proposing explanations, framing hypothesis, and making comparisons

first goal of poli sci

define and measure concepts

describe the measurement process

• begin with conceptual term and turn it into concrete and operationalized term

  • ex: what is political tolerance?

• describe it’s variation - transform variables to analyze them more effectively

• then we ask why it varies and how its variation affects other variables

second goal of poli sci

to propose and test explanations for political phenomena

• goals isn’t defined by what, but why

  • why do some people prefer abortion rights while some don’t?

  • why do some people vote and others don’t?

  • why do some people vote democratic and some republican?

theory

general statement about a causal relationship

• help us distill the essential features of social systems

• help us explain what happened in the past and predict what will happen in the future

model

simplified, abstract representation of some larger and more complicated subject

• ex: street map - not that realistic, but lets people know how it works and where to go

  • we don’t need all the info to know what’s going on

• used to develop ideas

• can theorize about the effect of variables on the outcome of interest

• makes the subject easier to understand

causal diagram

visual representation of the causal relationships between variables or factors in a system based on specific symbols and conventions

• you can depict variables as nodes on them

• draw arrows between nodes to show how they’re related

  • direction of arrow shows causality - which one influences the other

  • x → y

• adding nodes and edges: we can represent both direct and indirect causal representation among variables

probabilistic explanations

help up understand the general patterns and essential features of politics

two kinds of theories/explanations

probabilistic and deterministic

probabilistic theories

there are always exceptions to the rules

• not an iron-clad rule, other things could play as a factor

• dashed lines indicate that there is an unobserved variable that may be influencing the variables in the diagram

• time studying → test score ← - other factors

deterministic explanations

theories that leave no room for error

why is probabilistic thinking so important in poli sci research?

• involves important difference between causation in social sciences and deterministic models used in physical sciences

• humans are unpredictable, we can’t make any deterministic explanations because anything can happen

how to generate plausible explanations

break the subject into components you can scientifically study

• identify the outcome of interest; add outcome variable to causal diagram

• think about relevant decision-makers

  • ex: if interested in elections, realize that ballots and campaign ads don’t make the decision, the people do

  • ex: people who ran for elections: candidates who ran, people who gave them money, and people who voted for them

fishbone diagram

(aka cause-and-effect or ishikawa diagram) tool used to identify and organize potential causes

• provides structured approach to brainstorming by breaking down causes into categories

  • category examples: people, methods, machines, materials, measurements, and environment

  • can and should be customized based on the context and nature of the analysis

• encourages a comprehensive examination of potential causes

rational actors

makes thoughtful and deliberate decisions to advance their own interests

social-psychological actor

makes decisions based on gut feelings rather than thoughtful analysis

what can both rationals and social-psychological actors help us do?

develop and test theories about the way people make decisions and interact with one another

two kinds of reasoning

inductive and deductive

inductive reasoning

• starts by proposing explanations for a specific case, often based on personal experience and observation

  • specific observations → general conclusions

deductive reasoning

• starts with general premises and derives the logical implications of those premises

  • based on abstract logic and reasoning

  • can be stated explicitly and their implications derived by using logical and mathematical statements

  • tests using info about specific cases to which generalizations should apply

explaining why proposing explanations is the essence of creative research

• invites us to think up possible reasons for the observed differences between subjects

• but causes and explanations can be subject to change

three criteria for a good explanation

• describes connection between dependent variable (opinions about SS) and causal relationship (partisanship)

• asserts direction or tendency of this difference

• testable

according to that one author, why was paul revere successful on delivering the message about the british troops?

he was more popular than william dawes

causal mechanism

internal link that acts as a go-between or mediator between an independent and dependent variable

• british troops → messenger → public response

  • messenger would be the causal link

why might a causal mechanism be important in a study?

• enriches research and helps us identify additional hypotheses to test

• linkages also called mediators or intervening variables

• if the links on the causal chain are weak, then the explanation is weak

why is simplifying a causal diagram important?

to clearly represent causal relationships

• if some variables are empirical characteristics of the same concept or share a common cause, represent that underlying concept or common cause with a single node

• also possible to aggregate multiple variables using variable transformations

how to simplify a causal diagram

focus on the main variables rather than minor one

• if there are variables in the diagram that do not add substantial information, consider removing them

• if there are intervening variables that are not the primary focus, they can be omitted to simplify the diagram

• these strategies can reduce the number of variables without sacrificing essential ideas

types of junctions to describe causal relationships

collider, chain, fork

collider

• two variables are connected to an effect variable, but two causes are not directly connected to each other

  • A → C ← B

  • ex: time spent studying and other factors affect grades

  • ex: influence of electoral competition and election day weather on voter turnout

    • competitive campaigns likely increase voter turnout, as does good weather on election day, but there is no direct relationship between electoral competition and weather on election day

chain

• one variable causes the other, which in turn causally influences a third variable

  • connected in linear sequence

  • A → B → C

fork

• two variables share a common cause but are not directly related to each other

  • A ← C → B

  • bowling alone is one effect of declining civil engagement; it is related to the decline of civil associations because of a common cause, but there is no direct connection between bowling leagues and civil associations

hypothesis

testable statement about the empirical relationship between cause and effect

research hypothesis

testable statement about the empirical relationship between an independent variable and a dependent variable

• tells us exactly how different values of the IV are related to different values of the dependent variable

• ex: In a comparison of individuals, those who are Democrats will be more likely to favor increased spending on Social Security than will those who are Republicans.

  • tells us that when we compare units of analysis (individuals) having different values of the independent variable (partisanship), we will observe a difference in the dependent variable (support for Social Security spending)

name one mistake that people make when making their hypothesis

sometimes they tone down the technicality to make it more readable

• though this can prove effective, prioritize clarity first

null hypothesis

asserts the there is no relationship between the independent and dependent variables

• used to translate research and null hypotheses into inequality statements

• labeled as H₀

alternative hypothesis

research hypothesis, suggests that there is a relationship between IV and DV

• labeled as H_A

cross tabulations

table that summarizes the relationship between two variables measured at the nominal or ordinal level

• shows the distribution of cases across the values of a DV for cases that have different values on an IV

• each column contains raw frequency and percentage of cases falling into each category of the dependent variables

• column values show the conditional distribution of the DV

• total column on the right shows the marginal distribution of the DV

what are cross tabulations used for?

to make comparisons when the dependent and independent variable are both measured at the nominal or ordinal level

conditional distribution

distribution of the DV”s value within groups by the IV; reported column in a cross-tabulation

marginal distribution

overall distribution of a DV’s values; reported in the margin of a cross-tabulation

rules for cross tabulations

1. set it up so that the categories of the IV define the columns of the table and values of the DV define the rows

   1. IV: raw frequencies falling into each category of the DV are displayed totaled at the bottom of each column

2. always calculate percentages by categories of the IV, never the percentages by categories of the DV

   1. most essential but most frequently violated rule

   2. we are showing how changes in the IV affect the DV

3. interpret a cross-tabulation by comparing percentages w/ a given value of the DV as you move across from one column to the next

mean comparison table

shows the mean of a DV for cases that have different values on an IV