Over 40 years of research clearly indicates that African Americans tend to have lower levels of political trust and confidence in governmental institutions than their White counterparts (Dawson, 1994; Engel, 2005; Howell & Fagan, 1988; Hurwitz & Peffley, 2005; Kinder & Sanders, 1996; Sigelman & Welch, 1991; Tate, 1994; Tyler, 2001). African Americans' distrust of government has often been attributed to the historical marginalization of Blacks in every element of American society (Bobo, 2006; Hochschild, 1995; Kinder & Winter, 2001; Sigelman & Welch, 1991). In addition, research on racial identity suggests that this marginalization, in combination with the development of a strong individual racial identity, contributes to mistrust and lack of confidence by African Americans in political institutions (Hochschild, 1995; Kinder & Sanders, 1996; Kinder & Winter, 2001).

We apply this approach to develop an understanding of how the American public had such varied responses to government efforts to assist citizens in the aftermath of Hurricane Katrina. Many observers noted how the storm revealed continuing racial and economic disparities in America (Bartels, 2006; Bobo, 2006), highlighting the fact that the great majority of victims in the New Orleans area were African American (Bobo, 2006). Indeed, to many, the average victim of the storm appeared to be a low-income African American (Huddy & Feldman, 2006; Sweeney, 2006).1 Perhaps not surprisingly, national polls conducted in the wake of the storm clearly demonstrated large differences in the way that Blacks and Whites viewed government response to the storm (Huddy & Feldman, 2006; Newport, Ludwig, & Kearney, 2005; Saad, 2005). This disparity in attitudes seems to have resulted from the predominance of African Americans in media coverage of the storm's aftermath and a subsequent identification with the victims by African Americans around the country. Our research uses unique national survey data to examine whether or not the racial attitude disparity holds after accounting for additional respondent characteristics, such as partisanship, income, and gender, which might also explain individual attitudes.

We begin with a brief overview of the literature on racial differences in attitudes. Within this context, we explore framing as a means for better understanding the attitudes of Blacks toward government response to Hurricane Katrina. Using data from a national sample of adults, we then test specific hypotheses regarding attitudes about government response to Katrina. Our analysis reveals deep racial differences in the aftermath of the storm and suggests that perceptions of weak government performance may have long-term implications on the confidence of African Americans in the federal government's ability to protect citizens from future natural and man-made disasters.

Theoretical Overview

There is a large body of research on group membership and its effects upon those who believe themselves to be part of a group (Chong & Rogers, 2005; Demo & Hughes, 1990; Federico & Luks, 2005; Gay, 2004; Huddy & Feldman, 2006; Schildkraut, 2005; Simien & Clawson, 2004; Stets & Burke, 2000; Tajfel, 1978; Tate, 1994). Individuals tend to categorize themselves as part of a group for several different reasons, including a shared set of beliefs or values because of a sense of belongingness or connection with other group members (Broman, Neighbors, & Jackson, 1988; Demo & Hughes, 1990). Additionally, individuals coalesce around shared characteristics such as race, ethnicity, or gender, as well as religious or political affiliation (Glaser, 1995).

Our focus is on the extent to which race is a group category that affects political attitudes in the context of events. Research suggests that the attitudes of Blacks are shaped by their experience as African Americans (Conover, 1984, 1988; Dawson, 1994; Huddy & Feldman, 2006; Hurwitz & Peffley, 2005; Sigelman & Tuch, 1997; Tate, 1994; Tyler, 2001). For example, Tate (1994) suggests that African Americans tend to have lower levels of trust in government, in large part, because they continue to face social and political intolerance.

How then does racial group membership become salient for individuals in interpreting events? One way that membership is made salient is through the process of framing. According to Nelson and Oxley (1997, p. 221), framing can be understood as "the process by which a communication source constructs and defines a social or political issue for its audience." All political or social issues and events have multiple dimensions (Iyengar, 1990; Nelson, Sanbonmatsu, & McClerking, 2007) and the extent to which certain dimensions are stressed, or framed, over others helps to "determine what we think about a particular issue" (Gilens, 1996, p. 293). Framing can influence attitudes about virtually any issue or event, including poverty (Gilens, 1996; Iyengar, 1990), crime and violence (Valentino, 1999), inequality and discrimination (Gandy, Kopp, Hands, Frazer, & David, 1997), and the attribution of blame for events (Haider-Markel & Joslyn, 2001). Political elites, the media, and virtually any public figure can employ frames that have the potential to shape mass attitudes (Chong & Druckman, 2007).

Our focus is on news media framing of Hurricane Katrina's aftermath. The news media have an especially important role in being able to shape perceptions of political and social events (Chong & Druckman, 2007; Iyengar, 1990, 1991; Iyengar & Kinder, 1987). For example, empirical evidence demonstrates that African Americans are usually represented in a negative light and are usually portrayed as an "under-class" in U.S. society (Gilens, 1996, pp. 536-7). Television news media in particular tends to emphasize poverty in the African American community (Gilens, 1996, p. 522). Additionally, the race of individuals appearing on television influences public perceptions of that group (Iyengar & Kinder, 1987). Finally, nonnews media also tends to portray Blacks in a negative light (Dates & Barlow, 1993).

If media framing of Blacks can shape perceptions of Blacks, it seems likely that framing an event or issue in terms of race will make it more likely that the event is understood in terms of race (Nelson et al., 2007). For an event such as Katrina, the portrayal of African Americans as the main victims should invoke affect toward African Americans. As group members, African Americans have greater positive affect toward African Americans (Gilens, 1998); the framing of Katrina as a disaster largely affecting Blacks would invoke more negative feelings about government response among Blacks, relative to other groups.

In addition, research indicates that individual identification within a particular group varies so that some individuals may have stronger identification than others (Broman et al., 1988; Gay, 2004; Sigelman, Steven, Tuch, Jack, & Martin, 2005). Factors such as age, education, income, region of residence, and social environment will all contribute to an individual's identity strength (Broman et al., 1988; Gay, 2004; Sigelman et al., 2005). African Americans, like any other group, manifest intra-group differences in strength of identity; these differences are conditioned by variation between group members in terms of political ideology, partisanship, education, and socioeconomic status (Dawson, 1994; Gay, 2004; Kinder & Sanders, 1996; Kinder & Winter, 2001; Tate, 1994). As such, in predicting differences in attitudes based on race, several intervening variables must be taken into account

The Framing of Katrina: Racializing a Disaster

On August 29, 2005, category 4 Hurricane Katrina wreaked havoc along the nation's Gulf Coast (Gugliotta & Whoriskey, 2005). Mississippi Governor Haley Barbour described the damage in his state as looking like "Hiroshima" (Treaster & Kleinfield, 2005). In Louisiana, Governor Kathleen Babineaux Blanco said, "I can't say we've escaped the worst I think there is still damage that can be inflicted on [New Orleans]" (Treaster & Zernike, 2005). The worst damage in New Orleans occurred when the levees around New Orleans failed. As New Orleans flooded, news media reports focused attention on groups of armed individuals looting, the over 10,000 people stranded in the Louisiana Superdome without power, and the isolation of those left behind (Treaster & Kleinfield, 2005; Whoriskey & Coates, 2005).

As images of victims made homeless by the flooding cascaded across television screens, Americans became increasingly upset about the slow and inadequate government response to the storm and its aftermath (Bartels, 2006). Government officials hastened the sense of failure by blaming each other for the lack of preparation and inept response to the flooding in New Orleans (Bobo, 2006; Sweeney, 2006).

But perhaps most striking about the media coverage of these events was the dominance of images of storm victims who were poor and African American (Alter, 2005; Bartels, 2006; DeParle, 2005; Huddy & Feldman, 2006; Street, 2005; Sweeney, 2006; Zeleza, 2005). At least one analysis clearly demonstrated that newspaper coverage of the aftermath doubled its use of the terms race and African American over the weekend following Katrina's landfall (Project for Excellence in Journalism, 2005). The images of Black victims very likely made the average Black viewer feel more affinity toward the victims than viewers of other races (Huddy & Feldman, 2006). Polling firms reinforced this perspective. Although we were unable to locate any poll prior to Katrina that asked respondents to assess government response in terms of race of the victims or attribute poor response to the race of victims, virtually every major poll conducted by ABC, CBS, NBS, FOX, CNN, Gallup, and the Pew Research center in the days following Katrina's landfall asked questions about the race of hurricane victims (see also Huddy & Feldman, 2006). In identifying more closely with the Black victims and having a stronger positive affect toward members of their own group, Blacks should have come to more strongly believe that government was slow, inadequate, or otherwise meager.

In addition, the images of Black storm victims were sometimes more negative than those shown of Whites (Kinney, 2005). In one well-publicized example, the Associated Press wire service distributed two stories on August 30, 2005. In one, a White couple is shown wading through floodwater and the caption reads "Two residents wade through chest-deep water after finding bread and soda from a local grocery store . . . (italics added)." In the other image, a young Black man is shown in nearly the exact same situation but the caption reads "A young man walks through chest deep flood water after looting a grocery store in New Orleans . . . (italics added)." The conclusion is that White victims of Katrina found food while Black victims stole food (Kinney, 2005). As a whole, Americans watching media coverage of the storm's aftermath saw victims who were predominantly Black and who were often painted in a less than positive or sympathetic light (Alter, 2005; Bartels, 2006; Street, 2005). Thus, the racialized framing of the events in New Orleans may have enhanced a negative attitude among Blacks about government response, but at the same time, led Whites to have less sympathy for the predominantly black victims of the flooding (Huddy & Feldman, 2006).

In sum, given the racialized news media frame of Black hurricane victims, we hypothesize that attitudes regarding the aftermath of Hurricane Katrina will be strongly shaped by an individual's race. In particular, we expect that African Americans, feeling a stronger connection to the image of storm victims, will hold more negative attitudes about government response than non-Blacks.

However, lenses other than race, including socioeconomic status, could shape attitudes about government response to Katrina's destruction. Therefore, any model predicting attitudes about government response should also account for factors such as income, age, gender, education, media exposure, and partisanship (Huddy & Feldman, 2006; Nelson et al., 2007).

Variables and Measurement

Dependent Variables

To assess the role race played in shaping attitudes following Katrina, we examine data from a national Gallup telephone poll conducted of American adults September 8-11, 2005. The survey was conducted as a random sample of 1,100 adults, however, an additional over-sample of 255 African Americans were also surveyed (N = 1,355).2 Thus, this survey provides a relatively unique opportunity to examine racial differences in attitudes regarding government response to a natural disaster. In addition, the over-sample of African Americans offers not only a sizable comparison group to Whites, but also allows for a comparison of differences among African Americans in terms of racial identity.

We measure attitudes about government response to Katrina with a series of questions from the survey (full question wording and descriptive statistics are in the Appendix). Respondents were asked whether President Bush's initial response to the storm and aftermath was poor or good, whether the Bush administration or local or state officials deserve the most blame for New Orleans residents being trapped in the city, and whether the federal government was slow to rescue people because many were poor or because many were Black. Across each of these questions, we expect that Blacks will provide a more negative assessment than non-Blacks.

We also assess the impact of the storm on general perceptions of President Bush and federal government response to future disasters. Respondents were asked whether or not they believed President Bush cares about Black people, whether they would have confidence in the ability of the federal government to respond to similar future natural disaster, and whether they would have confidence in the ability of the federal government to respond to future terrorist attacks. These questions assess the potential legacy of perceived poor federal government response to the storm. We expect that Black respondents will be less likely to indicate that President Bush cares about Black people and less likely to have confidence in the federal government to respond to future natural disasters or terrorist attacks. For those questions that allowed for dichotomous responses, we estimated equations using probit. For those categorical responses with three or more choices, we estimated equations using ordered probit (see Appendix for response categories).

Independent Variables

By reason of the manner in which the news media framed the aftermath of Katrina, we expect race to play a central role in explaining attitudes following Katrina. We simply coded African Americans as one and all others as zero. As a whole, we expect that African Americans will have less positive attitudes about all issues relating to government response to the storm and less confidence in the government to handle similar events in the future.

In addition, we expected that respondents who live in areas with a higher African American population might have different views about government response than respondents who live in areas with a lower African American population. In particular, those living in areas with a higher black population, having more familiarity with African Americans, might have more empathy for the media-portrayed Black victim of Katrina (Gay, 2004; Welch, Sigelman, Bledsoe, & Combs 2001). We captured this concept by including a contextual variable provided by Gallup that is simply the percentage of the population in the respondent's zip code area that is African American.

Given the disheartening and often tragic images of Katrina victims in the media, we expect that respondents who closely followed the news about Katrina and the aftermath would have more negative feelings about government response, regardless of race.3 To account for this, our models include respondents' answers to the following question: "How closely have you been following the news about Hurricane Katrina and the subsequent flooding in the city of New Orleans: not at all, not too closely, somewhat closely, very closely?"4

Likewise, we expected that individuals who had more emotional responses to Katrina might have more negative feelings about government response. To capture this potential impact, we constructed a scale based on the sum of yes responses to the following: "Please say whether you, personally, have felt each of the following emotions in response to Hurricane Katrina and its aftermath, or not How about Shock, Sadness, Anger?" Respondents could indicate yes or no for each. In preliminary analysis, we estimated our full models using this scale of emotion question as a dependent variable to determine if there were significant racial differences in the emotional response to Katrina. The results (not shown) suggested that African Americans did not have a significantly different emotional response, but that young people, women, Democrats, and those who paid the most attention to the news about Katrina were more likely to indicate they had a strong emotional response.

Finally, we include a number of control variables in each of our models. Based on previous research on attitudes about government more generally, we also include controls for gender, proximity to the hurricane,5 living in the South, size of place of residence, and respondent's age, education level, partisanship, income, and church attendance (Chong & Rogers, 2005; Conover, 1984, 1988; Cook & Gronke, 2005; Davis & Brown, 2002; Gay, 2004; Kinder & Winter, 2001; Simien & Clawson, 2004).

Results and Discussion

Racial Differences in Attitudes

Our results testing for attitude differences between Blacks and non-Blacks are displayed in Tables 1 and 2. In each of these models, we examined differences in Katrina-related attitudes between African Americans and those who are not African Americans, controlling for a variety of individual characteristics. In terms of President Bush's response to the storm (column 2) and who was to blame most for the slow and inadequate response (column 3), the results suggest that those who had an emotional response to Katrina, younger respondents, those with more education, those who lean toward a Democrat affiliation, respondents from the South, those who attend church less often, and to some extent, those who paid more attention to the news about Katrina and lived in closer proximity to the storm, were more likely to indicate that Bush's response was poor and that Bush was more to blame than local or state government officials for a poor response. Meanwhile, household income, gender, size of place of residence, and percentage of Black people in the area of residence appears to have little influence on attitudes about Bush's response to the storm.

Most importantly, the results indicate that race played an important role in shaping attitudes. Blacks were more likely than non-Blacks to indicate that President Bush's response to the storm was poor and that President Bush was more to blame than local or state officials in the slow and ineffective response to the storm's aftermath. This finding is consistent with our expectation that news media frames of storm victims as predominately Black likely invoked more empathy among Blacks with the victims and produced more negative responses concerning federal government's efforts to help victims (Table 1).

Given the racialized news media frame of Black hurricane victims and the initial evidence that Blacks tended to blame President Bush for poor government response, we also explored if there were racial differences in beliefs on whether the President cares about Black people (column 1). The results suggest that those who had an emotional response to Katrina, Democrats, those who lived closer to the storm, and those who rarely attend church were more likely to indicate that President Bush does not care about Blacks. Respondent's education, income, gender, age, size of place of residence, and percentage of Black people in the area of residence appears to have little influence on attitudes on whether President Bush cares about Black people. As expected, Blacks were significantly more likely than non-Blacks to believe that the President does not care about Blacks.

Likewise, we expected that the media depiction of Black victims, sometimes in a negative light, would make Blacks less likely to view looters in New Orleans as criminals and instead as desperate individuals trying to survive. The results of this analysis are in column 4. The results confirm our expectations: Blacks were significantly more likely to view looters as desperate individuals looking for food, not criminals, as were respondents who lived outside the South, had a strong emotional response to the storm, those with less income, respondents with more education, and those who leaned toward being Democrat. Clearly, African Americans were more sympathetic with victims than non-Blacks.

Respondents were also asked if government's response was less than adequate because the victims tended to be poor or because they tended to be Black. Our analyses of responses to these two questions (columns 5 and 6) produce results similar to analysis of attitudes toward government response generally. Respondents who responded emotionally to the storm, were younger, lived outside of the South, and had Democratic partisan leanings were more likely to indicate that that government responded poorly because the victims were poor and were more likely to suggest poor government response because the victims were Black. Interestingly, those who paid the most attention to news media coverage of Katrina were also more likely to say that government response was inadequate because the victims were poor and because they were Black. Most importantly, African American respondents were significantly more likely to indicate that the poverty and race of victims was a reason for poor government response. Here our findings on media exposure and race tend to confirm our arguments concerning the racialization of news media coverage of Katrina, and its subsequent impact on the attitudes of those most exposed and Blacks.

Given initial attitudes about government response to Katrina, we can now explore how race might have shaped confidence in the federal government to respond to future events. Table 2 displays the results for questions regarding confidence in the federal government's ability to respond to future natural disasters as well as future terrorist attacks. The results suggest that those respondents who had a high emotional response to the storm, lived outside of the South, had more education, attended church less often, and leaned toward Democratic affiliation were more likely to have less confidence on the federal government's ability to respond to a future natural disaster or a future terrorist attack.

Most importantly, race played a significant role in predicting attitudes. African Americans were less likely to have confidence in the federal government's ability to respond to a natural disaster or a terrorist attack in the future. From this, we can likely infer that the negative attitudes held by African Americans toward federal government's response to Katrina translated into less confidence in the federal government's ability to respond to similar events in the future. Indeed, given the historically low levels of Black confidence in government, the perception of a poor federal government response to Katrina may have served to lower Black confidence in government even further.

To obtain a clearer picture of the substantive effects of race on attitudes following Katrina, we estimated predicted probabilities for all of the independent variables in the models with dichotomous responses. The predicted probabilities are displayed in Table 3. The results demonstrate the powerful influence of race on attitudes. For example, being Black increases the probability of blaming President Bush more than local officials by almost 13 percent and increases the probability of believing that Bush does not care about Black people by 42 percent In comparison, a standard deviation increase in partisanship (from Republican toward Democrat) increases the probability of blaming President Bush more than local officials by almost 11 percent, and increases the probability of believing that Bush does not care about Black people by almost 27 percent. Meanwhile, a standard deviation increase in emotional response to the storm only increases the probability of blaming President Bush more than local officials by 5 percent and increases the probability of believing that Bush does not care about Black people by almost 5 percent Thus, overall, the effect of race on attitudes following Katrina was relatively substantial and often more significant than even a change in respondent partisanship.

In sum, our analyses support our original hypothesis that African Americans would have different views from non-African Americans regarding the aftermath of Katrina. Recall that in part, we believe that the difference in attitudes is based on the fact that media coverage of the storm's aftermath in New Orleans tended to show African American victims, sometimes in a less than positive light Given the African American "face" of victims, we expected Blacks to more closely identify with the victims, and therefore, be more negative toward government response. Our results support these arguments.

Alternative Explanations

Although our findings are consistent with our expectations and previous studies on attitudes in the aftermath of Katrina (Huddy & Feldman, 2006), some observers might argue that there are alternative explanations for the racial differences in attitudes that we observe. Although the differences clearly do not simply reflect racial differences in income or partisan identification, one explanation for our findings that race shaped attitudes about government response to Katrina is that Blacks, having lower trust and confidence in government institutions generally, would consistently hold more negative views of government response to virtually any disaster.

Few studies have been conducted on this topic, but two studies are illustrative. In one early study on the political impact of hurricane flooding on an election, Abney and Hill (1966) found that predominantly Black precincts in New Orleans were no less likely to vote against the incumbent, who was criticized for his response to the flooding, than predominantly White precincts. These results held even when limiting the analysis to those precincts that faced the worst flooding and when the authors accounted for differences in socioeconomic status across precincts. Likewise, Arceneaux and Stein's (2006) analysis of the impact of flooding from Tropical Storm Allison on the 2001 Houston Mayoral election found that Blacks were actually more likely to support the incumbent mayor, even if they blamed local officials for flooding problems following the storm. The findings of both studies indicate that Blacks do not automatically view government response to storms more negatively than Whites.

In addition, Katrina was of course not the first disaster President Bush faced, nor was he the first president to be criticized for his response to a hurricane. To assess whether the opinions of Blacks systematically differ from those of non-Blacks and do not reflect the racialized nature of media coverage of Katrina as we suggested, we briefly explored responses to other disasters. For President George W. Bush, we were only able to locate opinion polls about his response to one other disaster, the terrorist attacks of 9/11. In a September 2001 ABC /Washington Post national poll of adults, respondents were asked: "Do you approve or disapprove of the way George W. Bush is handling the U.S. response to last week's terrorist attacks?" Respondents overwhelmingly approved of Bush's response. Although no observers have seriously suggested that media coverage of the 9/11 attacks provided a racial frame, we compared the approval of Blacks to non-Blacks regarding Bush's response. Blacks were no more likely than non-Blacks to disapprove of the president's response (p < 0.121) even without controlling for respondent partisanship.

But what of past presidential responses to hurricanes? Like his son, President George H. W. Bush was roundly criticized in 1992 for slow and inept federal government response in the aftermath of Hurricane Andrew in Florida (Andrews, 1992). However, the devastation from Andrew was largely in Dade County, Florida and media coverage of the aftermath did not contain significant references to race or ethnicity. Thus, attitudes about President George H. W Bush's response to Hurricane Andrew provides a reasonable comparison case for assessing whether African Americans are simply more likely to rate the president and the federal government negatively following a disaster.

We analyzed data from two CBS/New York Times polls conducted just after Hurricane Andrew. The first survey was a national random sample of 1,329 adults and the second a random sample survey of 1,112 adults in Florida.6 We replicated our analysis of the Katrina poll as closely as possible. Respondents in the Andrew surveys were asked: (i) "Do you approve or disapprove of the way George Bush has handled the response to Hurricane Andrew," and (ii) "Do you think the federal government responded adequately to Hurricane Andrew or do you think it could have done much better." We used the dichotomous responses to these two questions as dependent variables. Our independent variables are similar to those used earlier, including race, gender, education, income, party affiliation, size of place, living in the South, and age. For proximity, we include responses to a question that asked respondents if Andrew directly affected them or someone they knew. For the Florida survey, we also included a dichotomous variable coded 1 if the respondent lived in Dade County and 0 otherwise.7

Analyses of the national survey responses using a probit estimation model reveals that party affiliation and living in the South were strong predictors of approval of the Bush response to Andrew. However, race was not-Blacks were no more or less likely to approve of the Bush response.8 Likewise, in assessing the overall federal government response, Democrats, women, and those from more urban areas were more likely to indicate that the response could have been much better, while Blacks were not significantly more or less likely to indicate that federal government response could have been much better.9 Probit estimations with the Florida poll reveal similar results except that those living in Dade County or who were directly affected or had relatives who were affected by the storm were somewhat more likely to disapprove of Bush's response and suggest that the federal government could have done much better. Blacks were no more or less likely to approve of the Bush response and no more or less likely to indicate that the federal government could have done much better.10

Thus, our analyses and previous research suggests that race does not automatically shape perceptions of government response to disasters. Indeed, attitudes about government response to disasters are shaped by a broader context, including the manner in which the news media or political elites frame the disaster. As such, our evidence of media framing Katrina in terms of race, and a subsequent significant role in race-shaping attitudes about government response to the storm, are not easily disputed by alternative explanations.

Conclusion

In the aftermath of Hurricane Katrina, many Americans held negative perceptions about government response to the storm. African Americans seemed to have especially negative attitudes about federal government's actions and different beliefs about why government response was slow. We employed framing theory to examine the notion that news media racialization of Katrina invoked an attitude difference between Blacks and non-Blacks about government response and victims following the storm. We hypothesized that the dominant media frame of Black storm victims and the media's portrayal of African Americans in a negative light, led African Americans to be more empathetic with storm victims. This identification led African Americans to more often hold negative views about government response to the storm and hold more positive views about storm victims. We tested these hypotheses using a unique national poll of adults, with an over-sample of African Americans, conducted in September 2005. Our findings allow us to draw several important conclusions.

First, our results clearly support our argument that African Americans held distinctly different views about government response to Hurricane Katrina, even controlling for partisanship, income, and education, among other factors. In particular, Blacks were more likely than non-Blacks to believe that the federal government had done a poor job responding to the storm. Much of this blame was placed at the feet of President Bush. Black respondents were also more likely to believe that the government's slow response in the aftermath of the storms was partly a function of victims being poor and Black. Indeed, the substantive impact of race on attitudes was greater than even partisanship in predicting respondent attitudes.

Second, and perhaps the most problematic finding for policymakers, is that African Americans had less confidence in the federal government's ability to respond to future natural disasters and terror attacks following Katrina. Given the historically lower levels of confidence in government held by African Americans, it seems especially troubling that the aftermath of Katrina's devastation includes a lowering of Black's confidence in government. Moreover, 2006 polls suggest that the decline in confidence may not be short term. A February 2006 AP-Ipsos poll of national adults suggests that Blacks still have more negative feelings about government response to the storm, are more likely to say that not enough is being spent on the recovery, and are less likely to express confidence on how the government would handle a future disaster (McClam, 2006).

Third, although we cannot definitively establish that media framing of Katrina in terms of race leads to the development of a stronger affinity between African-Americans and the victims of the storm, our results lend support to this argument We provided clear evidence that media coverage of Katrina was racialized and we clearly established dramatic racial differences in attitudes. In addition, through analysis of attitudes about government response to previous disasters, we were able to establish that Blacks are not always more or less negative about government response to disasters and that attitudes about government response are highly dependent on the broader context in which a disaster is perceived.

Nonfindings in our results also support this contention. For example, although most media images of storm victims were of low-income people in New Orleans, in our models, personal income level played almost no role in shaping opinions about government response to the storm. Thus, it seems less likely that low-income individuals developed the same type of connection to storm victims as African Americans appear to have developed.

Finally, as a whole, our results lend support to linking framing and group membership together as a tool for understanding political attitudes. We believe that our findings add an additional dimension to past research and we encourage further research along these lines. We further believe that focus should be placed on media framing to better understand how marginalized groups in society come to develop less positive views of government, policy, and the political system. Likewise, research should more fully explore how race shapes not only confidence in government, but how race might also shape beliefs about the legitimacy of government policy and the political system more generally.