February 5, 2018

Concepts

From Theory to Empirics

Until now:

Posing questions, making claims

Now

Answering questions, evaluating claims

From Theory to Empirics

What is the first step?

Some claims:

The election of ethnic political parties cause an increase in ethnic violence.

Democratic governments improve the health of their citizens.

Technologies that favor offensive over defensive warfare make war more likely to occur.

Introducing proportional representation causes a decline in the accountability of legislators.

Before we evalute these claims

What are the first things we need to know?

Before we evalute these claims

The election of ethnic political parties cause an increase in ethnic violence.

What are…

  • ethnic political parties?
  • ethnic violence?

What do these mean? How would we recognize them?

Before we evalute these claims

Democratic governments improve the health of their citizens.

What are…

  • democracies?
  • health?

What do these mean? How would we recognize them?

Before we evalute these claims

Technologies that favor offensive over defensive warfare make war more likely to occur.

What are…

  • Offensive technologies?
  • Defensive technologies?
  • Wars?

What do these mean? How would we recognize them?

Before we evalute these claims

Introducing proportional representation causes a decline in the accountability of legislators.

What is

  • Proportional representation?
  • Accountability?

What do these mean? How would we recognize them?

What we need:

Concepts: abstract definitions that are used to describe phenomena, groups, individuals, and their attributes.

  • Help us clarify what causal or descriptive claims mean.

Concepts:

More attributes

  • General, not specific
    • no proper nouns

Concepts:

Aren't these really easy?

Concepts:

What is a …

Democracy

Concepts:

What is a …

War

A procedure:

  1. List all cases where concept clearly applies…
    • E.g., everyone would agree that Canada is a democracy
  2. Propose a "covering definition".
    • A definition that would apply concept to all clear cases
  3. Does this definition include other cases that intuitively do not belong
    • This includes both actual or hypothetical cases
  4. Go back to 2, reformulate definition
    • avoid many of conditions using "or", add criteria to fit a sintle idiosyncratic case

A procedure:

Repeat 3 and 4 until:

  • No more exceptions you can think of
  • Remaining exceptions are borderline cases where no one has a strong intuition on whether they fit the category

An example: "Ethnic violence"

1. What are the "clear" cases?

  • Holocaust, Rwandan Genocide, Ethnic Cleansing in Balkans, KKK attacks against African Americans

An example: "Ethnic violence"

2. A "covering" definition?

  • "ethnic violence is violence between ethnic groups"

An example: "Ethnic violence"

"ethnic violence is violence between ethnic groups"

3. Are there exceptions to this rule?

  • white American mugged by an African American?
  • fight between fans of Brazil and Argentina national football teams?

Fit the rule, but are they "ethnic violence"

  • Is mugging "economic crime" rather than "hate crime"?

An example: "Ethnic violence"

  • Are attacks on an ethnic group to take their land, jobs, material goods "ethnic violence" in absence of "hatred"?
  • What if soldier from separatist armed group plants a bomb that kills many from a different ethnic group. Soldier may lack hatred, does it count?
  • Could mugging be ethnic violence if ethnicity of victim is pretext for theft?
  • Can wars between countries be "ethnic violence"?

An example: "Ethnic violence"

Refined definition:

Violence is "ethnic" if it involves members of different ethnic groups and either:

  1. it is motivated by hatred or dislike of ethnic others;

  2. The criterion for selecting victims is ethnicity;

  3. it is committed on behalf of in in the name of an ethnic group, or against those claiming to represent an ethnic group (and because of that)

and at least one of the actors is not a state.

Another concept: "natural amenities"

In the 1990s, the US government sought to evaluate the best and worst places in American with respect to the natural scenery and climate.

natural amenities: "physical characteristics of an area that enhance the location as a place to live"

Natural amenities

Natural amenities

Vancouver would probably rank "very high" on natural amenities.

But

Just what physical characteristics "enhance a location as a place to live"?

Natural amenities

Several dimensions of "natural amenities"

According to the US Department of Agriculture:

  • Warm winters
  • Winter sunlight
  • Temperate summer
  • Low summer humidity
  • Topographic variation
  • Surface water

Dimensions of a concept:

dimensions of a concept: phenomena that both:

  1. Are part of the concept

  2. Could exist, disappear, change independently of one another. They are not redundant

Dimensions are like "sub-concepts"

Why do dimensions matter?

  • Clarify causal logics
  • Moving from concept to measurement

Example: Multiculturalism and conflict

claim:

"Multicultural policy reduces ethnic conflict"

Multicultural policy:

Defining the concept:

"Forms of public recognition, support, or accommodation for ethnocultural minorities to maintain and express their distinct identities and practices."

Multicultural policy:

A multidimensional concept (for immigrant minorities):

  • Formal state commitment to MCP:
  • Multiculturalism in school curricula
  • Required diversity in media
  • Exemptions from dress codes
  • Permitting dual citizenship
  • Funding ethnic groups/organizations
  • Funding bilingual/mother-tongue instruction
  • Affirmative action for disadvantaged minorities

Multicultural policy:

Multicultural policy:

Does it matter which dimensions are present and absent?

Multicultural policy:

Country Commit Curric. Media Dress Code Dual Citizens Ethnic Funding Bilingual Ed Affirm. Action Total
Germany Kinda No No Kinda No Yes Kinda No 2.5
Greece No Kinda Kinda No Yes No No Kinda 2.5

Kinda = 0.5 Yes = 1

Multicultural policy:

Does it matter which dimensions are present and absent?

  • Different dimensions may have different effects
  • Should inform our theory, causal logics
  • Conversely, need to measure correct dimension of concept as it corresponds to theory

From Concepts to Measurement

Once we have concepts…

What do we do?

Identify variables

Propose measures

Concept to Measurement:

Concept

\(\xrightarrow{}\)

Variable

  • Something observable that captures/maps onto concept

\(\xrightarrow{}\)

Measure:

  • Procedure for observing the variable

Concept to Measurement:

Concept: Height (of a mountain)

Prominence of peak from the earth's surface

\(\xrightarrow{}\)

Variable: Vertical distance from top of peak to sea level

\(\xrightarrow{}\)

Measure:

Use difference in barometric pressure to calculate difference in elevation

Concept to Measurement:

Concept: Democracy

Regime in which government decided by contested elections and incumbents lose and leave office peacefully.

\(\xrightarrow{}\)

Variable: Occurrence of contested elections, turn-over in government

\(\xrightarrow{}\)

Measure:

  • Historical data on each countries electoral rules, their application, results of elections

Concept to Measurement:

Concept: Accountability

Legislators vote in a manner similar to (median) preferences of constituents

\(\xrightarrow{}\)

Variable: Difference between legislator behavior and constituent preferences

\(\xrightarrow{}\)

Measure:

  • Random survey of constituents on policy areas on which legislator voted. List of votes on these policies by the legislator.

Concept to Measurement:

Concept:

  • An abstract definition for characteristics of or types of phenomena, groups, or individuals.

Variable:

A measurable property of a phenomena, group, or individual that takes on different values (it varies). Maps onto a concept.

Measure:

A procedure for determining the value of a variable for specific cases based on observation.

Concept to Measurement:

Concepts and Variables vs. Measures:

Concepts and variables are general

  • Concept: Intensity of civil war
  • Variable: Casualties in a civil war
  • NOT: "intensity of civil war in Syria"

Let's try it:

Claim:

"Ethnic diversity in a country causes poor economic growth"

Causal Logic:

  • Ethnic diversity leads to group-based politics
  • This leads to fewer public goods (e.g. education), policies that enrich one group
  • This discourages investment and innovation
  • Weaker economic growth

Ethnic diversity:

  • concept?
  • variable?
  • measure?

Ethnic diversity: concept

  • Presence of many different ethnic groups?
  • Presence of many distinct and large ethnic groups?

Ethnic diversity: variable

Probability that any two random individuals belong to different ethnic groups

"Ethno-linguistic fractionalization": "ELF"

\[ELF = 1 - \sum_{i=1}^{n} s_i^2\]

Where \(s_i\) is fraction of population for group \(i \in \lbrace 1 \ldots n \rbrace\)

Ethnic diversity: measure

Measure size of ethnic groups using Atlas Narodov Mira:

Soviet anthropological text from 1964

  • lists ethnic groups in each country
  • lists the size of these groups

Anything wrong here?

Ethnic diversity: revisions

Concept:

  • Different dimensions of ethnicity (e.g. language, religion, race, caste, tribe)
  • Causal logic suggests political relevance matters (differences between Greek, Italian, German, Irish immigrant descendants in US are not relevant today)

Variable:

  • Country A: Group 1 50%, Group 2 50%
    • \(ELF = 0.5\)
  • Country B: Group 1 66.6%, Group 2 16.6%, Group 3 16.6%
    • \(ELF = 0.5\)

Ethnic diversity: revisions

Measure:

Soviet anthropologists:

  • Are group sizes from 1964 correct?
  • Distinct ethnic groups treated as the same, unified ethnic groups treated as different
  • Rwanda: Hutus and Tutsis coded as one ethnic group.

Ethnic diversity: revisions

Concept:

Presence of distinct and large politically relevant ethnic groups

Variable:

Probability that any two random individuals belong to different politically relevant ethnic groups

Measure:

  • Identify relevant groups in each country using histories, news, academic books
  • Calculate size of these groups using up-to-date census counts

Variables

Definition

variable : a measurable property of a phenomenon, group, person that can potentially take on different values.

  • Derived to capture a concept
  • Variation across cases or over time.

In practice:

Variables take on values for each specific case at a specific point in time

  • "Annual household income"
  • Case is a household, time is a year
  • Can vary between households in year
  • Can vary across years within a household

Annual Family Income:

This is a variable. What values does it take?

  • Numeric values
  • Range: 0 to ?
  • Range: -? to ? (if business losses included)

Variations:

  • Gross Annual Household Income
  • Annual after-tax family income

Annual Family Income:

Languages

Language spoken:

This is a variable: what values does it take?

  • Categories
  • No numeric range
  • One language is not "higher"/"lower" than another.

But what if we counted the fraction of people?

Another example: