Flooded Communities

• Published date: 01 June 2012
• Date: 01 June 2012
• DOI: 10.1177/1065912911398050
• Subject Matter: Articles

Political Research Quarterly

65(2) 443 –459

© 2012 University of Utah

Reprints and permission:

sagepub.com/journalsPermissions.nav

DOI: 10.1177/1065912911398050

http://prq.sagepub.com

398050PRQ65210.1177/106591291139

8050HopkinsPolitical Research Quarterly

1Georgetown University, Washington, DC, USA

Corresponding Author:

Daniel J. Hopkins, Georgetown University,

Department of Government, Washington, DC 20057

Email: dh335@georgetown.edu

Flooded Communities: Explaining

Local Reactions to the

Post-Katrina Migrants

Daniel J. Hopkins1

Abstract

This article uses the post-Katrina migration as an exogenous shock to test theories of racial threat while minimizing

concerns about selection bias. Drawing in part on a new survey of 3,879 respondents, it demonstrates that despite

the national concern about issues of race and poverty following Katrina, people in communities that took in evacuees

became less supportive of spending to help the poor and African Americans. The results suggest a novel hypothesis

that threatened responses to newcomers hinge on both local conditions and the frames that develop around their

arrival.

Keywords

political methodology, public opinion, political participation, race, ethnicity, politics

Before Hurricane Katrina, one town in rural Arkansas

had fewer than one thousand residents and a single Afri-

can American. According to a state official, the Ku Klux

Klan was active nearby. Yet for a short while, Hurricane

Katrina changed those demographics dramatically. Over-

night, a local church camp became a shelter to 350 evacu-

ees, almost all of whom were poor African Americans. In

the words of one community leader, “The majority of the

people here in [town] . . . were very angry with us. In the

beginning, people were very prejudiced. Once these peo-

ple got to know them, everyone’s outlook totally changed.”

This town was far from unique. In the weeks following

the storm, Katrina evacuees were scattered across the

nation. Treating the post-Katrina migration as an exoge-

nous shock, this article investigates how people’s

politi cal attitudes changed in response to the sudden

demographic shifts in their communities that followed in

the storm’s wake.

Since at least the work of Key (1949), scholars in the

racial threat tradition have been interested in how the

racial, ethnic, partisan, and class composition of one’s

surroundings shapes one’s social and political views.

Scholars have been interested as well in the extent to which

direct contact across lines of race or class can influence

intergroup attitudes (Pettigrew 1998). Yet past work in

these traditions has faced significant methodological

challenges. On account of geographic mobility, it is

impossible to know whether observed correlations result

from contextual effects or from individuals’ self-selection

into differing environments.1 The post-Katrina migration

provides an unusual opportunity to test theories of racial

threat and intergroup contact in a case where those

encounters could not have been anticipated.

The next section details these two long-standing theo-

retical approaches and also identifies a theoretical chal-

lenge common to both. If it is true that Americans think

about politics in terms of collective evaluations and out-

comes, then it is unclear what role local experiences play

in shaping political views. Information gleaned from con-

tact or from one’s immediate environment, while readily

available, might remain separate from one’s political or

social attitudes. In response, that section develops an

alternative one might term “constructivist” emphasizing

the interplay of local conditions and the frames that jour-

nalists and community members develop to make sense

of those conditions. Local conditions do not give rise to

any immediate political conclusions until they are framed

in politically relevant ways. Put differently, the changing

demographic environment and salient frames interact to

produce contextual effects. This “politicized places”

approach explains not just the valence of individuals’

444 Political Research Quarterly 65(2)

responses to contact or local contexts but the specific atti-

tudes that change.

To test the competing predictions, this article’s third

section introduces the 2006 Social Capital Community

Benchmark Survey, a study of 3,879 southern respon-

dents with oversamples in heavily affected communities

such as Harris County, Texas (Houston), and East Baton

Rouge Parish, Louisiana. The exogenous variation in

community demographics induced by the hurricane gives

researchers an unusual opportunity to confront concerns

about self-selection since residents could not have chosen

their cities knowing that they would later become home

to the evacuees.2 This research design also provides unusu-

ally good balance on many covariates (or independent

variables), as there are many surveyed individuals who

are highly similar but for the city in which they hap-

pen t o live. To provide additional information about a

key cou nterfactual—what people’s attitudes would have

been absent the post-Katrina migration in Harris County—

this article turns to supplemental evidence from the

General Social Survey (GSS). As discussed in the subse-

quent section, the core results prove robust to multiple

estimation techniques including matching. These addi-

tional tests indicate that the results are not dependent on

the spe cific statistical model or missing data strategy used.

This research represents a rare test of theories of framing

with real-world events rather than survey experiments.

The study’s design maximizes its leverage over the

problem of self-selection. None of the prior research cited

here exploits exogenous demographic shocks.3 By com-

paring survey respondents in Harris County and Baton

Rouge with those from more than 700 zip codes and 312

counties in unaffected areas, this article can convincingly

demonstrate attitudinal differences between affected and

unaffected areas.4 However, determining why communi-

ties responded in particular ways is a harder problem.

While the evacuees’ arrival was exogenous, the local

frames, media coverage, and other responses were not.

Moreover, the evacuees in Houston and the surrounding

Harris County were not identical to those in Baton Rouge:

the latter group included more middle-class evacuees.

The article thus introduces a wide range of additional evi-

dence to rule out alternative explanations. Key proposi-

tions are reinforced with fifty-nine in-depth interviews.

In places, the article draws on data about crime rates and

the geographic distribution of homicides. Elsewhere, it

draws on other surveys of the affected communities to

trace trends before and after Katrina. Its fifth section

presents a content analysis of local television and news-

paper coverage.

Based on the remarks of the relief worker above, we

might expect that the demographic shocks after Katrina

increased support for assistance to the needy or to blacks.

The presence of a visible natural disaster—as well as the

halting governmental response—made it clear that the

victims were not to blame for their predicament.4 In some

respects, the aftermath of Hurricane Katrina was a likely

case for positive contextual influence given the highly

public ordeal suffered by the evacuees. And indeed, in the

days after the storm, residents of the host communities

mounted a tremendous response that undermines simplis-

tic notions of racial threat (see also Aldrich and Crook

2008). One church leader was told by an evacuee, “We

didn’t know white people could love us.”

Yet this extraordinary volunteerism makes the central

finding of this article all the more puzzling: Why was it

that during a period of unusual altruism, attitudes toward

African Americans and the poor actually hardened in

some places? As this article demonstrates, residence in

host communities like Baton Rouge led to reduced sup-

port for spending on the poor and more negative affect

toward African Americans. Meanwhile, in Harris County,

the influx of evacuees led to increased demand for anti-

crime spending. The anticrime results do not appear to

reflect preexisting differences since no such attitudinal

differences are detected in prior surveys. Nor do they

appear to be direct responses to rising crime since the

spikes in violent crime were similar in both places. The

sixth section probes alternative explanations. Comparing

2002 and 2006, it shows that Houston residents became

more worried about lower levels of crime.

The challenge is to explain the negative attitudinal

response and the differing forms it took. As this article

illustrates, those who had direct contact with evacuees

did not have more positive attitudes, ruling out the notion

that these effects were driven by interpersonal contact.

Since political elites in the host communities studied here

received acclaim for their efforts on the evacuees’ behalf,

it seems unlikely that they led any visible effort to demon-

ize the evacuees.6 Instead, the sudden change in local

demographics appears to have interacted with frames

about evacuee benefits, joblessness, and criminality to

produce marked changes in many community members’

views. In Harris County, prominent frames linked the

evacuees to crime, and residents responded to their

changing context by becoming more anticrime. In Baton

Rouge, a dominant frame associated the evacuees with

government benefits, again suggesting an explanation for

the specific attitudes that changed. In both places, the

combination of physical proximity and framing appears

to have produced contextual effects. The next section

provides the theoretical background for these claims.

Theorizing Contexts

Scholars developed contextual theories primarily to

explain how Americans respond to local racial demo-

graphics. The dominant “racial threat” approach contends