---
title: "[WEEK 7 TITLE]"
subtitle: "[WEEK 7 SUBTITLE]"
date: last-modified
date-format: "[Updated ]MMM D, YYYY"
format: 
  revealjs:
    theme: brownslides.scss
    logo: images/pols1140_hex.png
    footer: "[COURSE CODE]"
    multiplex: false
    transition: fade
    slide-number: c
    incremental: true
    center: false
    menu: true
    scrollable: true
    highlight-style: github
    progress: true
    code-overflow: wrap
    chalkboard: true
    # include-after-body: title-slide.html
    title-slide-attributes:
      align: left
      data-background-image: images/pols1140_hex.png
      data-background-position: 90% 50%
      data-background-size: 40%
filters:
  - openlinksinnewpage
execute: 
  eval: true
  echo: true
  warning: false
  message: false
  cache: false
---


```{r}
#| label: init
#| echo: false
#| results: hide
#| warning: false 
#| message: false

library(tidyverse)
library(labelled)
library(haven)
library(DeclareDesign)
library(easystats)
library(texreg)

```

# {{< fa map-location>}} Wednesday {.inverse}

## Plan 

- Current Events/Piloting Survey

- Retrospective/Economic Voting

- A realist theory of democracy (Chapter 8)

- Evidence of the political Relevance of Group Identity (Chapter 9)

- Thursday: Pitfalls of group identity

## Group Project

```{r}
#| echo: false

library(tidyverse)

df <- haven::read_spss("surveys/wk07.sav") 


df %>% 
  mutate(
    Project = fct_rev(fct_inorder(as_factor(Q5))),
    `Namesake` = name.0
  ) -> df

df %>% 
  filter(share == 1) -> class_df


# df %>%
#   mutate(
#     # Turn numeric values into factor labels 
#     Reincarnation = forcats::as_factor(reincarnation),
#     # Order factor in decreasing frequency of levels
#     Reincarnation = forcats::fct_infreq(Reincarnation),
#     # Reverse order so levels are increasing in frequency
#     Reincarnation = forcats::fct_rev(Reincarnation),
#     # Rename explanations
#     Why = reincarnation_why
#   ) -> df
# 
# df %>% # Data
#   # Aesthetics
#   ggplot(aes(x = Reincarnation, 
#              fill = Reincarnation))+
#   # Geometry
#   geom_bar(stat = "count")+ # Statistic
#   ## Include levels of Reincarnation w/ no values
#   scale_x_discrete(drop=FALSE)+
#   # Don't include a legend
#   scale_fill_discrete(drop=FALSE, guide="none")+
#   # Flip x and y
#   coord_flip()+
#   # Remove lines
#   theme_classic() -> fig1
# 
# fig1
df %>% 
    ggplot(aes(Project, fill=Project))+
  geom_bar()+
  geom_text(stat='count', aes(label=..count..), hjust=-0.25)+
  coord_flip()+
  ylim(0,30)+
  scale_fill_brewer()+
  theme_minimal()+
  labs(
    y="Count",
    x="",
    fill = "",
    title = "What should we do?"
  )

```


## Attendance Survey



Please click [here](https://brown.co1.qualtrics.com/jfe/form/SV_8ABVBmBLJTiTch0) to register your attendance today.

## What's in a name?

```{r }
#| label: namesake
#| echo: false


DT::datatable(class_df %>% 
                select(Namesake),
               fillContainer = F,
              height = "90%",
              options = list(
                pageLength = 5
              )
              )
```

## Public opinion and Iran

::: panel-tabset

## Discuss

- What are some questions/expectations we might ask?

- What does a model like RAS lead us to expect?

- What might Achen & Bartels expect?

## Questions

- How does the public feel about the current conflict?

- How will attitudes change?

- Will we see a rally around the flag effect?

- What are the consequences for Trump's Approval/Midterm elections?

## CNN

![](images/07_groups/07_cnn.png)

## Reuters

![](images/07_groups/07_reuters.png)

## WaPo

![](images/07_groups/07_wapo_2.png)

:::

# {{<fa lightbulb>}} Retrospective Voting {.inverse}

## Overview

- Retrospective voting reflects an alternative response to problems raised by Converse (1964)


- Redefine the problem of citizen competence
    - Democracy doesn't need perfectly informed, ideal citizens
    - Just requires citizens to **select** competent leaders and **sanction** bad leaders

    
## Two Models of Retrospective Voting

- Leadership selection:
    - Select the most competent candidate
- Sanctioning
    - Punish "bad" candidates who fail to work on citizens behalf


:::{.fragment}
Both models depend on the quality of the information or signal citizens have
:::

## Leadership Selection {.smaller}

::::{.columns}

:::{.column width="50%"}
As the information environment becomes noisier, it becomes harder to select good leaders

:::

:::{.column width="45%"}

![](images/06_voting/a1.png)

:::
::::

## Leadership Sanctioning

::::{.columns}
:::{.column width="50%"}

As the information environment becomes noisier, it becomes 

- Harder for voters to sanction ineffective leaders

- Easier for leaders to shirk their duties


:::

:::{.column width="45%"}

![](images/06_voting/a2.png)


:::
::::



## Retrospective Voting {.smaller}

::::{.columns}
:::{.column width="50%"}
So what should people base their retrospective evaluations on?



> In  order  to ascertain whether  the incumbents  have performed  poorly  or well, citizens need only calculate the changes in their own welfare. (Fiorina 1981)


:::

:::{.columns width="45%"}
![](https://media.giphy.com/media/JIJs7LYcWVmwxBShXD/giphy.gif)
:::

::::

## Economic Voting

![](https://delta.creativecirclecdn.com/richardson/original/20240829-112012-c99-its-the-economy-stupid-james-carville-rent-home-own.jpg)


## Economic Voting (Fiorina 1978)

![](images/06_voting/f1.png)


## Aggregate Evidence of Retrospective Voting {.smaller}

![](https://fivethirtyeight.com/wp-content/uploads/2014/03/5794984500_e79a01b99e1.jpg?w=1150)

## Debates in Economic Voting {.smaller}

Broad consensus that economic factors matter, but lots of ongoing debates within the field of economic voting:


- Macro vs Micro | Sociotropic vs Egocentric | National vs Pocketbook

  - Do national or individual economic factors matter?



- Time horizons | Myopic voters

  - Do voters maintain a "running tally" of long term events or are they overly swayed by recent changes


- Negative vs positive shocks

  - Negativity bias: weight bad news more heavily than good?


- Mechanisms and moderators

  - Partisans biases in evaluations of the economy




## Achen and Bartel's Critique of Retrospective Voting {.smaller}

A&B's critique boils down to two claims:



1. Voters retrospective capabilities appear haphazard at best
    - Punish politicians for things that are out of their control (Blind Retrospectiion)
    - Ignoring policy failures they could address (e.g. Spanish Influenza)


2. Voter's Economic evaluations are:
    - Short sighted
    - Poor predicters of competence
    - Open to Manipulation



# {{<fa lightbulb>}} Blind Retrospection {.inverse}

## Review

Take a moment to review the arguments in Chapter 5. Specifically:

- Table 5.1 and Figure 5.1 and 5.3

- Table 5.2 and Figure 5.4


## Shark Attacks and Blind Retrospection

:::: {.columns}
::: {.column width="60%"}

![](images/06_voting/a3.png)

:::

::: {.column width="40%"}


```{ojs}
//| echo: false

viewof q1 = 
  Inputs.textarea({
    label: "",
    placeholder: "Write your interpretation here",
    width: "500"
  })

```


:::
::::


## Shark Attacks and Blind Retrospection

:::: {.columns}
::: {.column width="60%"}
![](images/06_voting/a4.png)

:::
::: {.column width="40%"}


```{ojs}
//| echo: false

viewof q2 = 
  Inputs.textarea({
    label: "",
    placeholder: "Write your interpretation here",
    width: "500"
  })

```


:::
::::

## Shark Attacks and Blind Retrospection

:::: {.columns}
::: {.column width="60%"}
![](images/06_voting/a5.png)

:::
::: {.column width="40%"}


```{ojs}
//| echo: false

viewof q3 = 
  Inputs.textarea({
    label: "",
    placeholder: "Write your interpretation here",
    width: "500"
  })

```


:::
::::

## Droughts and Blind Retrospection

:::: {.columns}
::: {.column width="60%"}
![](images/06_voting/a6.png)

:::
::: {.column width="40%"}


```{ojs}
//| echo: false

viewof q4 = 
  Inputs.textarea({
    label: "",
    placeholder: "Write your interpretation here",
    width: "500"
  })

```


:::
::::

## Droughts and Blind Retrospection

:::: {.columns}
::: {.column width="60%"}
![](images/06_voting/a7.png)

:::
::: {.column width="40%"}


```{ojs}
//| echo: false

viewof q5 = 
  Inputs.textarea({
    label: "",
    placeholder: "Write your interpretation here",
    width: "500"
  })

```


:::
::::

## Summary {.smaller}

In Chapter 5, Achen and Bartels present evidence that voters enage in [blind retrospection]{.blue}, punishing elected officials for events outside of their control.

- They show that Woodrow Wilson's vote totals in 1916 appear to be lower in beach counties during a summer when shark attacks in New Jersey were particularly salient

- They suggest this phenomena is more general, by showing how extreme weather (droughts and floods) negatively impacts incumbent vote share.

:::{.fragment}
But how robust are these results?
:::

## Do Shark Attacks Really Sway Elections

![](images/06_voting/f0.png)


## Do Shark Attacks Really Sway Elections

![](images/07_groups/f1.png)

## Did Shark Attacks Influence the 1912 Election

![](images/07_groups/f2.png)

## The Garden of Forking Paths

![](images/07_groups/f3.png)

## Summary

- The shark attack example is flashy and surprising



- [Fowler and Hall (2018)](https://www.journals.uchicago.edu/doi/full/10.1086/699244?casa_token=q3ZyXyUgLooAAAAA%3AjFKe0Bdo7l7yIhCtTu9PdJt_dQwQjiQFe1gyngtDDPbNrnzRWGrAGBUQspzqrSePd16uQRUw49V6) offer compelling critiques of this particular finding
  - Achen and Bartels (2018) [response](https://www.journals.uchicago.edu/doi/full/10.1086/699245?casa_token=AUiqvr8tSbkAAAAA%3Aw6o2lX0jGU-FMKNvT-a9-R8oKYoYTVbCxdKbnid6HE_lr8Hyb0i20pWneULUjrIjBeFtBr9MlkY8)
- But still others find evidence of irrelevant events influencing electoral behavior:
    - [Healy et al. 2010](https://www.pnas.org/content/107/29/12804.short) analyze football games ( [Fowler and Montange](https://www.pnas.org/content/112/45/13800.short) offer a similar critique)
    - [Busby and Druckman (2017)](https://www.journals.uchicago.edu/doi/pdfplus/10.1086/688585) offer experimental evidence using Ohio State - Oregon game

- What can we conclude?

# {{<fa lightbulb>}} Economic Voting {.inverse}

## Achen and Bartel's Critique of Economic Voting

If voters are rational retrospective evaluators, we should observe:

- Long time horizons (running tally)

- Effects tied to actual executive responsibility (No shark attacks/droughts)

- Strong relationship between performance and re-election

If A&B are right, we should observe:

- Myopia (only recent quarters matter)

- Weak relationship between performance and competence

- Effects consistent with manipulation

Now let's walk through some of the evidence.

## Table 6.1: Voters are Short sighted and influenced only by recent economic events

What Is Being Estimated?

- Dependent variable: Incumbent vote share

- Key predictors: Quarterly economic growth

- Question: Do voters respond to the full term… or just the last year?

- If voters use a "running tally":
  - growth over entire term should matter

- If voters are myopic: only the final year (or final quarters) should matter


## Table 6.1: Voters are Short sighted and influenced only by recent economic events

![](images/06_voting/a8.png)
## Interpretation

- Which quarters have statistically significant effects?

- How large are the effects?

- Do early-term economic shocks matter?

Key takeaway:

**Only growth in the last year meaningfully predicts incumbent vote share.**

- Voters appear to overweight recent economic conditions.

- That creates incentives for short-term manipulation.

## What Would “Competence Selection” Look Like?

If voters select competent leaders:

Strong economic performance under Party A

- Party A should win repeatedly
- Future performance should be predictable

If voters do poorly at selection:

- Re-election does not strongly predict future performance
- Electoral outcomes may not track underlying competence

## Table 6.4: Weak relationship between performance and competence

![](images/06_voting/a9.png)





## Table 6.5: Weak relationship between performance and competence

![](images/06_voting/a10.png)

## What Do These Results Suggest?

- Re-electing incumbents does not reliably improve economic outcomes (Table 6.4 & 6.5)

- Implication:

  - Even if voters respond to economic conditions, they may not be effectively selecting competent leaders.
  - Retrospection $\neq$ competence selection.

## What Would Manipulation Look Like?

If incumbents can stimulate the economy before elections we should observe:

- Economic growth spikes before elections

- Weak growth afterward

This creates:

- A moral hazard problem

- Incentives for short-term economic distortion

## Figure 6.3: Voters May be Subject to Manipulation/Electoral Business Cycles

![](images/06_voting/a11.png)
## Poltical Business Cycle:

- Economic growth appears to increase before elections.

- Suggests potential electoral timing.

- If voters focus heavily on recent growth, incumbents may game the system.

- Democratic accountability becomes vulnerable to short-term engineering.

## Summary {.smaller}

- Theories of Retrospective Voting seek to offer an alternative account of democratic accountability that more realistically reflects the abilities of average citizens

- Rather than assuming coherent beliefs, complete knowledge, RV asks citizens to select good leaders and sanction bad leaders using assessments of their welfare as indicator of competence

- Critics of RV contend that retrospective evaluations are:

  - Haphazard: Citizens punish elected officials for things they have know control over
  - Myopic: Only recent economic evaluations seem to matter, can be manipulated/biased.




# {{<fa lightbulb>}} A Realist Theory of Democracy (Chapter 8) {.inverse}


## Competing Theories of Democracy {.smaller}

**Populist Folk Theory**
- Voters hold coherent policy preferences
- Elections aggregate those preferences

**Rational Retrospective Theory**
- Voters evaluate performance
- Elections reward competence


## How This Section Fits the Course

Achen & Bartels argue:

- Voters are not primarily policy maximizers
- Nor especially competent retrospective evaluators
- Instead, **partisan loyalties flow from social identities**

If that’s true, we should observe:

1. Voting shifts when group attachments shift  
2. Party identification changing more slowly than votes  
3. Attitudes following partisanship (or other core identities)  


## A brief introduction to Social Identities

- Social identities are going to play a significant role in A&B's realist theory of democracy

- Social identity is a person’s sense of who they are based on their group membership


## Social Identity

- Take a moment to write down all the social identities you possess. 

## My Social Identities

- White
- Male
- Sports fan (I agree, go Cavs)
- Catholic
- Democrat
- Married
- Irish and Italian
- Ithacan
- Academic
- Culture vulture

## Social Identity




- Now cross off the least important identity.



- Now cross off the next least important identity...

## Social Identity {.smaller}

- Social identities vary in 
    - Origin (voluntary vs involuntary)
    - Strength
    - Salience
    - Effect
- People tend to 
    - Possess multiple identities
    - Understand identities relationally (in-groups vs out groups)
    - Exaggerate differences between groups and emphasize similarities within groups


## What Would Identity-Based Politics Look Like?

If identity drives politics, we should observe:

- Group loyalties preceding policy positions
- Party attachments shifting with group realignment
- Stable social identities predicting partisan change
- Attitudes adjusting to match partisan identity

# It feels like we're thinking (Chapter 9){.inverse}

## Evidence of Group Identity (Chapter 9) {.smaller}

Chapter 9 presents evidence of the important of "identities" to understanding political behavior using three types of evidence:

- Historical analysis of Catholic voting behavior

- Time series cross sectional survey analysis of the partisan identity and policy beliefs of White Southerners

- Panel survey analysis of abortion attitudes and partisanship

## Catholic Voting: What Should We See?

If policy preferences dominate:

- Kennedy’s Catholic identity should not matter much.

If identity matters:

- Catholic voters should temporarily rally behind Kennedy.
- But long-term voting patterns should revert once the identity shock fades.

## Figure 9.1

![](images/07_groups/a1.png)

## Historical analysis of Catholic voting behavior {.smaller}

- What are the key takeaways from Figure 9.1?
- What evidence supports the claims:
    - "[T]he impact of Kennedy's candidacy on Catholic support for the Democratic Party was temporary" (p. 245)
    -  "It is hard to imagine a clear demonstration of the political impact of group attachments and the trade offs among them" (p. 244)

- Why was "the social significance of a Catholic presidential candidacy ... no longer sufficient to produce substantial deviations from accustomed voting behavior" (p. 246) 

## The Realignment of Partisan Identities in the South {.smaller}

Achen and Bartels present an alternative interpretation of realignment in the south emphasizing the role of social identities over standard accounts that emphasized partisan policies using the following evidence:

- Analyzing trends in PID and Voting overtime (Fig 9.1) and by age cohort (Fig 9.2)

- Analyzing trends in PID by policy position (Fig 9.4, 9.5)

- Regression analysis predicting PID with feelings toward Southerners over time (Table 9.1)


## Southern Realignment: Competing Explanations

Standard Account:
- Policy evolution (especially race) drove partisan change.

Achen & Bartels' Claim:
- Social identity as “Southerner” reshaped partisan attachments.
- Voting moved first.
- Party identification adjusted more slowly.


## The Realignment of Partisan Identities in the South {.smaller}

Achen and Bartels present an alternative interpretation of realignment in the south emphasizing the role of social identities over standard accounts that emphasized partisan policies using the following evidence:

- Analyzing trends in PID and Voting overtime (Fig 9.1) and by age cohort (Fig 9.2)

- Analyzing trends in PID by policy position (Fig 9.4, 9.5)

- Regression analysis predicting PID with feelings toward Southerners over time (Table 9.1)


## White Southerners went from solid Democrats to consistent Republicans

::: panel-tabset

## Fig 9.2

![](images/07_groups/a2.png)

## Summary

- Voting shifts faster than partisan identification.
- PID lags behind electoral behavior.

Implication:

- Identity-based attachments adjust slowly.
- Elections move before self-identification catches up.

:::

## If Issue Evolution Drove Realignment...

We would expect:

- Strong and consistent relationships between racial policy views and PID
- Clear sorting by policy position
- Identity feelings to be secondary

## Issue evolution alone doesn't explain partisan realignment

What would issue evolution predict?

![](images/07_groups/a3.png)



## Regression Analysis

What Is Being Estimated?

Dependent variable:
- Democratic identification **among White Southerners**

Key predictor:
- Feelings toward "Southerners"

Question:
Does warmth toward the in-group predict partisanship?

## Regression Analysis


![](images/07_groups/a4.png)



## Regression Analysis{.smaller}

::::{.columns}
:::{.column width="45%"}

> The  results  indicate  that  the  Democratic partisan advantage in 1964 among white southerners who expressed neutral  attitudes  toward  “southerners”  was  almost  25  percentage points,  while  the  corresponding  advantage  among  those  who expressed very warm feelings toward “southerners” was more than twice as large, almost  55 percentage points

:::
:::{.column width="55%"}

![](images/07_groups/a4a.png)

:::
::::


## What are these models showing{.smaller}

::::{.columns}
:::{.column width="55%"}

```{r}
#| echo: false

plot_df <- data.frame(
  x = seq(-1,1,length.out=11),
  y = seq(-50,50,length.out=11)
)
plot_df %>% ggplot(aes(x=x,y=y))+
  theme_classic(base_size = 12)+
  labs(
    y = "Democratic Advantage in Identification",
    x = "Feelings toward Southerners",
    title = "1964"
  ) -> p1964

plot_df %>% ggplot(aes(x=x,y=y))+
  theme_classic(base_size = 12)+
  labs(
    y = "Percent Identifying as Democrat",
    x = "Feelings toward Southerners",
    title = "2008"
  ) -> p2008

```

```{r}
#| echo: false
#| cache: false

p1964
```

```{r}
#| echo: false
#| cache: false


p2008
```

:::
:::{.column width="45%"}

![](images/07_groups/a4a.png)

:::
::::





## Regression Analysis 

::::{.columns}
:::{.column width="45%"}

Was  southern  identity  really  the  basis  of  these  very  different responses  to the political  events of the long southern  realignment era?

:::
:::{.column width="55%"}


![](images/07_groups/a4a.png)

:::
::::

## Southern Identity and Partisan Identification {.smaller}

```{r}
#| echo: false

library(anesr)
library(labelled)
library(tidyverse)
library(haven)
data("timeseries_cum")

df <- timeseries_cum
rm(timeseries_cum)

df %>% 
  mutate(
    year = VCF0004,
    linear_trend = year - 1964,
    is_2008 = ifelse(year == 2008, 1, 0),
    is_white = ifelse(VCF0105a == 1, 1, 0),
    is_southern = ifelse(VCF0112 == 3, 1,0),
    pid_7pt = VCF0301,
    dem_adv = case_when(
      pid_7pt < 4 ~ 1,
      pid_7pt == 4 ~ 0,
      pid_7pt > 4 ~ -1,
    ),
    is_dem = ifelse(pid_7pt < 4, 1, 0),
    is_ind = ifelse(pid_7pt == 4, 1, 0),
    is_rep = ifelse(pid_7pt > 4, 1, 0),
    pid_3cat = case_when(
      pid_7pt < 4 ~ "Democrat",
      pid_7pt == 4 ~ "Independent",
      pid_7pt > 4 ~ "Republican",
    ) %>%  factor(),
    ft_southerners = VCF0208,
    ft_blacks = VCF0206,
    ft_whites = VCF0207,
    ft_southerners_ab = (ft_southerners - 50)/50,
    ft_blacks_ab = (ft_blacks - 50)/50,
    ft_whites_ab = (ft_whites - 50)/50,
    aid_to_blacks = ifelse(VCF0830 == 9, NA, (VCF0830-8)*-1)
  ) -> df
```

```{r}
#| echo: false

df %>% 
  filter(!is.na(ft_southerners)) %>%
  filter(is_white == 1, is_southern ==1) %>% 
  ggplot(aes(ft_southerners, is_dem))+
  geom_smooth(method = "lm")+
  labs(y = "Percent Democrat", x = "Feelings toward Southerners",
       title = "Democratic Identification of White Southerners")+
  facet_wrap(~year, ncol =4 ) -> p1

df %>% 
  filter(!is.na(ft_whites)) %>%
  filter(year <= 1982) %>% 
  filter(is_white == 1, is_southern ==1) %>% 
  ggplot(aes(ft_whites, is_dem))+
  geom_smooth(method = "lm")+
  labs(y = "Percent Democrat", x = "Feelings toward Whites",
       title = "Democratic Identification of White Southerners")+
  facet_wrap(~year, ncol =4 ) -> p2a

df %>% 
  filter(!is.na(ft_whites)) %>%
  filter(year > 1982) %>% 
  filter(is_white == 1, is_southern ==1) %>% 
  ggplot(aes(ft_whites, is_dem))+
  geom_smooth(method = "lm")+
  labs(y = "Percent Democrat", x = "Feelings toward Whites",
       title = "Democratic Identification of White Southerners")+
  facet_wrap(~year, ncol =4 ) -> p2b

df %>% 
  filter(!is.na(ft_blacks)) %>%
  filter(year <= 1982) %>% 
  filter(is_white == 1, is_southern ==1) %>% 
  ggplot(aes(ft_blacks, is_dem))+
  geom_smooth(method = "lm")+
  labs(y = "Percent Democrat", x = "Feelings toward Blacks",
       title = "Democratic Identification of White Southerners")+
  facet_wrap(~year, ncol =4 ) -> p3a

df %>% 
  filter(!is.na(ft_blacks)) %>%
  filter(year > 1982) %>% 
  filter(is_white == 1, is_southern ==1) %>% 
  ggplot(aes(ft_blacks, is_dem))+
  geom_smooth(method = "lm")+
  labs(y = "Percent Democrat", x = "Feelings toward Blacks",
       title = "Democratic Identification of White Southerners")+
  facet_wrap(~year, ncol =4 ) -> p3b

df %>% 
  filter(!is.na(aid_to_blacks)) %>%
  filter(year < 1989) %>% 
  filter(is_white == 1, is_southern ==1) %>% 
  ggplot(aes(aid_to_blacks, is_dem))+
  geom_smooth(method = "lm")+
  labs(y = "Percent Democrat", 
       x = "Support for Govt Aid to Blacks",
       title = "Democratic Identification of White Southerners"
       )+
  facet_wrap(~year, ncol =4 ) -> p4a

df %>% 
  filter(!is.na(aid_to_blacks)) %>%
  filter(year > 1989) %>% 
  filter(is_white == 1, is_southern ==1) %>% 
  ggplot(aes(aid_to_blacks, is_dem))+
  geom_smooth(method = "lm")+
  labs(y = "Percent Democrat", 
       x = "Support for Govt Aid to Blacks",
       title = "Democratic Identification of White Southerners"
       )+
  facet_wrap(~year, ncol =4 ) -> p4b


```


::::{.panel-tabset}

## Overview

I find the model(s) in Table 9.1 a little confusing.

In the following tabs, I've plot some linear trends in Democratic Identification by year for:

- Feelings toward Southerners
- Feelings toward Whites
- Feelings toward Blacks
- Govt aid to Blacks

## FT: Southerners

```{r}
#| echo: false


p1
```


## Whites

```{r}
#| echo: false


p2a
p2b
```


## Blacks

```{r}
#| echo: false


p3a
p3b
```


## Govt Aid to Blacks

```{r}
#| echo: false


p4a
p4b
```


::::

## Do the Raw Patterns Match A&B Regression Story?{.smaller}


- In 1964, there is little clear relationship between affect toward white Southerners and Democratic ID.
- The slope varies substantially across years.
- Some relationships are weak, noisy, or unstable.
- Policy attitudes (e.g., aid to Blacks) sometimes show similar patterns.

This suggests that:

The regression interpretation requires modeling assumptions and selective emphasis.

The identity story is plausible -- but it is not obvious in the raw data.

## PID, Gender and Abortion

Two possibilities:

1. Attitudes -> Party (policy drives identity)
2. Party ->  Attitudes (identity reshapes policy views)

Panel data allow us to test these claims (and test them separately for men and women)

## When Party and Abortion Conflict: What Moves?

Two conditioning exercises:

1. Among 1982 Republicans:
   Does abortion position predict party defection?

2. Among 1982 pro-life citizens:
   Does party predict abortion conversion?



## PID, Gender and Abortion

![](images/07_groups/a5.png)

## PID, Gender and Abortion

![](images/07_groups/a6.png)

## PID, Gender and Abortion

Key result:

- Women $\to$ more likely to change party  
- Men $\to$ more likely to change abortion attitudes

Interpretation:

The identity that is more central (gender for women, party for men) is less likely to move.

# Pitfalls of Group Identity (Chapter 10){.inverse}

## Pitfalls of Group Identity (Chapter 10)

Achen and Bartels conclude by considering the role of partisan identities in politics, and look at:

- Partisan misperceptions of party positions
- Partisan misperceptions of objective facts
- The impact of scandals on unrelated partisan policies

## What Would Identity-Based Perception Look Like?

If identity shapes cognition:

- People should perceive their party as closer to their own views.
- Even when objective distances are equal.

## People percieve parties as closer to the their own positions

![](images/07_groups/a7.png)

## Interpretation

- Partisans compress distance between themselves and their party.
- They exaggerate distance from the opposing party.

- Identity influences perception — not just preference.

## If Identity Shapes Belief Formation...

We should observe:

- Partisan bias in factual beliefs
- Persistence of misperceptions even among the politically informed


## Partisanship increases misperceptions of objective facts

![](images/07_groups/a8.png)

## Misperceptions remain common among even as political information increases 

![](images/07_groups/a9.png)

## Interpretation

- Partisanship predicts factual misperceptions.
- Information does not eliminate identity-based distortions.

- Implication:
  - Democratic reasoning is filtered through group attachment.

## What Would Identity-Driven Reactions to Scandals Look Like?

If partisanship structures evaluation:

- Scandals should spill over to unrelated policy attitudes.
- In-group members defend; out-group members generalize negativity.

## The impact of scandals on unrelated partisan policies

![](images/07_groups/a10.png)

## The impact of scandals on unrelated partisan policies

![](images/07_groups/a11.png)

## Interpretation

- Scandal effects extend beyond the scandal domain.
- Partisan identity structures policy reactions.

- Identity organizes political reasoning broadly.

## What Have We Learned?

Across cases:

- Catholic voting
- Southern realignment
- Abortion attitudes
- Perceptual biases
- Scandal spillovers

The pattern is consistent:

Social identity shapes:

- Voting
- Issue attitudes
- Perceptions
- Interpretation of events

Is this a problem for democracy?



## Next week: Huddy's Critique of Achen and Bartels

Huddy offers several critiques of Democracy for Realists:

- A&B's discussion of partisanship and abortion
- The nature of group politics
- The relevance and rationality of group interests
- The consequences of multiple identities of varying strengths

Why does she make these critiques? How compelling are they?

Let's pick these questions up at the start of class on Tuesday