The riot as politics
THE RIOT AS POLITICS Devin Bent In the 1960s America's cities were shaken by a series of urban riots. The most severe disorder that took place in Detroit in the summer of 1967 resulted in more than 40 deaths--primarily blacks at the hands of the authorities--and in excess of $40 million in property damage. The major riots provoked a number of attempts to explain intercity differences in riot experience. These studies generally agree that region and total population (or numerical black population) are important in the explanation of inter- city differences--i.e., that riots are more likely to occur, and to be severe, in northern cities with large numbers of people (or Blacks in particular). Beyond this there is little agreement among the studies, although perhaps the modal finding is that controls for region and total population negate the explanatory power of any other variable. Explanations of riot variation based on political structure, per capita spending, relative deprivation, ab- solute deprivation, social disorganization, and segregation all fail of con- firmation, i This study will offer a political explanation of the major urban riots of the 1960s. It is argued that the grievances underlying the major riots of 1964- 68 were not directly produced by the problems, be they conceptualized as relative deprivation, social disorganization, or what-have-you. Rather what is required for a major riot--an uprising of the nature of Watts, Detroit, or Newark--is the belief among the politically oriented segments of the black population that the government will not respond to demands for change. 2 I am not arguing here that all of the rioters were politically motivated. Obviously the rioters may have been individuals of varying degrees of political interest and motivation. Robert M. Fogelson, in his analysis of arrest records from 19 cities, found that the arsonists were the most likely to be under 25, employed, skilled, and born in the North-- 478 The Review of Black Political Economy factors generally associated with militance in the black community. The looters were the most likely to be over 25 and unemployed, and second most likely to be unskilled and born in the South--a pattern that suggests greater economic motivation. The assaulters--throwing stones, obstruct- ing policemen, and carrying weapons--fell between the two extremes.3 A major uprising requires the support of the politically oriented rioters, who would turn to arson or battling with police rather than looting and would thus play a major role in breaking the control of the authorities. As long as the politically oriented segments of the black community refrain from participation, the ingredients for a major upheaval are missing.' It might have been possible prior to the riots to survey black attitudes toward local government responsiveness in a set of cities and then correlate attitudes with eventual riot experience. In practice, of course, surveys of this nature were conducted after the riots had begun. While these surveys did find greater dissatisfaction in riot cities, it seems likely that these attitudes were as much a product of the riots as a cause. 5 Thus, for this study it will be necessary to infer the attitudes from the actual public policies of the cities--i.e., to assume that Blacks have more confidence in local government responsiveness when local government has actually re- sponded to their demands for change. While it may seem questionable to infer black attitudes toward local government response from the actual response, certainly the inference is as defensible as the common practice of inferring relative deprivation or grievance levels from census data. A question that arises, however, is: Does greater response lessen disaf- fection or might it actually increase disaffection? One hypothesis, associ- ated with Alexis De Tocqueville, is that public policy response generates higher levels of expectations and thus eventuates in greater dissatisfaction.6 This latter hypothesis is based explicitly on the notion that government action or inaction will significantly influence the level of expectations. This may be true in many instances, but it was probably not true of large cities and their black citizens in the 1960s. It seems more reasonable to assume that the expectations of Blacks were influenced by forces that were national or international in scope--the activities of the civil rights move- ment, the policies of the national government (including the Supreme Court), and the independence of African nations. If black expectations were not significantly influenced by local public policy, then local government response could only increase satisfaction by narrowing the gap between expectations and performance. Thus, the hy- pothesis of this study is that the more responsive a city to black demands the less likely it was to have a major riot (assuming that factors such as region THE RIOT AS POLITICS 479 and city size are controlled). The hypothesis will be tested twice with two groups of cities: first, the nine non-southern municipalities with the numer- ically largest black populations, and second, an additional set of 27 cities. The independent variable will be the actual public policy response of the cities to a set of black demands. The nine cities will be ranked on eight public policies that are in sufficient agreement to provide the basis for a summary measure of response. Four of the eight policies are in particularly close agreement and provide the basis for a second summary measure of response for the nine cities. The same four policies are then used to estab- lish a similar summary measure of public policy response for the 27 cities. The dependent variable is riot magnitude using the best available data for each set of cities. It might seem reasonable to combine the two sets of cities into one; however, the two sets will be kept distinct for two reasons. First, the data are much better for the nine non-southern municipalities with the largest black populations. Combining the two sets would necessitate use of the less adequate data of the second set of cities. Second, as I have already suggested, there may be different types of rioters distinguished by their motivation. Similarly, there may be different types of riots. A number of observers have suggested riot typologies, and some have critiqued riot studies for their failure to distinguish between types. 7 Different types of riots may have different correlates; thus grouping all types together would blur relationships. The nine municipalities experienced all but one of the major uprisings of the 1960s." Separating out the nine allows a focus on this particular type of riot. THE CITIES STUDIED The first set comprises the nine non-southern municipalities with the numerically largest black populations (see Table 1). The nine range in 1960 total population from 7,800,000 (New York City) to 400,000 (New- ark), and in black population from 1,100,000 (New York City) to 140,000 (Newark). The 27 cities of the second set were selected using five criteria: (1) a 1960 population in excess of 50,000; (2) a black proportion greater than 10% and less than 45 %; (3) municipal authority to enact civil rights ordinances and to construct public housing; (4) mayor-council or council- manager form of government; and (5) location in the northeast quadrant of the United States. The end product is a set of 27 cities located in a nearly contiguous belt of states--Connecticut, New Jersey, Pennsylvania, Ohio, Indiana, Michigan, and Illinois. They range in total population from 48O The Review of Black Political Economy TABLE 1 Summary Response Rankings City Eight Central Po l ic ies Cluster Phi ladelphia 1 1.5 New York 2 1.5 Chicago 3 5 Baltimore 4,5 3 St. Louis 4.5 6 Newark 6 7 Los Angeles 7 8 Cleveland 8 4 Detroit 9 9 604,000 to 51,000 and in black population from 109,000 to 5,400. INDEPENDENT VARIABLE: PUBLIC POLICY RESPONSE The Nine Cities Eight measures of policy response to black demands have been devel- oped. Each measure ranks the response of the cities in the 1960s to a discrete controversial policy claim raised by Blacks and their allies. The cities are ranked by their response to the following demands: (1) a broad and enforceable fair employment ordinance; (2) a broad and enforceable fair housing ordinance; (3) a legally empowered, durable civilian review board; (4) the construction, acquisition, or leasing of public housing (not including that designated for the aged); (5) the hiring of black police officers in proportion to black population; (6) the promotion of black officers to the rank of captain and higher in proportion to black population; (7) target area control of the Community Action Program (CAP); and (8) the sparing or nondestmction by the city's urban renewal program of the housing of non-white families. Summary titles and dates of the eight rankings are shown in Table 2. The eight measures are the dependent variable of an earlier article and will not be individually discussed here. 9 THE RIOT AS POLITICS 481 TABLE 2 Policy Measure Intercorrelations Fa i r Employment (1965) - - Fa i r Housing (1965) .64" - - C iv i l i an Review (1965-8) .71" .55 - - Pub l i c Housing (1962-5) .61" ,82"* .68* -- . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Police Hiring (1967) Police Promotion (1967) Control of CAP (1967) Nonwhite Housin~ Spared ~1962-5~ Significance levels: ** .01; Central Cluster .39 .63" .32 .55 - - .61" .30 .29 .33 .64" - - .07 - .16 .51 .13 - .13 .01 - - .06 ~19 134 ,23 .38 .02 - .02 * .05 (N=9). Note: The cor re la t ion coe f f i c ient i s Spearman 's rho cor rec ted fo r t ies . resu l t s a re obta ined us ing the Pearson product -moment coe f f i c ient . Similar Summary Measures of Response In order to develop a summary measure of response it is necessary to demonstrate that there is some agreement among the public policies. The Kendall coefficient of concordance (W) is a measure of agreement among rankings that ranges from zero to unity as agreement increases.'~ For the eight policies, W equals 0.42, which is significant of the 0.001 level (X~ -- 26.9, df ---- 8). Since W is significant, it is possible to rank the cities on response, ranking according to the sum of the ranks allotted each city. The eight policy summary ranking is shown in Table 1. Another way to examine agreement is to compute the correlation coeffi- cients between all possible pairs of policy measures. The coefficients-- using the Spearman rank correlation coefficient (r,)--are shown in Ta- ble 2." Four of the eight policies have been identified as a central cluster. All six correlations among the four policies are 0.55 or greater, and five of the six are significant at the 0.05 level. A summary ranking of public policy response based on the central cluster is also presented in Table 1. 482 The Review of Black Political Economy The Twenty-Seven Cities Data were not available for all eight public policies for the second set of cities. However, data were available for the central cluster policies, and thus a single summary measure of central cluster policy response was developed. The four central cluster policies were dichotomized because a large number of ties renders the ranking procedure unsuitable. The dichot- omies employ the same criteria as the rankings; thus if city A is classified as high response and city B as low response, then city A would rank higher than B. For central cluster summary response, the cities are rated from zero to four according to the number of policies for which they have been classified high response. DEPENDENT VARIABLE: RIOT SEVERITY The Nine Cities The measure of riot magnitude or severity proved more difficult to construct than expected. The most accurate available information mea- sures official response to the riot rather than the activities of the rioters themselves. Arrest statistics, or the use of National Guard or regular U.S. Army troops, may to some extent reflect riot magnitude, but they also are influenced by the strategy of public officials. The number of arrests made and troops used in Washington in 1968 exceeded that for any other riot even though most observers would agree that the Detroit riot of 1967 was larger. The massive use of arrests and manpower was the strategy of Cyrus Vance, derived from his experiences in Detroit the previous summer. ~2 The number of deaths is also a measure of response, or perhaps "counter-riot," since most deaths were caused by law enforcement officers or National Guard troops. In Los Angeles, for instance, coroner's inquests were held on 32 deaths connected with the Watts riot. Twenty-six were ruled justifiable homicide, of which 23 were attributed to police or National Guardsmen..3 A measure of riot severity should reflect the activities of the rioters-- activities that included the looting and burning of buildings, primarily retail stores. Thus, for the nine cities, the measure of riot severity is based on estimated property damage, which encompasses damage caused by both looting and arson. The estimates were prepared by the American Insurance Association (AIA). ,4 The advantage in using the AIA estimates is that they come from a single source, and thus were prepared by a more or less uniform method using insurance industry data. The disadvantage is THE RIOT AS POLITICS 483 that these estimates measure insured damages and not total damages: what appear to be differences in riot magnitude may actually be differences in insurance coverage. However, no other single source of damage estimates provides reliable data for all nine cities, and many of the alternative sources ultimately rely on newspaper estimates--estimates that have often proven to be wildly inflated. ~5 For each city, the estimated property damage of the largest riot was divided by the numerical black population. Each city is thus ranked by estimated insured property damage per black person (see Table 3). The data reveal that there was considerable variation in riot experience among these cities: damage estimates range from $91 per black person to less than $2. The ranking also contains a few surprises. While Detroit had the most total destruction, Los Angeles ranks first on a per capita basis. Several riots that dominated media attention at the time--the Harem and Bedford- Stuyvesant riot of 1964 and the Hough (Cleveland) riot of 1966--do not appear on the list. Both of these riots were relatively minor outbreaks that would have received very little media attention had they occurred in the summer of 1967 or the spring of 1968. The 2 7 Cities With few exceptions it proved impossible to obtain reliable property damage estimates for the smaller cities. Moreover, it appears that none of these cities had sufficient black population to support a major uprising of the nature of Detroit, Watts, or Newark. However, it is possible to argue that any riot that occurred in these cities was more likely to become serious if it had the support of the politically oriented segment of the black commu- nity.'6 Thus, the more responsive cities should have fewer serious riots than the less responsive cities. Two measures of riot experience were developed for the 27 cities. First, each city is rated according to the number of riots, both serious and minor, occurring from January 1964 through August 1968. A number of sources were consulted, and any listed riot, disorder, or civil disturbance was counted as a riot--with no double counting, of course. The courses overlap in their time periods so that the entire period with the exception of August 1968 is covered by at least two sources.'7 The Lemberg Center for the Study of Violence was not directly utilized, but its data were used by two of the other sources. There were a total of 60 riots (broadly defined) in the 27 cities. Second, each city is rated by the number of serious riots during the same 484 The Review of Black Political Economy TABLE 3 Riot Magnitude Insured 1965 Black Insured Date Property Populat ion Property City of Pamage In terpo la ted Damage Rank Riot in in per ~llions Tho1~ands ~lac~ of Person Dol lars Los Angeles Aug. 6S 38.0 419 $91 1 Detroi t Ju ly 67 41.5 571 73 2 Newark Ju ly 67 11.0 173 64 3 Baltimore Apr. 68 12.0 373 32 4 Chicago Apr. 68 13.0 958 14 S Cleveland Ju ly 68 1.5 269 5.6 6 Nq:~,'.York Apr, 68 4.2 1,377 3.0 7 St. Louis Apr. 68 0.5 234 2.1 8 Phi ladelphia Aug. 64 1.0 592 1.7 9 Sources: see text . time period (January 1964 through August 1968). A serious riot is one that meets the test developed in a staff study for the Permanent Subcommittee on Investigations of the U.S. Senate Committee on Government Opera- tions: " I f any community reported a riot-connected death, it was automati- cally included. Otherwise, two or more of the following criteria had to be met for the incident to be included: (l) two or more injuries; (2) sniping; (3) looting; (4) 20 or more fires; and (5) 50 or more arrests. '''8 The staff study was used as the source of the serious riot data except for 1964 and August 1968, which are not included in the staff study. The other sources were consulted for these periods uncovering two riots that met the criteria, and resulting in a total of 24 serious riots in the 27 cities during the period January 1964 through August 1968. There are several riots listed in other sources for the period January 1965 through July 1968 that appear to satisfy the criteria, yet are not included in the subcommittee's list. These riots all occurred in low response cities and are not counted as serious riots for this study. Thus, omissions from the subcommittee's list would seem to work against the hypothesis. The advantage of relying upon the subcom- mittee's data is that they come primarily from a survey of mayors and thus avoid some of the problems associated with newspaper riot data. ,9 THE RIOT AS POLITICS 485 TABLE 4 The Nine Cities: Correlates of Riot Magnitude Stmmmry Public Policy Response Population Variables Eight Central Total Ntmmrical Policy Cluster Population Black Population 9 70" -.75"* -.18 -.03 Significance levels: *.05; **.01. Note: the correlation coefficient is Spearman's rho corrected for ties. Similar results are obtained using the Pearson product-moment coefficient. FINDINGS For the nine cities the relationship between public policy response to black demands and per capita riot damages is quite strong and in the hypothesized direction: --0.70 for eight policy summary response and --0.75 for central cluster summary response (Table 4). For these large non-southern cities, public policy response and relative freedom from violence are definitely associated. Since riot magnitude is a per black person measure, it was not expected that total population or numerical black population would be closely related to riot magnitude and they are not: the relationships are weak and in the wrong direction (Table 4). For the 27 cities, however, the number of riots and the number of seious riots are not per capita measures, and from all previous studies both should be related to total population and numerical black population. In fact, both population variables are positively related to both riot measures, although only one of the four zero-order correlations are significant at the 0.05 level (Table 5). 20 Once central cluster response is controlled, however, then three of four partials are significant (Table 5). Policy response is negatively associated with frequency of (all) riot(s) once either total population or numerical black population is controlled, but the partial coefficients are not significant at the 0.05 level (Table 6). However, public policy response and frequency of serious riot are more 486 The Review of Black Political Economy TABLE 5 The 27 Cities: Population Variables and Number of Riots Ntrnber of Riots Population Variables Total Population Nt~nerical Black Population partial a r part ia l a r r All Riots .27 .33 .35 * .41" Serious Riots .28 .42* .29 .40* Significance level: *.05. Note: the Pearson product-moment coefficient is used in this table, acontrolling for central cluster stmmmry response. closely associated: the partials, controlling either population variable, are significant at the 0.05 level (see Table 6). Thus, the hypothesis is con- firmed for the 27 cities: riots occur for a variety of reasons, but are more likely to become serious in the less responsive cities. CONCLUSION Two independent tests with a total of 36 large and medium-sized non- southern cities have demonstrated support for a political explanation of the major riots of the 1960s. It is argued that politically interested and moti- vated segments of the black community were disaffected because of the failure of government to respond to their demands. Their participation turned minor disorders into major uprisings. An alternative hypothesis, that public policy response generates increased expectations and ultimately greater violence, has been rejected. However, the findings rest upon the riot experience of a small number of cities in the 1960s. It seems very unlikely that the public policies of these cities would have much impact on the expectations of their black citizens and thus the rationale of the alterna- tive hypothesis is undermined. The public policies of other governments, at other times, might have more influence on the expectations of their THE RIOT AS POLITICS 487 TABLE 6 The 27 Cities: Central Cluster Public Policy Ntmber of Riots Central Cluster Response par t ia l r Controll ing for Total Population Numerical Black Population All Riots -.06 -.20 -.23 Serious Riots -.20 -.37" -,35" Significance level: *.05. Note: the Pearson product-moment coeff ic ient is used in th is table. citizens. Additional research would be required to identify circumstances under which public policy response might lead to increased expectations, disaffection, and ultimately greater violence. NOTES 1. William R. Morgan and Terry N. Clark, "The Causes of Racial Disorders: A Griev- ance Level Explanation." American Sociological Review 38 (October 1973): 611-24; Seymour Spilerman, "The Causes of Racial Disturbances: A Comparison of Alternative Explanations," American Sociological Review 35 (August 1970): 627-49; Seymour Spiler- man, "The Causes of Racial Disturbances: Tests of an Explanation," American Sociologi- cal Review 36 (June 1971): 427-42; Bryan T. Downes, "Social and Political Characteristics of Riot Cities: A Comparative Study," Social Science Quarterly 49 (December 1968): 504-20; Bryan T. Dowries, "A Critical Reexamination of the Social and Political Charac- teristics of Riot Cities," Social Science Quarterly 51 (September 1970): 349-60; Larry D. Singell, "The Socio-Economic Causes of the Recent Urban Disorders: Some Empirical Evidence," Land Economics 47 (August 1971): 225-34; Jerome L. McElroy and Larry D. Singell, "Riot and Non-riot Cities: An Examination of Structural Contours," Urban Af- fairs Quarterly 8 (March 1973): 281-302; Joel A. Lieske, "The Conditions of Racial Violence in American Cities: A Developmental Synthesis," American Political Science Review 72 (December 1978): 1324-1340; Jules J. Wanderer, "An Index of Riot Severity and Some Correlates," American Journal of Sociology 74 (March 1969): 500-5; William E Ford and John H. Moore, "Additional Evidence on the Social Characteristics of Riot Cities," Social Science Quarterly 51 (September 1970): 339-48; and Robert M. Jiobu, "City Characteristics, Differential Stratification, and the Occurrence of Interracial Vio- lence," Social Science Quarterly 52 (December 1971): 508-20. 2. Cf. Chalmers Johnson, Revolutionary Change (Boston: Little, Brown, 1966), pp. 92- 3. 488 The Review of Black Political Economy 3. Violence As Protest: A Study of Riots and Ghettos (Garden City, N.Y.: Doubleday, Anchor Books, 1971), pp. 43-44, 4. Cf. Edward C. Banfield, The Unheavenly City: The Nature and Future of Our Urban Crisis (Boston: Little, Brown, 1970), pp. 204-5. 5. Howard Schuman and Barry Gmenberg, "The Impact of City on Racial Attitudes," American Journal of Sociology 76 (September 1970): 245. See also Lemberg Center for the Study of Violence, "Six-City Study, A Survey of Racial Attitudes in Six Northern Cities: Preliminary Findings," Waltham, Mass.: 1967 (mimeographed). 6. The Old Regime and the French Revolution, trans. Stuart Gilbert (Garden City, N. Y. : Doubleday, Anchor Books, 1955), pp. 176-7. See also Banfield, pp. 205-6; and Bernard N. Grofman and Edward N. Mueller, "The Strange Case of Relative Gratification and Poten- tial for Political Violence: The V-curve Hypothesis," American Political Science Review 67 (June 1973): 514-39. 7. Banfield, pp. 184-209; Gary T. Marx, "Issueless Riots," TheAnnals 391 (September 1970): 21-33; Louis C. Golderg, "Ghetto Riots and Others: The Faces of Civil Disorder in 1967," in Cities Under Siege: An Anatomy of the Ghetto Riots, 1964-8, ed. David Boesel and Peter H. Rossi (New York: Basic Books, 1971), pp. 138-45; Peter A. Lupsha, "On Theories of Urban Violence," Urban Affairs Quarterly 4 (March 1969): 272-95; George Rude, The Crowd in History: A Study of Popular Disturbances in France and England, 1730-1848 (New York: Wiley, 1964), pp. 214-36; and Neii J. Smelser, Theory of Collective Behavior (New York: Free Press, 1963), pp. 257-61. 8. The exception is Washington, D.C., which experienced a major uprising in the spring of 1968, but has not been included because it lacked the local self-government characteristic of the other cities. 9. Devin Bent, "Partisan Elections and Public Policy: Response to Black Demands in Large American Cities," Journal of Black Studies (September 1981), forthcoming. 10. Maurice G. Kendall, Rank Correlation Methods (New York: Hafner, 1962), pp. 94-103. 11. Spearman's rho has been corrected for ties. For the larger study, the Pearson product-moment coefficient has also been computed and supports the results reported here. 12. Ben W. Gilbert and the staff of the Washington Post, Ten Blocks from the White House: An Anatomy of the Washington Riots of 1968 (New York: Praeger, 1968), pp. 32-5, 118-9. 13. California Governor's Commission on the Los Angeles Riots, Violence in the City-- An End or a Beginning? (1965), p. 20, 14. Although the estimates were prepared by the American Insurance Association, they were obtained from a variety of sources: Insurance Information Institute, "News from Insurance Information Institute," 15 May 1968 press release (New York); Insurance Infor- mation Institute, Insurance Facts, 1970 (New York: Insurance Information Institute, 1970); Wall Street Journal, 17 April 1968,p. 17, andl6 May1968, p. 6; Wilfred J. Perry, Executive Assistant, Property Claims Services, American Insurance Association, personal letter, 24 February 1972. Philadelphia presented special problems. The largest riot was apparently the first one, in August 1964. No property damage estimate could be obtained from the American Insurance Association for this riot. However, the estimated property damage for all riots in 1964 was $2.5 million (Perry, letter). According to the California Governor's Commission, five persons were killed in the riots of 1964, 952 were injured, 2,484 arrested, and 1,080 stores were damaged. Of these, none of the deaths occurred in Philadelphia, 341 or 36 % of the injuries, 774 or 31% of the arrest and 225 or 21% of the stores damaged. From this, it would seem reasonable to assign an insured damage of $1.0 million (40%) as an upper limit estimate. 15. National Advisory Commission on Civil Disorders, Report (New York: Bantam Books, 1968), p. 115. THE R IOT AS POL IT ICS 489 16. Cf. Morgan and Clark, pp. 614-5. 17. California Governor's Commission, p. 2; Congressional Quarterly Weekly Report, 8 September 1967, pp. 1709-12; Permanent Subcommittee on Investigations, U.S. Senate Committee on Government Operations, Staff Study of Major Riots and Civil Disorders (Washington, D.C.: 1968); International City Managers Association, Urban Data Service 1 (January 1969): 23-31; National Advisory Commission on Civil Disorders, pp. 112-113. 18. Ibid., p. 2. 19. Ontheuseofnewspaperdata, seeM. Herbert Danzger, "Validating Conflict Data," American Sociological Review 40 (October 1975): 570-84. 20. In Tables 5 and 6, I have used the Pearson product-moment coefficient because it has a significance test for the partial. Spearman's rho actually reports a stronger relationship between response and frequency of serious riot, but it is not used here because it lacks a significance test for the partial.