewest elements
• General: explains multiple events
• Falsifiable: capable of being proven wrong

Theory Development

• Balance theory-building and theory-testing research
• Theory-testing research (deductive) tests a hypothesis from a theory using quantitative methods
• Goal is to test against real-world data, not to "prove" the theory

Theory Review

• More confidence results from supporting evidence
• Revision is required for partially supporting evidence
• Reconsideration is required for non-supporting evidence
• Theories guide political scientists and provide frameworks

Concepts

• Abstract tools to organize and classify phenomena
• Theories are made of concepts to focus on relationships between variables
• Theories explore relationships between dependent (outcome) and independent (explanatory) events
• Theories aim to establish a causal explanation of concept relations

Concepts are useful in research if they have

• A clear specified definition: define what the concept and is not
• Concepts can take forms such as typologies or continuums and can be unidimensional or multidimensional
• Concepts can be concrete or abstract

Concepts for Research

• Clearly expressed to prevent misunderstanding
• Contextually relevant and links to the broader literature
• Well-defined: measured in a way that aligns with the conceptual definition

Researcher's concept applications

• Select from existing definitions
• Adapt existing definitions
• Create entirely new concepts

Concept's Value

• Relies on how well its conceptualized and operationalized in the study

Measurement

• Operationalization creates a working version of a concept for research
• A measure is a tool to gather observable evidence and the project will fail if it does not work

Measurement based on research type

• Qualitative researchers identify themes in data
• Quantitative researchers define variables and indicators

Variables

• More concrete representation of a concept being measured

Indicators

• How to capture value of a variable
• It can be challenging to categorize and record information accurately

Indexes

• Created if a concept is too complex
• Combination of several variables representing different facets of a concept

Correlation vs Causality

• Correlation is when 2 variables have a predictable relationship
• Causality is when a change in one variable causes a change in another
• Correlation alone doesn't equals causality
• A hypothesis specifies the relationship including which variable causes the other

Criteria for Causality

• Correlation: variables have to be correlated
• Temporal Order: cause has to happen before the effect
• Absence of confounding variables. There should be no third factor influencing both variables
• Plausible causal mechanism: theory has to explain why it makes sense
• Consistency: relationship has to be consistent over time/in different contexts

Types of Hypotheses

• Null Hypothesis: There is no relationship between the variables
• Research Hypothesis: The relationship between variables exists and is usually directional (positive or negative correlation).

Additional Variable Types

• Intervening: Comes between the IV and DV
• Conditional: Influences the strength of the relationship
• Reinforcing: Amplifies the relationship
• Spurious: A third factor that influences both the IV and DV, creating misleading results

Levels of measurement

• Variables can be measured at different levels which influences the analysis

Types of concepts

• Categorical: difference in kind; produces nominal variables (no order)
• Continuous: Difference in amount; produces ordinal, interval or ratio variables
  • ranked
  • value distances
  • variables with a true zero

Determining Level of Measurement

• Align measurement with concept
• Higher-level measurements offer more options
• Accuracy and generalizability are afforded to lower-level measures

Measurement error

• Difference between true vs measured variable value
• Random error: random
• Non-random error: flaws

Reducing Measurement Error

• Ensure results across time and individuals
• Validate measure accurately represent a concept, and adjust if not

Small-N research

• N is # of cases/units in a research project
• Large-N research is quantitative, focusing on cases to make representative results
• Small-N research is qualitative, has fewer cases, and in-depth analysis is done

Case Study

• Case Study research: Studying a single case in detail to gain insights
• Focus: gathering detailed information about cases

Comparative research

• Comparative research: Analyzing multiple cases; trying to identify similar/different factors
• Goal: understand causal factors

Small-N Studies

• Good at:
  • Correlation: examining fewer cases to observe relationships/patterns
  • Plausible Causal Mechanism: detecting mechanisms due to in-depth focus on each case
  • Temporal Order: establishing causes can be easier with longitudinal approach
• Harder at:
  • Absence of confounding variables: challenging with fewer cases
  • consistency. Difficult due to focus on specific cases

Case Study Research (approach)

• Take approach w/o preconditions or artificial limitations
• Allows focus on comprehensive understanding

Types of case studies

• Descriptive case study:
  • Aims to describe a phenomenon in detail
  • it helps identify and gives foundation
  • helps improve theories
  • internal validity due to access
  • harder to generalize given its descriptive nature
• Theory-testing case study:
  • challenge or test theories
  • assesses whether it needs modification

Other aspects

• Focus: outliers; situations where theory does not apply
  • Examples:
    • failed most-likely case: theory supposed to apply
    • successful least-likely case: theory shouldn't apply but does

Process Tracing

• Purpose: to show links b/w observations within studies
• Key questions:
  • Is there a distinction b/w subject and object?
  • is conceptualization rigorous?
  • does the study employ tracing properly?
• Sources for case study data
  • Primary, archival, discourse analysis, interviews

Comparative Research

• Useful increasing understanding helping overcome assumptions
• Example used was race and voting, highlighting race's influence depending on party and political system

Comp. System traits

• The U.S and Canada share traits but differ in their political structures

Case selection

• Crucial, as comparative research is often small-N
• selection must capture representative cases

Purposive Sampling

• Useful selects for relevant cases

Types of Design

• Most similar System Design:
  • Goal: Investigate difference; helps remove certain variations
• Most different System Design:
  • Goal: Investigate similarities between different systems; helps find common systems

Choosing Designs

• depends on how closely cases reflect each other

Equivalent Measures

• Must measure underlying concepts
• Operationalizing defines what measurement has in store
• the challenge is ensuring accuracy