Weaver VIZ@UBC Talk

---

#Publicity and the Delegitimation of Lynching

Michael Weaver

The University of British Columbia

February 26, 2019

---

##Societies accept or tolerate some forms of violence while rejecting others.

---

###Domestic violence

###Dueling

###Police use of force

##How does violence become publicly unacceptable?

---

##Lynching in the United States

---
template:inverse

###Rebecca Felton, suffragette:

---
layout:false
class: left, middle

> "**I led the mob** which lynched Nelse Patton and **I'm proud of it**. I directed every movement of the mob. **I wanted him lynched**. I saw his body
dangling from a tree this morning, and I am glad of it. **I aroused the mob and
directed them to storm the jail**. I had my revolver but did not use it. I gave it to a deputy sheriff and **told him to shoot** Patton and shoot to kill."
>> ### William V. Sullivan, former United States Senator, Mississippi

---

template:inverse

### South Carolina Governor Coleman Blease:

---

template:inverse

---

template:inverse

---

template:inverse

---
template:inverse

*New York Age* 1/17/1931

---

template:inverse

*The Chillicothe Constitution-Tribune* 1/15/1931

---
template:inverse

## How did this transformation take place?

---

---
template:inverse

##The argument

---

- (Black 1983, Fiske and Rai 2014)
  
### Which violence gets to be "legitimate" in a context/locally?
]

---

##### When perpetrators are powerful.

#### 1. Able to frame violence
#### 2. Critics coerced into silence
#### 3. Challengers face reputational costs
#### 4. May be self-reinforcing dynamic
  - Determine both "facts" and norms

### How can this dynamic be challenged?
]

---
template:inverse

## Publicity Shocks

---

*Geographic scope of audience*

When it expands:
  
  * New audiences not share local norms on violence
  * Critics not silenced

### Inclusivity

*Inclusion of different voices in public debate*

When it expands:

* Victims and allies no longer silent
  * 'Facts' and justifications around violence contested 
  * **Enabled** by *reach*

]

---

### Expanded publicity can

#### *Challenge dominance of pro-violence discourse*

### and under [some conditions](#conditions):

#### *Increase reputational costs to violence*

#### *Lead to local efforts to stop violence*

]

---
template:inverse

##Lynching and Publicity

### Local critics silenced, but...

---

.right-column[
###1. Increase in publicity
* Technology + Economic/Political Integration `$\rightarrow$` greater reach
* Reach + activists/black press `$\rightarrow$` inclusion of black voices

]

---

template:inverse

---

template:inverse

"The negroes must have detectives who can go and find out the facts about each lynching and publish them side by side with the versions printed in the Southern newspapers"

---

template:inverse

---

.right-column[
###1. Increase in publicity
* Technology + Economic/Political Integration `$\rightarrow$` greater reach
* Reach + activists/black press `$\rightarrow$` inclusion of black voices

###2. Publicity breeds criticism and scandal
]

---
template:inverse

Threatened by Wells's publicity campaign

---
class: left, middle

> ##### "The Chattanooga News does not lift its voice in behalf of Sheriff Shipp, his deputies or any member of the mob that lynched Johnson. Its **voice is merely raised in behalf of this community.** It submits that even so high and mighty a person as the Attorney General of the United States **has not just cause to hold this community up in contempt of the world** and in effect **publish it abroad as lawless** and firmly set **against the enforcement of law.**"

>> "An Official in Contempt of Chattanooga," *The Chattanooga News*

---

.right-column[
###1. Increase in publicity
* Technology + Economic/Political Integration `$\rightarrow$` greater reach
* Reach + activists/black press `$\rightarrow$` inclusion of black voices

###2. Publicity breeds criticism and scandal

###3. Bad publicity turns Southern elites against lynching

]

---

template:inverse

"our lynching reputation ... is holding the state back in its development."

---

.right-column[
###1. Increase in publicity
* Technology + Economic/Political Integration `$\rightarrow$` greater reach
* Reach + activists/black press `$\rightarrow$` inclusion of black voices

###2. Publicity breeds criticism and scandal

###3. Bad publicity turns Southern elites against lynching

###4. With opposition of local elites, lynching declines
]

---

template:inverse

---

template:inverse

## Testing the Argument

---

### Emergence of a national public sphere...

- ##### many causes
- ##### monotonic growth

]

---

### ***Local* exposure** to publicity grew **uneven**ly

##### Differential access to communication technology

### `$\xrightarrow{implies}$`

##### Differential exposure to new publicity

### What are observable implications at **subnational** level?

]

---

##### Access determines reach

##### The **probability** that a **lynching is reported:**

##### (1) **increases** as travel times **decrease** between the lynching and the paper.

##### (2) **increases** when the lynching occurred in an area more **central in communication and transportation networks**.

]

---

##### Greater reach yields criticism, fewer lynchings.

##### **Coverage** of lynching is **more critical**:

##### (3) as the **distance** from the lynching to the newspaper **increases**.

##### Lynching **declines faster**:

##### (4) in places with **more exposure** to national public sphere

]

---

template:inverse

##The data

---

template:inverse

## "Reach" of Publicity

---

template:inverse

---

name:newspaper_data

###"Big data" meets history

* Digital newspapers archives

* More than 3,000 different papers
  * **National**, though **uneven** coverage
  * Big-city dailies, small town weeklies
  
* 9 million+ issues between [1880 and 1940](#digitization_time)

* Searchable content
  * 1.2 million articles mention "lynching"
  * Words/phrases for [lynching discourse](#discourse_measure)
]

---

Digitized Newspaper Issues by State

---

template:inverse

## Communication Networks

---

class:center, middle

---

###Railroad networks

Yearly data from 1880 and 1900
  * Period of rapid growth (more than doubles)
  * Network centrality of counties
  * Travel time between counties
  * Proxy for telegraph network
]

---

Collecting data on railroad/telegraph network

- US, Canada, Mexico
- yearly
- specifics of service provided

]

---

class:center, middle

---

template:inverse

## Lynching Events

---

###Lynching events
  
* 1880 to 1930s, ~ 4000 *reported* events
* Sources:
  * Historians, Sociologists
  * NAACP
  * Chicago Tribune

]

---

template:inverse

##Analysis

---

### Four implications of the argument

##### Access determines reach

##### The **probability** that a **lynching is reported:**

##### ![check][] (1) **increases** as travel times **decrease** between the lynching and the paper.

##### ![check][] (2) **increases** when the lynching occurred in an area more **central in communication and transportation networks**.

]

[check]:./check-mark-3-24.png

---

Probability of lynching coverage by railroad travel time (Residual)

---
class: center, middle

### Effects of Betweenness Centrality on Pr(Mention)

---

##### Greater reach yields criticism, fewer lynchings.

##### **Coverage** of lynching is **more critical**:

##### ![check][] (3) as the **distance** from the lynching to the newspaper **increases**.

##### Lynching **declines faster**:

##### (4) in places with **more exposure** to national public sphere

]

[check]:./check-mark-3-24.png

---

## Criticism of Lynching across Distance

---

##### Greater reach yields criticism, fewer lynchings.

##### **Coverage** of lynching is **more critical**:

##### ![check][] (3) as the **distance** from the lynching to the newspaper **increases**.

##### Lynching **declines faster**:

##### ![check][] (4) in places with **more exposure** to national public sphere

]

[check]:./check-mark-3-24.png

---

### [Access](#access) to daily newspaper circulation inhibits lynching

---

---
template:inverse

## Conclusion

---

###Did **activists** change the frame?

###Did external **criticism** stop lynching?
  
###Can changes in publicity **legitimize** violence?
]

---
template:inverse
class: center, middle

---
template:inverse

---

### Use of Visualizaiton

#### 1. Data exploration/management
#### 2. Preliminary Analyses
#### 3. Communicating

]

---

### Does publicity *constrain state violence* ?

Unclear at this point, but research

### Helps understand *legitimation of state violence*

]

---

### Legacy of Anti-Lynching Movement

Publicity effective, but should not be overly sanguine

]

---

template:inverse

#Thank you

---

---
template:inverse

#Extra slides

---

name:extras

#### Argument
- [conditions for effects](#effect_conditions)
- [newspaper partisanship](#partisanship)
- [discourse shifts over time](#discourse_shift_time)

#### Data
- [digitization rates by time](#digitization_time)
- [coverage validation (window)](#coverage_window)
- [coverage validation (accuracy)](#coverage_validation)
- [discourse keywords](#keywords)
- [keyword validation](#keyword_validation)
- [discourse measure](#discourse_measure)
- [media access measure](#access)

#### Analysis

- [model specification](#equations)
- [coverage over distance](#coverage_over_distance)
- [coverage results (aggregate)](#aggregate_result)
- [effects over distance](#distance_result)
- [travel time table](#travel_time_table)
- [limits](#limits)
- [extensions](#railroad_extension)

---

template:inverse

## Extras: Argument and Evidence

---

name:effect_conditions

1. **News of violence reaches new audience**
2. **New audience is interested**
3. **New audience is critical/persuaded by victims**
4. **News of criticism reaches locality**
5. Locals value relationship/reputation with audience

### ... increasing publicity can

1. **Challenge local hegemony of pro-violence discourses**

]

---

### ... increasing publicity can

1. Challenge local hegemony of pro-violence discourses
2. **Increase reputational costs to endorse or tacitly approve violence**

]

---

### ... increasing publicity can

1. Challenge local hegemony of pro-violence discourses
2. Increase reputational costs to endorse or tacitly approve violence
3. **Increase costs of permitting violence to occur**

]

---

Fraction of daily Republican papers Republican by region

[extras](#extras)

---
class: center, middle

Fraction of all papers Republican (south)

[extras](#extras)

---
class: center, middle

Average Criticism over Time

[extras](#extras)

---

template:inverse

## Extras: Data

---

Digitized Newspaper Issues by Year

[back](#newspaper_data)
[extras](#extras)

---

name:coverage_window

Fraction of event-related articles from time of event

[extras](#extras)

---
class: center, middle
name:coverage_validation

TPR/PPV of article-event matching over time from event

[extras](#extras)

---

name:keywords

[extras](#extras)

---

[extras](#extras)

---
name:keyword_validation

[extras](#extras)

---

name:discourse_measure

### Discourse scaling

`$$\begin{equation}\label{eq:discourse_scale}
Criticism_k =  \left( \sum_{i=1}^{n_a} AntiLynchingWord_{ik} \right) - \left(\sum_{j=1}^{n_p} ProLynchingWord_{jk} \right)
\end{equation}$$`

[extras](#extras)
[back](#newspaper_data)

---

name:access

### Media Access

Media access in **county** `$i$` is the average of:

* `$NewspaperCirculation_j$` (total circulation of daily newspapers in a county `$j$`)

* for **all other counties** `$j$`; `$i \neq j$`;

* Weighted by the inverse of `$\tau_{ij}$` (railroad travel time between `$i$` and `$j$`)

`$$MediaAccess_i = \sum\limits_{i \neq j}^{j} NewspaperCirculation_j * \tau_{ij}^{-1}$$`

[extras](#extras)

[return](#access_table)

---

template:inverse

## Extras: Analysis

---

name:equations

##Equations

### Coverage Model

`$$\begin{equation}\label{eq:coverage_dyads}
\begin{split}
Y_{ijt} =& \alpha_{year} + \alpha_{i} + \alpha_{j} + \boldsymbol{\delta}\mathbf{Centrality_{it}} + \\ &\boldsymbol{\beta}\mathbf{Distance_{ijt}} + 
\boldsymbol{\gamma}\mathbf{X_{it}} + 
\boldsymbol{\lambda}\mathbf{W_{jt}} + \\ 
&\varepsilon_{i} + \varepsilon_{j}
\end{split}
\end{equation}$$`

### Lynching model
`$$\begin{equation}\label{eq:lynching}
\begin{split}
Lynchings_{it} = \alpha_i + \alpha_t + \beta_1 Access_{it} + \mathbf{X_{it}\beta} +  
 \varepsilon_{county_i}
\end{split}
\end{equation}$$`

[extras](#extras)

---
class: center, middle
name: coverage_over_distance

Probability of lynching mention by distance

---

Centrality Results (aggregated to the event, fraction of papers)

[extras](#extras)

---

Effect across Distance

[extras](#extras)

---

name:travel_time_table

---

name:limits

.right-column[
###Railroads Endogenous?
* but network attributes not locally determined
* conditioned on local railroad construction

###Alternative interpretations
* Importance, not information
  * Condition on population, economic output
  * Access to circulation, not to population
* Reduced or hidden lynchings?
* "Supply" or "demand"?

###Text
* Broad but limited depth
* Keywords are drastic simplification
  * [Validation](#keyword_validation): manual coding
 
 ]

---

name:railroad_extension

]

###Railroad networks

* Railroad stations for each year between 1880 and 1910
  * Scanned station lists for each year
  * Digitizing text, geocoding
  * Proximity to rail stations

### Telegraph Offices
* Telegraph offices
  * Scanned lists
  * Digitizing text, geocoding

]

---
class: center, middle

[extras](#extras)

---

template:inverse
<iframe width='100%' height='95%' frameborder='0' src='https://mdweaver.github.io/station_map' allowfullscreen webkitallowfullscreen mozallowfullscreen oallowfullscreen msallowfullscreen></iframe>

[extras](#extras)