Theory Development and Hypothesis Generation Department of - - PowerPoint PPT Presentation

▶
theory development and hypothesis generation
SMART_READER_LITE
LIVE PREVIEW

Theory Development and Hypothesis Generation Department of - - PowerPoint PPT Presentation

Jun 11, 2023 318 likes •1.01k views

Theory Generating Hypotheses Hypothesis Testing Preview Theory Development and Hypothesis Generation Department of Government London School of Economics and Political Science Theory Generating Hypotheses Hypothesis Testing Preview 1

slide-1
SLIDE 1

Theory Generating Hypotheses Hypothesis Testing Preview

Theory Development and Hypothesis Generation

Department of Government London School of Economics and Political Science

slide-2
SLIDE 2

Theory Generating Hypotheses Hypothesis Testing Preview

1 Theory 2 Generating Hypotheses 3 Hypothesis Testing 4 Preview

slide-3
SLIDE 3

Theory Generating Hypotheses Hypothesis Testing Preview

1 Theory 2 Generating Hypotheses 3 Hypothesis Testing 4 Preview

slide-4
SLIDE 4

Theory Generating Hypotheses Hypothesis Testing Preview

Scientific method

1 Research question(s) 2 Clarify the core concepts 3 Develop theory 4 Derive specific, testable hypotheses 5 Plan data collection 6 Gather data/evidence 7 Analyze data 8 Draw inferences

slide-5
SLIDE 5

Theory Generating Hypotheses Hypothesis Testing Preview

What is a theory?

Definition: A tentative conjecture about the causes of some phenomenon of interest1

1Kellstedt and Whitten, p.3

slide-6
SLIDE 6

Theory Generating Hypotheses Hypothesis Testing Preview

What is a theory?

Definition: A tentative conjecture about the causes of some phenomenon of interest1

Another way of saying this: An argument that attempts to explain how concepts are causally related

1Kellstedt and Whitten, p.3

slide-7
SLIDE 7

Theory Generating Hypotheses Hypothesis Testing Preview

What is a theory?

Definition: A tentative conjecture about the causes of some phenomenon of interest1

Another way of saying this: An argument that attempts to explain how concepts are causally related

1Kellstedt and Whitten, p.3

slide-8
SLIDE 8

Theory Generating Hypotheses Hypothesis Testing Preview

What is a theory?

Definition: A tentative conjecture about the causes of some phenomenon of interest1

Another way of saying this: An argument that attempts to explain how concepts are causally related

Theories contain:

Outcome or dependent variable, to be explained

1Kellstedt and Whitten, p.3

slide-9
SLIDE 9

Theory Generating Hypotheses Hypothesis Testing Preview

What is a theory?

Definition: A tentative conjecture about the causes of some phenomenon of interest1

Another way of saying this: An argument that attempts to explain how concepts are causally related

Theories contain:

Outcome or dependent variable, to be explained Explanatory, independent, or causal variable(s) thought to affect the outcome

1Kellstedt and Whitten, p.3

slide-10
SLIDE 10

Theory Generating Hypotheses Hypothesis Testing Preview

What is a theory?

Definition: A tentative conjecture about the causes of some phenomenon of interest1

Another way of saying this: An argument that attempts to explain how concepts are causally related

Theories contain:

Outcome or dependent variable, to be explained Explanatory, independent, or causal variable(s) thought to affect the outcome A mechanism (“how”) that links the two

1Kellstedt and Whitten, p.3

slide-11
SLIDE 11

Theory Generating Hypotheses Hypothesis Testing Preview

Smoking Cancer Sex Environment Genetic Predisposition Parental Smoking

slide-12
SLIDE 12

Theory Generating Hypotheses Hypothesis Testing Preview

Theory vs. Framework

Theories are general statements about causal relationships Testing a theory involves:

Stating observable implications of theory (and rival theory/ies) Examining whether evidence is consistent

  • r inconsistent with expectations
slide-13
SLIDE 13

Theory Generating Hypotheses Hypothesis Testing Preview

Theory vs. Framework

Theories are general statements about causal relationships Testing a theory involves:

Stating observable implications of theory (and rival theory/ies) Examining whether evidence is consistent

  • r inconsistent with expectations

Applying an interpretive framework involves merely focusing on particular features of an empirical phenomenon

slide-14
SLIDE 14

Theory Generating Hypotheses Hypothesis Testing Preview

Key Points!

1 Theory is about concepts 2 Analysis is about measured variables

Everything from MT

3 So our task as scientists is to:

Find observable implications of theory Draw theoretical implications from measures

slide-15
SLIDE 15

Theory Generating Hypotheses Hypothesis Testing Preview

Independent Variable (Measured) Dependent Variable (Measured) Independent Variable (Concept) Dependent Variable (Concept)

slide-16
SLIDE 16

Theory Generating Hypotheses Hypothesis Testing Preview

Independent Variable (Measured) Dependent Variable (Measured) Independent Variable (Concept) Dependent Variable (Concept)

Analysis Theory

slide-17
SLIDE 17

Theory Generating Hypotheses Hypothesis Testing Preview

slide-18
SLIDE 18

Theory Generating Hypotheses Hypothesis Testing Preview

Generating Theory I

One way to theorize is to reason inductively Induction works by drawing generalities from specific observations Sometimes called “bottom-up” theorizing

slide-19
SLIDE 19

Theory Generating Hypotheses Hypothesis Testing Preview

Generating Theory II

An alternative way of developing theory is through deduction Deduction begins from general, assumed principles/axioms to reach more specific

  • bservable realities
slide-20
SLIDE 20

Theory Generating Hypotheses Hypothesis Testing Preview

Generating Theory II

An alternative way of developing theory is through deduction Deduction begins from general, assumed principles/axioms to reach more specific

  • bservable realities

Common example: Rational choice theory

slide-21
SLIDE 21

Theory Generating Hypotheses Hypothesis Testing Preview

Generating Theory III

“The Calculus of Voting” is a rational choice theory

Assumes utility maximization is the driver

  • f all behaviour

Understanding phenomena is a matter of figuring out utility structures, especially those created by institutions

slide-22
SLIDE 22

Theory Generating Hypotheses Hypothesis Testing Preview

The Calculus of Voting

Theory: Voting is explained by 3 factors Costs of voting Benefits from preferred alternative winning Probability of impacting result Pr(Voting) Costs Benefits Pr(Impact)

slide-23
SLIDE 23

Theory Generating Hypotheses Hypothesis Testing Preview

The Calculus of Voting

Theory: Voting is explained by 4 factors Costs of voting Benefits from preferred alternative winning Probability of impacting result Benefits from voting per se Pr(Voting) Costs Benefits Pr(Impact) “D term”

slide-24
SLIDE 24

Theory Generating Hypotheses Hypothesis Testing Preview

Aside: Assumptions

If a theory requires assumptions, is that theory credible?

slide-25
SLIDE 25

Theory Generating Hypotheses Hypothesis Testing Preview

Generating Theory III

“The Calculus of Voting” is a rational choice theory

Assumes utility maximization is the driver

  • f all behaviour

Understanding phenomena is a matter of figuring out utility structures, especially those created by institutions

slide-26
SLIDE 26

Theory Generating Hypotheses Hypothesis Testing Preview

Generating Theory III

“The Calculus of Voting” is a rational choice theory

Assumes utility maximization is the driver

  • f all behaviour

Understanding phenomena is a matter of figuring out utility structures, especially those created by institutions

Not the only broad theoretical paradigm

slide-27
SLIDE 27

Theory Generating Hypotheses Hypothesis Testing Preview

The Michigan Model

Theory: Vote choice is explained by long-standing partisan identification, which is in turn shaped by early socialization. Vote Choice Party Identification Parental Party ID Early Adult Experiences

slide-28
SLIDE 28

Theory Generating Hypotheses Hypothesis Testing Preview

Induction vs. Deduction?

Induction and deduction are both integral to science Theory testing and theory building both require observation

slide-29
SLIDE 29

Theory Generating Hypotheses Hypothesis Testing Preview

Theory Generation in Practice

As you theorize an explanation for some phenomenon, you will draw on: General principles Extant theory Specific evidence

slide-30
SLIDE 30

Theory Generating Hypotheses Hypothesis Testing Preview

What makes for a good theory?

Truth Relevance Coherence Falsifiability Precision Generality Parsimony

slide-31
SLIDE 31

Theory Generating Hypotheses Hypothesis Testing Preview

Generality & Parsimony

Think for 90 seconds about each of these principles:

Generality: Theories that can explain more are preferred over theories that can explain less Parsimony: Simple theories are preferred over complex theories

Are these principles defensible? Are they any good?

slide-32
SLIDE 32

Theory Generating Hypotheses Hypothesis Testing Preview

1 Theory 2 Generating Hypotheses 3 Hypothesis Testing 4 Preview

slide-33
SLIDE 33

Theory Generating Hypotheses Hypothesis Testing Preview

Hypotheses

2Toshkov p.64 3Kellstedt and Whitten (p.4)

slide-34
SLIDE 34

Theory Generating Hypotheses Hypothesis Testing Preview

Hypotheses

Definitions:

  • bservable implications; testable propositions

entailed by the logic of the theory2 a theory-based statement about a relationship that we expect to observe3

2Toshkov p.64 3Kellstedt and Whitten (p.4)

slide-35
SLIDE 35

Theory Generating Hypotheses Hypothesis Testing Preview

Hypotheses

Definitions:

  • bservable implications; testable propositions

entailed by the logic of the theory2 a theory-based statement about a relationship that we expect to observe3

Features

Derived from theory Specific Empirical/observable Causal (“if-then” logic)

2Toshkov p.64 3Kellstedt and Whitten (p.4)

slide-36
SLIDE 36

Theory Generating Hypotheses Hypothesis Testing Preview

How do we generate hypotheses?

Think about observable implications What would evidence consistent with this theory be? What would evidence inconsistent with this theory be?

slide-37
SLIDE 37

Theory Generating Hypotheses Hypothesis Testing Preview

How do we generate hypotheses?

Think about observable implications What would evidence consistent with this theory be? What would evidence inconsistent with this theory be?

This is falsifiability

slide-38
SLIDE 38

Theory Generating Hypotheses Hypothesis Testing Preview

Ex.: Broad Street Cholera

1854 outbreak of cholera in London

Around Broad Street (Soho) 616 eventual deaths

slide-39
SLIDE 39

Theory Generating Hypotheses Hypothesis Testing Preview

Ex.: Broad Street Cholera

1854 outbreak of cholera in London

Around Broad Street (Soho) 616 eventual deaths

What causes transmission of cholera?

slide-40
SLIDE 40

Theory Generating Hypotheses Hypothesis Testing Preview

Ex.: Broad Street Cholera

1854 outbreak of cholera in London

Around Broad Street (Soho) 616 eventual deaths

What causes transmission of cholera? Dominant theory at time: “miasma”

slide-41
SLIDE 41

Theory Generating Hypotheses Hypothesis Testing Preview

Ex.: Broad Street Cholera

1854 outbreak of cholera in London

Around Broad Street (Soho) 616 eventual deaths

What causes transmission of cholera? Dominant theory at time: “miasma” Hypotheses:

Clean up garbage → ↓ cholera Open windows → ↓ cholera

slide-42
SLIDE 42

Theory Generating Hypotheses Hypothesis Testing Preview

slide-43
SLIDE 43

Theory Generating Hypotheses Hypothesis Testing Preview

Observational Equivalence

Definition: All hypotheses for two (or more) theories are identical What to do?

Generate more specific expectations Move outside scope conditions Settle for lack of explanation

slide-44
SLIDE 44

Theory Generating Hypotheses Hypothesis Testing Preview

Median Voter Theory of Legislatures

slide-45
SLIDE 45

Theory Generating Hypotheses Hypothesis Testing Preview

Median Voter Theory of Legislatures

m

slide-46
SLIDE 46

Theory Generating Hypotheses Hypothesis Testing Preview

Median Voter Theory of Legislatures

m q

slide-47
SLIDE 47

Theory Generating Hypotheses Hypothesis Testing Preview

Median Voter Theory of Legislatures

m q

slide-48
SLIDE 48

Theory Generating Hypotheses Hypothesis Testing Preview

Median Voter Theory of Legislatures

m q If this is true, why do we sometimes see policies left of m in a legislature?

slide-49
SLIDE 49

Theory Generating Hypotheses Hypothesis Testing Preview

Three Competing Theories

Committee Gatekeeping

m q

Party Cartels

m q

Pivotal Politics

m q

Adapted From: Krehbiel, Keith. 1999. “Paradoxes of Parties in Congress.” Legislative Studies Quarterly 24(1): 31–64.

slide-50
SLIDE 50

Theory Generating Hypotheses Hypothesis Testing Preview

Three Competing Theories

Committee Gatekeeping

m q c

Party Cartels

m q

Pivotal Politics

m q

Adapted From: Krehbiel, Keith. 1999. “Paradoxes of Parties in Congress.” Legislative Studies Quarterly 24(1): 31–64.

slide-51
SLIDE 51

Theory Generating Hypotheses Hypothesis Testing Preview

Three Competing Theories

Committee Gatekeeping

m q c

Party Cartels

m q = r c

Pivotal Politics

m q

Adapted From: Krehbiel, Keith. 1999. “Paradoxes of Parties in Congress.” Legislative Studies Quarterly 24(1): 31–64.

slide-52
SLIDE 52

Theory Generating Hypotheses Hypothesis Testing Preview

Three Competing Theories

Committee Gatekeeping

m q c

Party Cartels

m q = r c

Pivotal Politics

m q p s

Adapted From: Krehbiel, Keith. 1999. “Paradoxes of Parties in Congress.” Legislative Studies Quarterly 24(1): 31–64.

slide-53
SLIDE 53

Theory Generating Hypotheses Hypothesis Testing Preview

1 Theory 2 Generating Hypotheses 3 Hypothesis Testing 4 Preview

slide-54
SLIDE 54

Theory Generating Hypotheses Hypothesis Testing Preview

Hypothesis Testing

Multiple schools of thought History is conflictual and murky Two strands of literature

Philosophy of science Statistics

slide-55
SLIDE 55

Theory Generating Hypotheses Hypothesis Testing Preview

Principle of Hypothesis Testing

1 Identify and collect data

slide-56
SLIDE 56

Theory Generating Hypotheses Hypothesis Testing Preview

Principle of Hypothesis Testing

1 Identify and collect data 2 Data should include:

Outcome variable(s), X Explanatory variable(s), Y Covariates (theoretically unimportant explanations of the outcome)

slide-57
SLIDE 57

Theory Generating Hypotheses Hypothesis Testing Preview

Principle of Hypothesis Testing

1 Identify and collect data 2 Data should include:

Outcome variable(s), X Explanatory variable(s), Y Covariates (theoretically unimportant explanations of the outcome)

3 Need variation on both X and Y

slide-58
SLIDE 58

Theory Generating Hypotheses Hypothesis Testing Preview

Principle of Hypothesis Testing

1 Identify and collect data 2 Data should include:

Outcome variable(s), X Explanatory variable(s), Y Covariates (theoretically unimportant explanations of the outcome)

3 Need variation on both X and Y 4 Test difference between outcomes when

(possibly) causal variable differs

slide-59
SLIDE 59

Theory Generating Hypotheses Hypothesis Testing Preview

Forms of Hypothesis Testing Null hypothesis

Begin with null hypothesis Your hypothesis expects an alternative state of the world c/o Ronald Fisher

Alternative hypotheses

Begin with 2(+) alternative hypotheses Accept hypothesis consistent with

  • bservation

c/o Jerzy Neyman and Egon Pearson

slide-60
SLIDE 60

Theory Generating Hypotheses Hypothesis Testing Preview

Fearon’s Counterfactuals

Sometimes we cannot test our hypothesis with actual observations What does Fearon suggest we do?

slide-61
SLIDE 61

Theory Generating Hypotheses Hypothesis Testing Preview

A Good Test

Correct level of analysis Within scope conditions of theory Well-defined concepts Measures of high construct validity, accuracy, and precision Possible to observe any correlation between potential cause and outcome Consistent with or an improvement upon past methods Test using different data than data used to generate theory

slide-62
SLIDE 62

Theory Generating Hypotheses Hypothesis Testing Preview

Some Testing Challenges

1 Deterministic and probabilistic causality 2 Effect heterogeneity 3 Multiple causation 4 Equifinality 5 Confirmation or disconfirmation bias

slide-63
SLIDE 63

Theory Generating Hypotheses Hypothesis Testing Preview

1 Theory 2 Generating Hypotheses 3 Hypothesis Testing 4 Preview

slide-64
SLIDE 64

Theory Generating Hypotheses Hypothesis Testing Preview

Theory testing involves: Between-case comparisons, or Across-time comparisons, or Between-case & across-time comparisons Within-case comparisons at a lower level

  • f analysis
slide-65
SLIDE 65

Theory Generating Hypotheses Hypothesis Testing Preview

Methods of theory testing include: Case comparisons Process-tracing Observational statistical comparisons

Tabulation/visualization Bivariate statistical inference (e.g., t-tests) Multivariate analysis (e.g., regression)

Experimentation

slide-66
SLIDE 66

Theory Generating Hypotheses Hypothesis Testing Preview

Preview of Next Week

How do we make between-case comparisons to test theories?

slide-67
SLIDE 67