ties that bind - cpb-us-e1.wpmucdn.com€¦ · ties that bind social heterogeneity and cohesive...

Cities on the HillCity Delegations

Conclusion

Ties That BindSocial Heterogeneity and Cohesive City Delegations

Tom Ogorzalek

Northwestern Political Science and Urban [email protected]

Tom Ogorzalek Ties That Bind


Conclusion

Social-Political GeographyA City Interest?

Synthesize Urban Politics,Race, and APD

• Liberalism as city-friendlypolicy

• Development ofurban-rural partisan divide

• City institutions innational politics

Today

• City delegations

• Local institutions ofhorizontal integration(IHIs): Support cohesiondespite heterogeneity

THE CITIES ON THE HILL

How Urban Institutions Transformed National Politics

THOMAS OGORZALEK



Conclusion


How do cities fit into American politics?

Important, distinctive communities

• Very challenging local governance

• No formal representation at national level

Urban-rural divide

• Chronic, but not always partisan

• What holds the hyper-diverse “Blue” side together?

• Across and within cities



Conclusion


NYTimes


http://www.nytimes.com/projects/census/2010/map.html


Conclusion


NYTimes, CityLab


http://www.nytimes.com/projects/census/2010/map.html

https://www.citylab.com/equity/2017/04/is-this-the-ultimate-2016-presidential-election-map/521622/


Conclusion


“The Ungovernable City”

NYtimes.com, US Census



Conclusion


“Hyperpluralism”

NYtimes.com, US Census



Conclusion


Proportion African American,

Chicago Census Tracts, 1940

0.0 - 0.010

0.011 - 0.050

0.051 - 0.10

0.11 - 0.25

0.26 - 0.50

0.51 - 1.0

Figure: Chicago, Percent blackin census tracts, 1940 Census.African Americans were heavily con-centrated within cities like Chicago.

0

3

6

7

2

10

4

1

9

8

5

1


Chicago Congressional Districts,

76th Congress (1939-1941)

0.00 - 0.05

0.06 - 0.10

0.11 - 0.25

0.26 - 0.50

0.51 - 0.74

Figure: Chicago, Percent black incongressional districts, 1940 Res-idential segregation led to politicalsegregation: few congressmen hadmany black constituents.



Conclusion




Conclusion


A paradox of city government

Urbanicity generates distinctive preferences and governance needs

• Density, Heterogeneity, Size

• Economic management: redistribution, regulation, publicgoods

• Culture: flexibility, adaptation, impersonality

Cities face distinctive challenges

• City limits: Formal and informal constraints to policymaking

• “Diversity problems” related to heterogeneous constituencies



Conclusion


Establishing political order

Cities develop institutions to overcome “diversity problems”

• Machines (Erie, etc)

• Monopolies (Trounstine)

• Urban regime (Stone)

• Pluralism (Dahl)

These are forms of horizontal integration across an uneven,heterogeneous preference space (IHIs).



Conclusion

Theory and MechanismsCohesionCivil Rights

Cities on the Hill: City Delegations

• Some cities send several representatives

• Pursue city-friendly policies in state and nation

• Urban reps act strategically as blocs (Weir et al)

• Cohesion a good strategy for success on “urban” issues(Burns et al)

• MCs represent city as well as district

• Transmit local solutions for political order to higher levels.



Conclusion


“Those of us who have served in the state legislatureknow of the power that is more than numerical that goeswith the organization of the big cities. . . [MCs fromcities] represent, not, in fact, their separate districts, butthe whole city, representatives who are responsible to thesame public opinion, and, in fact represent but onecombined interest of the citizens of the city.”

Rep. John Vorys (R-OH)1

“Local experience has taught them that in unity there ispower.”

Leo Snowiss,“Congressional Recruitment and Representation”2

1“Apportionment of State Legislatures,” Subcom No. 5, House Committee on Judiciary. Aug. 6, 1964,

HRG-1964-HJH-0043, p. 504-502

Snowiss (1966), p.630



Conclusion


Jurisdictional IHI: Municipal Boundary

4

16

35

28

6

15

98

7

27

14

12

13

1110

24

17

18

21

23

19

20

22

Percent Black,Congressional Districts

0.00 - 0.05

0.06 - 0.10

0.11 - 0.25

0.26 - 0.50

0.51 - 0.60



Conclusion


Forces for cohesion

Municipal

• Common local political identity

• Intergovernmental transfers go downtown, not to district

• Norms of reciprocity, repeated interaction



Conclusion


Jurisdictional IHI: City Government as Whip

NY Municipal Archives

“Legislative Prospects of the Fed-eral Urban Mass Transit Bill. . . asof now there is only a total ofabout 195 votes for the bill.”



Conclusion




“It is necessary that another hardcount be made of the following[NYC Congressmen]. . . it will beappreciated if the NYC TransitAuthority will. . . secure the kind ofcommitment that will stand up.”



Conclusion






Conclusion


Organizational IHI: Party

Chicago Tribune



Conclusion


Organizational IHI: Mayor as Slater

Pres. Johnson: Did you know [Libonati] had the best voting recordof any congressman, as far as I am concerned? 98 percent [partyloyalty votes].

Mayor Daley: Well, the fella we’ll send down there will have 99.3

(Frank Annunzio, the actual successor, was 100% party-loyal in thefollowing congress.)

3Conversation between Richard J. Daley and Lyndon Johnson, tape no. 6369, Miller Center Archives,

University of Virginia.



Conclusion


Mechanisms for cohesion

Municipal


• City government controls allocation to city districts

• Norms of reciprocity, repeated interaction

• Progressive ambition

Organizational/Partisan

• Control access to nomination (Mayhew 1986)

• Mobilize, win all elections, not just local offices

• Politicians “brung up” in local organizations remain “thatway” (Wilson, Snowiss)



Conclusion


Model of Representation: Electoral Connection



Conclusion


Model of Representation: City delegations

• Jurisdictional (Border) and Organizational (Party)

• IHIs vary over space and time

• H: Local IHIs will be related to representation

• H: Cities with strong IHIs will be more cohesive



Conclusion




Conclusion


Do IHIs Enhance Cohesion? Hypotheses

Jurisdictional:

HJurisdictional : CCity > CMetro,Suburbs (1)

Organizational:

HOrganizational : CStrongIHI > CWeakIHI (2)



Conclusion


Observable implications: Delegation level

City delegations will be more cohesive:

• Than sets of representatives not from single local politicaljurisdiction (eg. suburbs, metro area, national party)(Jurisdictional)

• If they have strong local partisan institutions (eg. Chicago v.LA) (Organizational)



Conclusion


Cities are more cohesive than other blocs

ChicagoA

vera

ge

Ric

e C

oh

esio

n S

co

res

CityMetro

Suburbs0.4

0.5

0.6

0.7

0.8

0.9

1.0

1920 1940 1960 1980

New York City

Ave

rag

e R

ice

Co

he

sio

n S

co

res

0.4

0.5

0.6

0.7

0.8

0.9

1.0

1900 1920 1940 1960 1980

Philadelphia

Ave

rag

e R

ice

Co

he

sio

n S

co

res

0.4

0.5

0.6

0.7

0.8

0.9

1.0

1900 1920 1940 1960 1980

Los Angeles

Ave

rag

e R

ice

Co

he

sio

n S

co

res

0.4

0.5

0.6

0.7

0.8

0.9

1.0

1900 1920 1940 1960 1980



Conclusion


Stronger organizations, more cohesionE

xpecte

d R

ice C

ohesio

n S

core

s

Chicago

Philadelphia

New York City

Los Angeles0.5

0.6

0.7

0.8

0.9

1.0

1920 1930 1940 1950 1960 1970 1980 1990



Conclusion


Cities with Stronger IHIs are Unanimous More Often

Unanimity Among City DelegationsUnanimity Among City Delegations, 70th-100th Congresses

City IHI strength All votes Domestic Votes City Votes IR votes

Philadelphia Strong 0.76 0.59 0.77 0.59Chicago Strong 0.67 0.53 0.64 0.72New York City Halfway 0.48 0.38 0.47 0.38Los Angeles Weak 0.46 0.34 0.40 0.45

All Democrats 0.10 0.07 0.06 0.11All Republicans 0.11 0.07 0.06 0.14

Total votes 13,962 10,185 1,099 2,814

Table: Proportion of votes on which city delegations were unani-mous, Congresses 70-100: Cities with strong IHIs are unanimous moreoften than cities with weaker IHIs. Source: USR Data, AIP



Conclusion


Dyadic regression model

Delegation-level cohesion may be due to other factors

• Partisan composition

• Demographic similarity

Analyze dyads

• Pairs of representatives are building blocks of delegations

• Can account for some alternative hypotheses: PARTY

• Two steps to representation: selection and voting



Conclusion




Conclusion


Factors Associated with Congressional Party Affinity, 78th-105th Congress

Model 1 2 3 4(District Dyads Included) All All City Only Same City

Same City 0.79** 0.49** 0.78**High TPO 0.68**Same State -0.21** -0.11** -0.45**Same Section 0.14** 0.11** 0.06*Same Region 0.30** 0.19** 0.05Urbanicity Similarity (7 category) 0.03** 0.02** 2.26** -0.90Similarity in FarLeft % -1.52** -0.93** -0.58 -8.69**Race (Similarity in % Non-native-white) 0.85** 0.53** -1.02** -1.63*Class (Similarity in % Blue collar) 0.13 5.98** 38.60**Union (Similarity in % in state) -0.54**

Adjusted R2 .01 .01 .02 .085N 2,171,755 2,171,755 158,504 6,851∗p <.05, ∗∗p <.01

Table: Congressional Affinity: Dyadic Regression. Dependent variable is membership in same congressionalparty, independent variables are measures of similarity on the variable listed at left. Cell entries are probit regressioncoefficients with robust standard errors, clustered by dyad. Significance levels for all key explanatory variables ofinterest (identified in grey) verified using nonparametric simulation technique in [?]. Estimated with constant termand Congress-fixed effects not listed here. * p < .05, ** p < .01. Sources: Voteview; CSR data; [?]; [?]



Conclusion


Marginal effects of IHI and other explanatory variableson Congressional Party Affinity, 78th-105th Congresses

Model 1 2 3 4(Dyads Included) All All City Only Same City

Same City .20 .19 .18High TPO .12State -.05 -.04 -.10Section .034 .044 .01Region .075 .07 NAUrbanicity .0082 .01 .51 NAFar Left -.37 -.37 NA -1.56Race .21 .21 -.23 -.29Class NA 1.35 6.9Union -.21

Table: Congressional Affinity: Marginal Effects. Cell entries are the marginal effects of similarity on dimensionsat left on dyad co- membership in congressional party. Estimated with covariates held at appropriate levels. Estimatesbased on coefficients that were not estimated to be significant at p < .05 in Table 4.3 marked NA.



Conclusion




Conclusion


Marginal Effects of City IHI Influenceon Roll Call Agreement, 78th-105th Congress

Model 1 2 3Dyads All City Pairs Same City & Party

Same City 0.05** 0.03**TPO 0.05**Party 0.26** 0.30**State 0.01** NASection 0.01** 0.058**Region 0.02** 0.011**Urbanicity 0.02** 0.042** NA% Far Left 0.41** 0.22** 0.26**% Dem -.04** 0.14** 0.36**Race 0.03** NA NAClass 0.07** -0.31** 1.3**

Table: Cell entries are the marginal effects of similarity on dimensions at left on dyad agreement in roll callvotes. Estimated with covariates held at appropriate levels. Estimates based on coefficients that were not estimatedsignificant at p < .05 in Table 4.5 marked NA.



Conclusion


Crucial Issue Area: Civil Rights

Realignment Issue, 1930s-1970s

• Second “Dimension”

• Divisive regionally, but also locally

Possible positions:

• Threat/Rivalry (Mass position)

• Contact-induced toleration (common but not modal)

• Reduced public coordination

• Pluralist accommodation (“City interest”)



Conclusion


Crucial Issue Area: Civil Rights

Local Division, National Cohesion



Conclusion


City Delegation Cohesion, Civil Rights Votes (1940-1970)

Expecte

d R

ice C

ohesio

n S

core

s

Chicago

Philadelphia

New York City

Los Angeles0.0

0.2

0.4

0.6

0.8

1.0

1940 1945 1950 1955 1960 1965 1970

Figure: Average City Delegation Cohesion on Civil Rights Votes. City delegations with strong IHIs were morecohesive on civil rights issues. Dotted line indicates congressional party average on all RCs.



Conclusion


Party Organization and Support for Civil Rights by Party, 1933-1963

.5.6

.7.8

.9P

r(A

gre

e_B

lack

)

0 1 2 3 4TPO(5)

Republicans Democrats

Probability of Supporting Civil Rights by Party and Local Organization Strength, 1933−1963

Figure: Probability of support for racially liberal position by TPO (Interaction interpretation). WhileRepublicans from strong-party places were no more likely to support the racially liberal position, Democrats fromstrong party organizations were about 37 percent more likely to take such a position than their co-partisans fromplaces with weak local parties.



Conclusion


Backlash



Conclusion



.6.7

.8.9

Pr(

Agre

e_B

lack

)

0 1 2 3 4TPO(5)

Republicans Democrats

Probability of Supporting Civil Rights byParty and Local Organization Strength, 1967−1973

Figure: Predicted Probability of support for racially liberal positionby Party and TPO (Interaction interpretation), 1967-1971. The re-lationship between party strength is positive for Democrats, but not forRepublicans.



ConclusionBack Matter

City delegations: Empirical summary

IHIs associated with cohesion

• City delegations more cohesive than non-city

• Cities with TPO more cohesive delegation

• Same-city dyads more likely to agree on party and vote,especially from strong-party cities

• . . . Despite heterogeneity, and even when we account for otherkinds of constituency similarities

• Crucially, this remains true on Civil Rights through therealignment era




Big picture

• Local institutions influence character of higher-levelrepresentation

• Bygone days?

• City unity supports pro-city policies: liberalism

• Democracy and Diversity




Thank you!




District Heterogeneity, not just diversity




District Heterogeneity, not just diversity

1960 1970 1980 1990

0.0

0.5

1.0

1.5

2.0

2.5

3.0

Chicago

2−

D C

ross−

dis

tric

t D

ivers

ity

City

Suburbs

New York

0.0

0.5

1.0

1.5

2.0

2.5

3.0

1960 1970 1980 1990

City

Suburbs

1960 1970 1980 1990

0.0

0.5

1.0

1.5

2.0

2.5

3.0

Los Angeles

2−

D C

ross−

dis

tric

t D

ivers

ity

City

Suburbs

Philadelphia

0.0

0.5

1.0

1.5

2.0

2.5

3.0

1960 1970 1980 1990

City

Suburbs

Figure: Cross-District Heterogeneity Among Congressional Delegations (1960-2000): City delegations arealmost always more internally heterogeneous than suburban delegations Source:USR data, Lublin (1997)




Number of Big−city Mayors Appearing Before Congress

0

20

40

60

83

1865 1891 1913 1931 1947 1976 1995 2007

Figure: Appearances by big-city mayors. Before the 1930s, mayors seldom appeared before Congressionalcommittees to provide testimony. They visited Washington with increasing frequency through the fiscal crises of the1970s. Their presence has diminished, but leveled off at a rate much higher than before the 30s. Source: ProQuestCongressional




1880 1900 1920 1940 1960 1980

010

2030

4050

60

year

NU

rban

Vot

es

0.00

0.02

0.04

0.06

0.08

0.10

0.12

Pro

p. U

rban

Vot

es

Number of Urban VotesProp. Urban Votes

Figure: Total urban roll call votes, and urban votes as proportion of all roll call votes, 45th-100th Congresses.The number of votes about urban issues has increased since the beginning of the urban interlude. The proportionof the overall agenda taken up by urban issues was generally high throughout the period, and peaked in the 88thcongress (1959-1961). Source: AIP




Urban-Rural Cleavage: Congress

Congressional Districts, by urbanicity−type

# o

f D

istr

icts

Urban

Suburban

Rural

1865 1913 1947 2015

0

100

200

300

400

District Urbanicity−type by Party

# o

f D

istr

icts

Urban

Suburban

Rural

1891 1933 1965 2015

0

100

200

300

400

Source: CSR data




Urban-Rural Cleavage: Congressional Leadership

Proportion of House cartel

members, by urbanicity

City

Suburban

Rural

1897 1933 1995 2015

0.0

0.2

0.4

0.6

0.8

1.0

Proportion of members of

prestige committees, by urbanicity

City

Suburban

Rural

1897 1933 1995 2015


banking committee, by urbanicity

City

Suburban

Rural

1897 1933 1995 2015

0.0

0.2

0.4

0.6

0.8

1.0

Source: CSR data




Urban Win Rates: Congress

City Bloc Win Rates by Vote Type

Pro

port

ion

All votes

Contested

Urbanicity0.0

0.2

0.4

0.6

0.8

1.0

1865 1895 1933 2011

City Bloc Leverage

Positive leverage

Potential leverage0.0

0.2

0.4

0.6

0.8

1.0

1865 1895 1933 2011

Source: CSR data




There hasn’t always been an urban-rural voting divide

−1

−0.5

00.5

1

Roosevelt 1932

Urban−Rural Divide in 20th Century Presidential Elections

% D

emocr

atic

−%

Rep

ub

lica

n

(ln) County Population Density

Kennedy 1960

−2 0 2 4 6 8 10

−1

−0.5

00

.51 Obama 2008

−2 0 2 4 6 8 10

Clinton 2016

Figure: County-level support for Democratic Candidates in 1932, 1960, and 2016 by population density in25 states. Y-axis is proportion for Democratic Candidate less proportion for Republican. X-Axis is log of populationdensity (persons/sqmi). Lines are local-fit curves for visual clarity. Support for FDR top left, JFK top right, Clintonbottom. Sources: Clubb Flanigan Zingale (2006), McGovern (2016), and National Historic GIS




Urban−Rural Divide, 1932

(Log of) Population Density

%D

em

−%

Repub

−0.5

0.0

0.5

AL

0 5 10

AZ CA

0 5 10

CO FL

IL LA MA MD

−0.5

0.0

0.5

MI

−0.5

0.0

0.5

MN MO MS NC NE

NJ NY OH OR

−0.5

0.0

0.5

PA

−0.5

0.0

0.5

0 5 10

TN TX

0 5 10

VA WA

0 5 10

WI






%D

em

−%

Repub

−0.5

0.0

0.5

AL

0 5 10

AZ CA

0 5 10

CO FL

IL LA MA MD

−0.5

0.0

0.5

MI

−0.5

0.0

0.5

MN MO MS NC NE

NJ NY OH OR

−0.5

0.0

0.5

PA

−0.5

0.0

0.5

0 5 10

TN TX

0 5 10

VA WA

0 5 10

WI




1800




1850




1900




1950




2000




Weighted Disproportionality in House Seats

GV

F (

in H

ouse S

eats

)

0.0

0.1

0.2

0.3

0.4

1865 1933 1965 2015

Distinctiveness in House Seats by Place−character

Dis

pro

port

ionalit

y (

in H

ouse S

eats

)

Rural

Suburban

City

0.0

0.1

0.2

0.3

0.4

0.5

0.6

1865 1933 1965 2015

Figure: Group-fractionalization by district place-character. At left, the summary measure of weighted dispro-portionality. Higher values mean City, Suburban, and Rural districts are increasingly different from each other inpartisan terms. At right, the general increase in place-character disproportionality is disaggregated. Higher valuesmean a given bloc is more different from the other blocs. Source: CSR data




Weighted Disproportionality in House Seats

GV

F (

in H

ouse S

eats

)

0.1

0.2

0.3

0.4

1865 1933 1965 2015

Distinctiveness in House Seats by Region

Dis

pro

port

ionalit

y (

in H

ouse S

eats

)

Northeast

Midwest

South

West

0.0

0.1

0.2

0.3

0.4

0.5

0.6

0.7

1865 1933 1965 1994

Figure: Group-fractionalization by region. At left, overall regional disproportionality. Higher values mean higherdifferences across regions. At right, distinctiveness disaggregated by region. Higher values mean a given bloc is moredifferent from the other 3 blocs. Source: CSR data




Proportion of City Representatives in Democratic Party

Pro

port

ion

1865 1891 1933 1965 2015

0.25

0.5

0.75

1

Proportion of Dem. Representatives from City Districts

1865 1891 1933 1965 2015

0.25

0.5

0.75

1

Figure: Partisanship and place-type over time. At left, the partisan balance of city districts. At right, theproportion of Democratic members of the House of Representatives that are from city districts. Source: CSR data




City Districts, 2013-

6

1

14

2

13

3

11

8

1

9

5

7

10

4

16

Chicago




1900 1920 1940 1960 1980 2000

−0.4

−0.2

0.0

0.2

0.4

Mean 1st−Dimension DW scores

ChamberUrbanSuburbanRural

Mean 1st−Dimension DW scores

ChamberUrban RepublicansNon−urban RepublicansNon−Southern Urban DemsNon−Southern, Non−urban DemsSouthern Dems

1900 1920 1940 1960 1980 2000

−0.5

0.0

0.5

Figure: At left: Mean DW-NOMINATE first dimension scores by urbanicity group over time. Lower scoresmean more liberal/Democratic At right:Mean DW-NOMINATE first dimension scores by party and urbanicity/regionsubgroup. Lower scores mean more liberal/Democratic.




Proportion of House Southerners in Democratic Party

1880 1900 1920 1940 1960 1980 2000

0.0

0.2

0.4

0.6

0.8

1.0

Figure: Share of Southern House seats held by Democrats, post-Civil War congresses. (Census RegionalDefinition)




1930 1935 1940 1945 1950 1955 1960

−0.4

−0.2

0.0

0.2

0.4

Mean 2nd−Dimension DW scores

City

Non−City

Chamber Mean

(More Conservative)

(More Liberal)

*All MCs

Mean 2nd−Dimension DW Scores

City

Non−City

Chamber Mean

*Non−southern MCS only

1930 1935 1940 1945 1950 1955 1960

−0.4

−0.2

0.0

0.2

0.4




1900 1910 1920 1930 1940 1950 1960

0.0

0.2

0.4

0.6

0.8

1.0

Civil Rights Liberalism,

1899−1963

NS Dem

GOP

S Dem

Civil Rights Liberalism,

Non−Southern Democrats, 1899−1963

NS Urban Dem

NS Non−urban Dem

1900 1910 1920 1930 1940 1950 1960

0.0

0.2

0.4

0.6

0.8

1.0




Urbanicity

TPO

Democrats

Democrats # TPO

Non−Southern Dem

Non−Southern

%Margin of Victory

%Far Left

%Black VEP (Alt)

union_std

%Black VEP

%Native−born white

% Union (in state)

Black Partner

%Blue Collar

−2 0 2 4 6 8Standardized Coefficients

Logit Coefficients for Alternative Specificationsof Model of Civil Rights Support, 1933−1963




Proportion of House cartel

members, by urbanicity

City

Suburban

Rural

1933 1965 1995 2011

0.0

0.2

0.4

0.6

0.8

1.0


prestige committees, by urbanicity

City

Suburban

Rural

1933 1965 1995 2011


banking committee, by urbanicity

City

Suburban

Rural

1933 1965 1995 2011

0.0

0.2

0.4

0.6

0.8

1.0




Yonkers

Worcester

Troy

Trenton

Syracuse

Springfield

Scranton

Rochester

Reading

Providence

Pittsburgh

Philadelphia

Paterson

Newark

New York

New Haven

New Bedford

Lowell

Jersey City

Hartford

Fall River

Camden

Cambridge

Buffalo

Brooklyn

Bridgeport

Boston

Allegheny

Albany

1790 1810 1830 1850 1870 1890 1910 1930 1950 1970 1990 2010

Northeast

Youngstown

Wichita

Toledo

St. Paul

St. Louis

St. Joseph

Omaha

Minneapolis

Milwaukee

Kansas City

Indianapolis

Grand Rapids

Flint

Detroit

Des Moines

Dayton

Columbus

Cleveland

Cincinnati

Chicago

Akron

1790 1810 1830 1850 1870 1890 1910 1930 1950 1970 1990 2010

Midwest

Wilmington

Washington

Virginia Beach

Tulsa

Tampa

St. Petersburg

San Antonio

Richmond

Raleigh

Oklahoma City

Norfolk

New Orleans

Nashville−Davidson

Mobile

Miami

Memphis

Louisville

Jacksonville

Houston

Fort Worth

El Paso

Dallas

Corpus Christi

Charlotte

Charleston

Birmingham

Baltimore

Austin

Atlanta

Arlington South

Tucson

Spokane

Seattle

Santa Ana

San Jose

San Francisco

San Diego

Salt Lake City

Sacramento

Portland

Phoenix

Oakland

Mesa

Los Angeles

Long Beach

Las Vegas

Honolulu

Fresno

Denver

Colorado Springs

Bakersfield

Aurora

Anaheim

Albuquerque West





New York Census Tracts,

1940

0.0 - 0.010

0.011 - 0.050

0.051 - 0.10

0.11 - 0.25

0.26 - 0.50

0.51 - 1.0




0 2

1

8

11

24

9

6

23

4

5

3

7

17

22

21

10

19

1516

18

1413

20

12

6


New York Congressional Districts,


0.01 - 0.05

0.06 - 0.10

0.11 - 0.25

0.26 - 0.36





Los Angeles Census Tracts,

1940

0.0 - 0.010

0.011 - 0.050

0.051 - 0.25

0.26 - 0.50

0.51 - 0.93




11

1216

1817

13

14

15

10

19


Los Angeles CongressionalDistricts,


0.00 - 0.05

0.06 - 0.10

0.11 - 0.25

0.26 - 0.50

0.51 - 0.59





Philadelphia Census Tracts,

1940

0.0 - 0.010

0.011 - 0.050

0.051 - 0.10

0.11 - 0.25

0.26 - 0.50

0.51 - 1.0




5

7

6

1

3

4

2

8

9


Philadelphia Congressional Districts,


0.00 - 0.05

0.06 - 0.10

0.11 - 0.25

0.26 - 0.50

0.51 - 0.59Tom Ogorzalek Ties That Bind



Antiracism without Antiracists

City IHIs associated with early support for civil rights in House

• Pursue urbanizing strategy, while GOP and non-urbanitessilent

• Unanimous support for civil rights by those who share localparty with Afam Rep

• 7 percent increase in likelihood of support for civil rights bythose in strong IHI context

Local institutions driving national political change




Urbanizing Issues: A Model of City Representation

How do cities pursue a “city interest” in higher arenas? (Burns etal 2009)

• Identify an issue as distinctively “urban”

• Articulate the city position

• Defend/represent that position cohesively

• Encourage non-city co-partisans to defer to city position




Unanimity Among City Delegations

City IHI strength All votes Domestic City IR

PHI Strong 0.76 0.59 0.77 0.59CHI Strong 0.67 0.53 0.64 0.72NYC Halfway 0.48 0.38 0.47 0.38LAX Weak 0.46 0.34 0.40 0.45

Dem 0.10 0.07 0.06 0.11GOP 0.11 0.07 0.06 0.14

Total votes 13962 10185 1099 2814

Table: Proportion of votes on which city delegations were unani-mous, Congresses 70-100: Cities with strong IHIs are unanimous moreoften than cities with weaker IHIs. Source: USR Data, AIP




Forces for cohesion/Vertical integration

Municipal


• Intergovernmental transfers go downtown, not to district

• Norms of reciprocity within organization, repeated interaction

Organizational/Partisan

• Strong local organizations control access to nomination(Mayhew 1986)

• Strong local organizations can mobilize, win all elections, notjust local

• Politicians “brung up” in local organizations remain “thatway” (Wilson, Snowiss)




Percent African American, Census TractsChicago 1940

0 - 0.010

0.011 - 0.050

0.051 - 0.25

0.26 - 0.50

0.51 - 0.75

0.76 - 1.0

Figure: Chicago, Percent blackin census tracts, 1940 Census.African Americans were heavily con-centrated within cities like Chicago.

3

7

2

6

4

10

1

9

8

5

Percent African American, Congressional DistrictsChicago 1940

0.00 - 0.05

0.06 - 0.10

0.11 - 0.25

0.26 - 0.50

0.51 - 0.60

Figure: Chicago, Percent black incongressional districts, 1945 Res-idential segregation led to politicalsegregation: few congressmen hadmany black constituents.




Pairwise model

Figure: City Delegation Theory: Local city IHIs complicate the tradi-tional model of Congressional representation (denoted by the relationshipsof A1, A2, and C ). Factors external to both district and chamber willinfluence representation.




Representation: selection

Jurisdictional:

Pr(Party) = City + Region + Section + State + Race + Class + Urbanicity + ε (3)

Organizational:

Pr(Party) = City*TPO+City+TPO+Region+Section+State+Race+Class+Urbanicity+ε(4)

• Measures are similarity on dimension in question.

• Vote- and Congress-level fixed effects

• Robust SEs and non-parametric estimation




Tests of City IHI Association with Congressional Party Affinity

Model 1 2 3 4 5 6Pool (Dyads) All All All All City Only Same City OnlyCongresses (45+) (77+) (77+) (77+) (77+) (77+)City*TPO 0.28* 0.29*

(0.039) (0.039)SameCity .44* 0.51* .52* 0.31* 0.39*

(.011) (0.015) (.016) (0.035) (0.036)High TPO -0.068* -.08* 0.37*

(0.002) (0.004) (0.047)SameState -.088* -0.12* -.11* -0.11* -0.24*

(.003) (0.004) (.004) (0.004) (0.007)SameSection .25* 0.09* .12* 0.11* -.14*

(.0012) (0.001) (.002) (0.0019) (0.003)SameRegion .17* 0.17* .17* 0.18* .12*

(.001) (0.002) (.002) (0.0022) (0.004)Urbanicity .0097* 0.013* 0.12* 0.012* 0.077* .026

(.0038) (0.0062) (.001) (0.001) (0.0011) (0.11)Race -0.12* -.12* -0.14* -.39* -1.63*

(0.012) (.012 (0.012) (0.019) (0.23)Class -0.006 -0.11 -1.14* 13.83*

(0.06) (0.0659) (.12) (1.81)Union -.005*

(.0002)

Pseudo-R2 .02 0.007 .007 0.0067 0.013 .0727N 5,880,127 2,625,712 2,625,712 2,625,712 736,162 7,735

Table: Probit Regression Results: City Delegation Models with different samples of congressional dyads.Dependent variable is membership in same congressional party, independent variables are measures of similarity onthe variable listed at left. Cell entries are probit regression coefficients with robust standard errors, clustered by dyad.Congress and vote fixed effects not listed here. *p < .05




Marginal effects of IHI and other explanatory variableson Congressional Party Affinity, 78th-105th Congress

Model 1 2 4 5 6Dyads All All All City Only Same City Only

(Post-78) (Post-78) (Post-78) (Post-78)City*TPO (Fig. 18) 0.11*City .17* 0.20* 0.12* 0.15*TPO -0.027* -.032* .13*State -.034* -0.050* -0.042* -0.10*Section .10* 0.040* 0.043* 0.056*Region .069* 0.070* 0.070* -0.049*Urbanicity .0039* 0.0050* 0.0046* 0.031* -0.008RaceSim -0.052* -0.055* -.16 -0.52*ClassSim -0.002 -0.0045 -.45* 4.35*

Table: Marginal Effects: City Delegation Models with different samplesof congressional dyads. *p < .05




Effects of SameCity for Cities with Different strength TPOs

Pr(

Sam

e P

art

y)

HiTPO

LoTPO

Not Same City Same City

0.2

0.3

0.4

0.5

0.6

Figure: Strength of Same-city effect on large-city dyads byorganization-type. Congressional dyads, 1939-1999. Difference in slopesignificant at p < .05. Diff-in-diff: .09




Jursidictional:

Pr(AgreeVote) = City+Party+Region+Section+State+Race+Class+Urbanicity+ε(5)

Organizational (Same City Only):

Pr(AgreeVote) = TPO + Race + Class + Urbanicity + ε (6)




Tests of City IHI Influence on Roll Call Voting

Model 1 2 3 4 5Pool All All All City Votes Same City, Party(Congresses) (73-89) (78+) (78+) (78+) (78+)City .18* .24* .24* 0.37*

(.01) (0.013) (.012) (0.014)Party .65* 0.72* .71* 0.74*

(.001) (0.001) (.001) (0.0016)High TPO .86*

(.27)State .065* 0.050* .03* 0.034*

(.003) (0.004) (.004) (0.004)Section .045* .005* -.02* -.068*

(.001) (0.002) (.002) (0.002)Region .01* -.004* .0002 -.035*

(.002) (0.002) (.002) (0.002)Urbanicity .023* 0.030* 0.032* 0.050* -.28

(.0016) (0.0005) (.0005) (0.0006) (.17)Race -0.3* -0.20* -.24* .37* -.59*

(0.016) (0.018) (.018) (0.019) (.35)Class 0.015* .014* .15*

(0.00030) (.0005) (.024)Union .007*

(.0002)

Pseudo-R2 .06 0.06 .07 0.119 .16N 6,407,860 4,621,162 4,621,162 8,272,388 11671

Table: Probit Regression Results: DV: agreement on vote, independent variables are measures of similarityon the variable listed at left. Cell entries are probit regression coefficients with robust standard errors, clustered bydyad. Shaded rows are coefficients of interest. All models include (unlisted) congress- and vote-specific fixed effects,and model 7 also includes city fixed effects. *p < .05




Probit regression: Dyad-level agreement on civil rights

Model 1 2 3Dyad Pools All Dyads Big-city only Same City, Same partyVariable Coeff. Mfx Coeff. Mfx Coeff. Mfx

SameCity .21** .07 .03* .015hiTPO .44** .11

SameParty .015* .0047 .09** .02State .081** .03 .000 –Region -.08** -.025 -.039** -.01Section 1.13** .32 1.38** .39CSRsim .07** .02 .416* .11Racesim 1.38** .44 .57** .16 .99** .24Classsim 4.56** 1.45 .004 – 5.36 –%FarLeftsim -.24* -.07 .16 .05 -.15 –

Psuedo-R2 .19 .24 .16N 2,288,558 177,655 6,430

Table: Dyad-level agreement on Civil Rights, 1933-1963. Dyads from strong-party cities were more likely toagree on civil rights roll calls. Key explanatory variables of interest highlighted in gray. Marginal effects for statisticallysignificant coefficients estimated as expected change in likelihood of dyad agreement given one-category shift in thedependent variable, other variables held constant at appropriate values (*p < .10, **p < .05). Significanceestimated with non-parametric shuffling procedure described in Rader et al 2014; significance at p < .10 means thatobserved test statistic (Z-score) lies outside 5-95 percentile range of test statisticsfor that coefficient in simulatedmodels. Estimated with an intercept, congress-, and vote-level fixed effects not listed here.




1930 1940 1950 1960

−0.4

−0.2

0.0

0.2

0.4

Mean 2nd−Dim. DW scores

City

Non−City

Chamber Mean

(More Conservative)

(More Liberal)

Mean 2nd−Dim. DW scores

City

Non−City

Chamber Mean

1930 1940 1950 1960

−0.4

−0.2

0.0

0.2

0.4

Figure: Group means on DW-NOMINATE 2nd-Dimension scores, 1930-1970. In the chamber as a whole,and outside the South, city representatives were more liberal on average.




Table: Linear regression of DW-NOMINATE second-dimension scores

Variable (#categories) Coeff. (Robust Std. Err.)Urbanicity(7) -0.053 (0.029)*Democrat(2) .353 (0.018)*South(2) .524 (0.021)*Intercept -.166 (0.013)*

Table: Linear regression of DW-NOMINATE 2nd Dimension scores,1930-1960. City representatives were on average more “liberal” thansuburban or rural representatives on issues of race and region. (*p < .01,N=6313, R2 = .70)




Multivariate Analysis, Civil Rights Liberalism 1933-1963DV: Agreement with African American position

Predictors of interest

• Urbanicity (City-Suburban-Rural district indicator)• City delegation descriptor (Indicator for white MCs from city

delegation with black MC of the same party)• Local party strength (TPO score, interacted with Democrat

indicator)

Alternative explanations

• Congressional Party (Democrat indicator)• Section (Non-south indicator)• Constituency Pressures (Union Density, % Afam, %Far Left)

Sample: Civil Rights roll calls, 73rd-88th Congress

Logit, robust SEs clustered by legislator, vote and congress fixed effects




Civil Rights Roll Calls

Year* # of votes Subject(s)1899 3 Frederick Douglass Statue1921 22 Anti-lynching1937 4 Anti-lynching1939 2 Anti-lynching1945 3 School Integration1949 11 Fair Employment1951 1 **1957 6 Civil Rights Commission, Act1959 7 Civil Rights Commission, Act1961 1 Civil Rights Commission

Total 35

Table: House roll calls about Civil Rights for African Americans by Congress, 1899-1963. *Year Congressbegan. **Unclear from AIP data what civil rights dimension of this appropriations roll call was. (Source: AIP data)




Logit regression: Agreement with African American position on civil rights

Variable (#categories) Coeff. (Robust SE) MfxUrbanicity (7) 0.173** (0.038) .020BlackPartner *** – –TPO*Dem 0.519* (0.211) (Fig.20)TPO (5) 0.083 (0.126) –Non-south Dem. Seat (2) 2.705** (0.452) .21Dem. Seat (2) -1.064* (0.501) -.12Non-south(2) 1.311** (0.368) .18BlackVEP (%) 1.728 (3.755) –Union (%) 0.067** (0.011) .007Margin of Vic. (%) -1.060** (0.322) -.12% Dem -1.705** (0.550) -.19% FarLeft 3.393 (8.862) –

Table: Civil Rights Liberalism, 1933-1963. ***City representatives and those with a black partner in their citydelegation were more likely to support civil rights. Local black partnership was perfect predictor of support, thoseobservations are dropped from this model. Exclusion of the variable does not reduce the magnitude or significance ofthe other predictors of interest. Marginal effects estimated as expected change in likelihood of agreement given one-category shift in the dependent variable, other variables held constant at appropriate values (*p < .05, **p < .01

N=7620, Psuedo-R2 = .49. Estimated with an intercept, vote-level fixed effects, and robust standard errors clusteredby legislator.)





.5.6

.7.8

.9P

r(A

gre

e A

fam

)

0 1 2 3 4TPO

Republican Democrat

Figure: Probability of support for racially liberal position by TPO (Interaction interpretation). WhileRepublicans from strong-party places were no more likely to support the racially liberal position, Democrats fromstrong party organizations were about 37 percent more likely to take such a position than their copartisans fromplaces with weak local parties.




City IHIs

City representatives supported civil rights liberalism

• Related to local institutions and “city interest”

• Racism present in streets and city hall, but not in nationalrepresentation

Analyze IHIs and cohesion directly




Dyadic Estimation: Significance Test

Dyadic Agreement as DV

• Observations not independent (voters appear in many dyads)

• Higher change of Type 1 Error with Robust Standard Errors

• Non-Parametric Significance Test (Rader et al 2014)

• Shuffle explanatory variable of interest 1000x

• Save simulated test statistics, compare distribution toobserved




Dyadic Estimation: Significance Test

010

20

30

40

−4 −2 0 2 4

Simulated 5−95 Range Simulated Distribution

Observed test stat

Figure: Significance of SameCity, Model 5, Dyad Vote Analysis. Density plot of simulated Z-scores




Cohesion scores

Cohesion: To what extent is a bloc voting cohesively?

• 1=Perfect Cohesion, 0=Evenly split

• Cohesion= |Yea−Nay |Total

• Bias correction when comparing blocs of very different sizes(Desposato 2005)

• E (C |Yea,Nay ,Total) = Yea(Yea−1)+Nay(Nay−1)Total(Total−1)

Likeness: How alike are two blocs of voters?

• 1=Identical in proportions, 0=Completely in disagreement

• LikenessAB = 1− | YeaATotalA− YeaB

TotalB|


ties that bind - cpb-us-e1.wpmucdn.com€¦ · ties that bind social heterogeneity and cohesive...

Documents