THE J-CURVE THEORY AND THE BLACK URBAN RIOTS: AN EMPIRICAL TEST OF PROGRESSIVE RELATIVE DEPRIVATION THEORY*

• Document Type: CREST
• Collection: General CIA Records
• Document Number (FOIA) /ESDN (CREST): CIA-RDP86B00985R000100070018-6
• Release Decision: RIFPUB
• Original Classification: K
• Document Page Count: 19
• Document Creation Date: December 16, 2016
• Document Release Date: June 22, 2005
• Sequence Number: 18
• Publication Date: September 1, 1977
• Content Type: PAPER

`r/
Approved Foelease 2005/07/01 : CIA-RDP86B00985R000100070018-6
Reprinted from THE AMERICAN POLITICAL SCIENCE REVIEW
Vol. LXXI, No. 3, September, 1977
The J-Curve Theory and the Black Urban Riots:
An Empirical Test of Progressive Relative Deprivation Theory
ABRAHAM H. MILLER
University of Cincinnati
Louis H. BOLCE
Fordham University, Lincoln Center
and
MARK HALLIGAN
Northwestern University Law School
Introduction
In the intellectual quest for a systematic
understanding of the etiology of political and
collective violence, relative deprivation theory
has emerged as a prevalent explanatory device.
One form of relative deprivation theory has
received particularly strong scholarly attention
through theoretical elaboration and empirical
application to a variety of significant episodes
of political and collective violence. This form is
the progressive variety of relative deprivation
and is most commonly referred to as the
"J-curve.
"1
*This research was funded by the University of
Concinnati Research Council through an award to
Professor Miller. The data were obtained from the
Inter-University Consortium for Political Research
through the facilities of the University of Cincinnati
Behavioral Sciences Laboratory. The authors alone are
responsible for the use and interpretation of the data.
At the time of this study, Louis H. Bocce was a Charles
Phelps Taft Graduate Fellow in Political Science and
Mark Halligan was an Undergraduate Fellow in Politi-
cal Science. William Klecka and Alfred Tuchfarber,
University of Cincinnati, Ted Robert Gun, North-
western University, and four anonymous reviewers
contributed useful and insightful suggestions which are
gratefully acknowledged.
I James C. Davies, "Toward a Theory of Revolu-
tion," in Studies in Social Movements, ed. Barry
McLaughlin (New York: The Free Press, 1969), pp.
85-109. Hardly an anthology on revolution or social
movements exists without including this work. James
A. Geschwender, "Explorations in the Theory of
Social Movements and Revolutions," Social Forces 47,
2 (December, 1968), 127-135, has attempted to
elaborate the theory in terms of cognitive dissonance.
The J-curve resulted from an attempt by
James C. Davies to reconcile what he perceived
as two antithetical explanations of the phenom-
enon of revolution. According to Davies, two of
the most famous students of revolution, Karl
Marx and Alexis de Tocqueville, had arrived at
completely opposite conclusions about the
causes of revolution. Drawing on Marx's work
in The Communist Manifesto, Davies argued
that Marx perceived revolutions as most likely
to occur when things get worse, when the
misery of the proletariate increased relative to
the economic standard of living of the bour-
geoise. In contrast, Tocqueville had advanced
the theory that revolutions occurred when a
previously oppressive regime released the yoke
and 'created expectations it could not fulfill.2
Davies attempted to reconcile these diver-
gent explanations by arguing that each of these
perspicacious students of revolution had ac-
curately rendered part of the picture of revolu-
2Alexis de Tocqueville, The Old Regime and the
French Revolution, translated by Stuart Gilbert (Gar-
den City, N.J.: Doubleday Anchor, 1955), p. 177.
Tocqueville's observation that those parts of France
which experienced the greatest degree of positive
change also showed the greatest popular discontent is
sometimes referred to as Tocqueville's paradox. Two
different explanations of the paradox exist. The most
common, and the one used by Davies and in this
research, is the rising expectations thesis. A less
well-known interpretation is referred to as "the
present value of the past." This argues that present
achievement is devaluated by the perception of~ past
costs. See, in this regard, Charles Wolf, Jr., "The
Present Value of the Past," mimeo (Santa Monica: The
Rand Corporation, 1969).
Approved For Release 2005/07/01`!!'IA-RDP86B00985R000100070018-6
 Approved For Release 2005/07/01 : CIA-RDP86B009R000100070018-6
%AW
1977 The J-Curve Theory and the Black Urban Riots 965
tion, but not the total picture. The complete
depiction of the causes of revolution awaited
the fashioning of both components into a single
entity, which Davies engagingly accomplished.
Davies argued that revolutions occur when a
period of progress is followed by a period of
sharp reversal and decline.
This rise-an.d-decline phenomenon can be
read as: first Tocqueville then Marx. As Davies
notes:. "Revolutions are most likely to occur
when a prolonged period of objective economic
and social development is followed by a short
period of sharp reversal. The all-important
effect on the minds of people in a particular
society is to produce, during the former period,
an expectation of continued ability to satisfy
needs - which continue to rise - and, during
the latter, a mental state of anxiety and
frustration when manifest reality breaks away
from anticipated reality."3 The theory is de-
picted in Figure 1 below.
3Davies, "Toward a Theory of Revolution," p. 86.
At a conceptual level the theory has a great
deal of appeal. It is the type of theory that is
very attractive to social scientists. It reconciles
apparent contradictions in the work of two
leading theorists and creates a parsimonious
explanatory structure. In the search for con-
ceptual simplicity, clarity and phenom-
enological regularity, it yields, at a minimum,
heuristic success and beyond that has had
ostensible empirical corroboration.
From the perspective of empirical theory,
however, there are several primary empirical,
methodological, and conceptual problems in-
herent in the structure and application of the
J-curve. These problems have consequences for
the J-curve theory specifically and relative
deprivation theory generally.
It is the purpose of this research to examine
some of these problems in the context of the
application of the J-curve to the black urban
riots. Although the primary objective of the
research is to test the J-curve theory, considera-
tion will also be given to the substantive
EXPECTED NEED SATISFACTION
U)
w
W
z
1956
ACTUAL NEED
SATISFACTION
1
A TOLERABLE GAP
AN INTOLERABLE
GAP BETWEEN WHAT
PEOPLE WANT AND
WHAT THEY GET
BETWEEN WHAT PEOPLE
WANT AND WHAT THEY GET
1964
TIME
1968
Figure 1.
Pictorial Representation of Davies's J-Curve
(Adapted from James C. Davies, "The J-Curve...," p. 691)
Approved For Release 2005/07/01 : CIA-RDP86B00985R000100070018-6
 jow~
Approved For Release 2005/07/01 : CIA-RDP86B00985R100070018-6
966 The American Political Science Review
implications of the analysis as it affects our research designs based on relative deprivation
understanding of the riots. theory have played a prominent role in the
Seemingly, a number of different events attempts to explain these events. The results of
could have been chosen to examine the theo- these studies, in making a case for relative
retical and empirical application of the J-curve. deprivation as. a primary explanation of the
The black urban riots were chosen, in part, riots, have been largely disappointing, despite
because Davies, himself, has applied his theory some rather sophisticated and well-conceived
to this set of events.4 Consequently, it is methodological procedures.8 As McPhail notes
possible to follow his operationalizations, ". . . there is considerable reason for rejecting
avoiding questions of operational comparabili- the sociological and popular cliche that abso-
ty, and to assess the theory in an appropriate lute or relative deprivation and the ensuing
empirical setting. Moreover, one of the major frustration or discontent or despair is the root
criticisms which we level against the application cause of rebellion."9 On the basis of his
of the J-curve, and most relative deprivation observations, McPhail suggests reconsideration
theories, is the problem of reductionism, some- of Guff's assertion that relative deprivation is
times referred to in the philosophy of science as ". . . as fundamental to understanding civil
the level-of-analysis issue: the validity of using strife as the law of gravity is to atmospheric
data and concepts developed at one level of physics......10
analysis to test theories formulated at a differ- Davies, in his assessment of the urban riots,
ent level.5 (This problem is comparable to the strikes a more confident note: "The Negro
one raised by W. S. Robinson's6 criticism of the rebellion appears to have been preceded by the
use of aggregate correlations to draw inferences same J-curve of expectations that are first
concerning the behavior of individuals.) A gratified and then frustrated."11 In contrast to
variety of data exists for the study of the black Davies's work, the studies cited by McPhail are
urban riots, making it possible to conduct generally based on analyses of socioeconomic
analytical comparisons and reconstruct oper- and political characteristics of riot versus non-
utionalizations at various analytical levels. As a riot cities. Relative deprivation is usually rnea-
result, the level-of-analysis problem can be sured in these studies by comparing various
empirically assessed rather than simply noted as socioeconomic, census data characteristics of
a caveat affecting the research. the black community with those of the white
The Black Urban Riots and community. The underlying logic is that the
Relative Deprivation Theory riot cities should manifest greater disparities
along these dimensions than the nonriot cities.
In his revrew of the sociological literature on Davies's work is based on population charac-
the black urban riots, Clark McPhail7 notes that tarictire? nnnn14-1-
4James C. Davies, "The J-Curve of Rising and
Declining Satisfactions as a Cause of S P.
ea' Sociological Review, 36 (December, 1971),
Revolutions and a Contained Rebellion," in The 1058-1073.
history of Violence in America, ed. Hugh Davis
Graham and Ted Robert Gurr (New York: Bantam 8See, in this regard, Milton Bloombaum, "The
Books, 1970), pp. 690-731. Conditions Underlying, Race Riots as Portrayed by
Sin the philosophy of science this problem is Multidimensional Scalogram Analysis: A Reanalysis of
sometimes alluded to as a problem in reductionism. Lieberson and Silverman's Data," American S'oci-
See, for example, May Brodbeck, "Methodological ological Review, 33 (February, 1968), 76-91; Bryan
Individualism: Definition and Reduction," in Readings T. Downes, "Social Characteristics of Riot Cities: A
in the Philosophy of the Social Sciences, ed. May (D
Comparative ecember, r96Study, 504-520; Stanley Quarterly, 49
Brodbeck (New York: The MacMillian Co., 1968). Arnold Silverman, "The Precipitants Lieberson and
6W. S. Robinson, "Ecological Correlations and the Conditions of Race Riots," AmericandSociological
Behavior of Individuals," American Sociological Re- Review, 30 (December, 1965), 887-898; Seymour
view, 15 (June, 1950), 351-357. See also: L. A. Spilerman, "The Causes of Racial Disturbances: A
Goodman, "Some Alternatives to Ecological Correla- Comparison of Alternative Explanations," American
tion," American Journal of Sociology, 64 (May, Sociological Review, 35 (August, 1970), 627-649;
1959), 610-625; 0. D. Duncan and B. Davis, "An and Jules Wanderer, "An Index of Riot Severity and
Alternative to Ecological Correlation," American Soci- Some Correlates," American Journal of Sociology, 74
ological Review, 18 (December, 1953), 665-666. 0. (March, 1969), 500-505.
1). Duncan, R. P. Cuzzort, and B. Duncan, Statistical q
Geography (Glencoe: The Free Press, 1961), note, as McPhail, "Civil Disorder Participation," p. 1064.
we do, that the problem arises wherever the units of t 0Ted Robert Gurr, "Urban Disorder: Perspectives
analysis are changed. The change need not be from From the Comparative Study of Civil Strife," in Riots
one distinct type of unit (e.g., individual) to a and Rebellion: Civil Violence in the Urban Com-
d i fferent type (e.g., group). munity, ed. Louis H. Mascotti and Don R. Bowen
70ark McPhail, "Civil Disorder Participation: A (Beverly Hills: Sage Publications, 1968), p. 52.
Critical Examination of Recent Research," American 11Davies, "The J-curve," p. 717.
Approved For Release 2005/07/01 : CIA-RDP86B00985R000100070018-6
 Approved Forlease 2005/07/01 : CIA-RDP86B00985R000100070018-6
1977 The J-Curve Theory and the Black Urban Riots 967
The index for measuring relative deprivation is This eclectic interpretation of the data raises
constructed by dividing family income by issues concerning the validity of the J-curve as
average years of schooling for the total and an explanation of the urban riots. Moreover,
nonwhite populations. The ratio of the result- Davies's logic, as that of his compeers in
ing nonwhite quotient to the white quotient is
described as a measure of nonwhite satisfaction.
The nonwhite satisfaction ratio is then plotted
from 1940 to 1967 to determine whether a
J-curve can be traced for the period of the riots.
The numeric data and plot are presented in
Table 1 and Figure 2 below.
Although in some remote fashion Figure 2
might be construed to illustrate a J-curve, the
reversal in need fulfillment takes place during
the wrong years - predicting a peak of unrest
in the mid-1950s. Despite this, the data from
the black urban riots are not seen as disconfirm-
ing the theory. Davies simply argues that the
riots needed the exacerbation of southern
violence a decade later to produce the aggres-
sion that began with the frustration demon-
strated by the ratio of education to income.12
12Jbid., p. 724.
TIME
Figure 2.
Index of Nonwhite Economic Satisfaction
(Adapted from James C. Davies,
"The JCurve... ," p. 724)
Table 1. Davies's Origin and Time Sequence of Black Frustration
(adapted from James C. Davies, "The J-Curve... ," p. 726)
Average Family Income Divided by
Average Years of Schooling
Nonwhite
Satisfaction
Nonwhite
Frustration
Year
Total
Population
Nonwhite
Population
Column 2 as
% of Column I
Column 3
from 100%
1940
$1231
= 146.3
$ 489
= 84.3
57.5
42.5
8.4
5.8
3031
= 336.8
1614
= 233.9
69.4
30.6
9.0
6.9
1950
3319
= 356.9
1869
= 274.9
77.0
23.0
9.3
6.8
3890
= 385.1
2338
= 329.3
85.8
14.5
10.1
7.1
4971
= 469.0
2764
= 359.0
76.5
23.5
10.6
7.7
5620
= 530.2
3233
= 394.3
74.4
25.6
10.6
8.2
5956
= 522.5
3330
= 387.2
74.1
25.9
11.4
8.6
6559
= 560.6
3839
= 431.3
77.0
23.0
11.7
8.9
3995
= 443.8
75.3
23.7
11.8
9.0
7436
= 619.7
4628
= 503.0
81.2
18.8
12.0
9.2
a
a
78.9
20.2
Approved For Release 2005/07/01 : CIA-RDP86B00985R000100070018-6
 Approved For Release 2005/07/01 : CIA-RDP86B00985900100070018-6
968 The American Political Science Review
sociology cited by McPhail,' 3 is not congruent
with the empirical and theoretical requirements
of relative deprivation theory. The requisite
level of analysis has been violated in the
assertion that the presence of aggregate, non-
perceptual characteristics ostensibly related to
black frustration is an appropriate indication
that individual black people are indeed exper-
iencing frustration.
The importance of this logical gap is best
understood by looking at the psychological
basis of relative deprivation theory. Relative
deprivation theory claims theoretical descent
from the psychological research of John Dol-
lard and his Yale colleagues.14 The Yale team
established what has come to be known as the
frustration-aggression hypothesis. This hypoth-
esis, as it has evolved, proposes that frustration
produces various responses, one of which is
aggression, and if nonaggressive responses do
not relieve the frustration, the probability of an
aggressive response increases. 15
Relative deprivation theory has built upon
this hypothesis in viewing relative deprivation
as a major component of the chain of events
leading to aggression. It is this deprivation that
produces frustration; frustration produces ag-
gression; aggression produces civil violence; and
civil violence can be translated into political
violence. 16
The language of relative deprivation theory
is replete with reference to this psychological
foundation. As Davies notes:-"Political stability
and instability are ultimately dependent on a
state of mind, a mood, in a society. Satisfied or
apathetic people who are poor in goods, status,
and power can remain politically quiet and
their opposites can revolt .... 1117
Relative deprivation is, thus, an individual
experience, and, despite objective, aggregate
indicators, relative deprivation can only take
place if people perceive themselves to be
deprived. Economic indicators may demon-
strate that the availability of goods, status, or
power may be low or may have diminished
13McPhail, "Civil Disorder Participation," p. 1059.
14John Dollard et al., Frustration and Aggression
(New Haven: Yale University Press, 1939).
15This statement is based on the elaboration of
Dollard's work by Neal E. Miller et al., "The Frustra-
tion-Aggression Hypothesis," Psychological Review,
48 (July, 1941), 337-342. See also: Leonard Berko-
witz, Aggression: A Social Psychological Analysis
(New York: McGraw Hill, 1962), in Advances in
Experimental Social Psychology, Volume 2, ed. Leon-
ard Berkowitz (New York: Academic Press, 1965).
t6Gurr, Why Men Rebel (Princeton: Princeton
University Press, paperback, 1970).
1 Davies, "The J-curve," pp. 86-87.
from some previous level, but this does not
demonstrate that people perceive themselves to
have been deprived and are consequently dis-
content.
With .the exception of work by Grofman and
Muller,' 8 relative deprivation theorists general-
ly use the language and theoretical foundation
of individual data but perform their empirical
testing with aggregate social and economic
indicators. The chasm that separates objective
societal conditions from subjective perceptions
seems to have trapped many researchers who
have made enthusiastic and confident assertions
about the perceptual basis of individual be-
havior from theoretically remote aggregate in-
dicators.19
Given the above considerations, how would
Davies's index appear if he had obtained indi-
vidual-level data (overlooking momentarily the
entire question of perceptual vs. nonperceptual
data and simply looking at the difference in the
unit of analysis)? From our individual data,
taken from the Michigan Survey Research
Center election studies from 1956-1968,
Davies's operationalizations were replicated and
his index was reconstructed. The data are
displayed below for both whites and blacks and
by region in Table 2 and Figure 3.
The above data and Figure 3 do not demon-
strate anything as parsimonious as a J-curve.
The index of black frustration varies without
consistent pattern, and the overall impression it
imparts is that the differences between blacks
and whites are so trivial as to be substantively
inconsequential. In four out of the seven years
analyzed, blacks, contrary to the interpretation
originally derived from Davies's data, receive a
slightly higher income return for years of
"Bernard N. Grofman and Edward N. Muller,
"The Strange Case of Relative Gratification and
Potential for Political Violence: The V-Curve Hy-
pothesis," The American Political Science Review, 67
(August, 1973), 514-539.
191n addition to the sociological literature noted
above in footnote 8, see, for example: Iva K.
Feierabend, R. L. Feierabend, and B. A. Nesvold,
"Social Change and Political Violence: Cross-National
Patterns," in Anger, Violence and Politics: Theories
and Research, ed. No K. Feierabend, Rosalind L.
Feierabend, and Ted Robert Gurr (Englewood Cliffs,
N.J.: Prentice Hall, No K. Feierabend, Rosalind L.
Feierabend, "Aggressive Behaviors within Polities
1948-1962: A Cross-National Study," Journal of
Conflict Resolution, 10 (September, 1966), 249-271;
Raymond Tanter and Manus Midlarsky, "A Theory of
Revolution," Journal of Conflict Resolution, 11 (Sep-
tember, 1967), 264-280; and Ted Robert Gurr, "A
Causal Model of Civil Strife: Using New Indices," in
Anger, Violence and Politics, ed. I. K. Feierabend et
al. Much to his credit, of all the authors cited, Gurr,
Ibid., p. 210, is the only one to acknowledge the
level-of-analysis problem.
Approved For Release 2005/07/01 : CIA-RDP86B00985R000100070018-6
 Approved For Release 2005/07/01 : CIA-RDP86B009R000100070018-6
1977 The J-Curve Theory and the Black Urban Riots 969
66 58 60 62 64 66 68
TIME
TOTAL BLACKS
NORTHERN BLACKS
SOUTHERN BLACKS
Figure 3.
Index of Black Economic Satisfaction
from Individual Data
education than do whites. This is not as
surprising a finding as it first appears (as we will
demonstrate below), but is rather a penetrating
commentary on the inappropriateness of the
index as a measure of black frustration. Al-
though income generally varies with education,
it is those with lower education who receive a
higher per annum income return for unit of
education. Higher education yields a con-
siderable amount more money over a working
lifetime. On a per annum basis, however, it
takes many increments in years of schooling to
produce modest increments in income. Black
people tend to cluster at the bottom end of the
education distribution where the ratio of in-
come to education is highest. As a result, black
people have higher income to education ratios.
This is not to argue the absurd position that
whites are frustrated relative to blacks. Rather,
what is discerned from this finding is the
futility of attempting to measure a perceptual
concept such as frustration with aggregate,
objective indicators. Moreover, it is solely be-
cause Davies's index is based on a ratio of two
aggregates that the inappropriateness of the
index is not readily discernible. The manner in
which the aggregations conceal what is occur-
ring can be readily observed in Table 3 below.
In Table 3 the level of education is controlled
by partition design, and the data indicate that it
is the lower levels of education that yield the
best ratio of income to years of schooling.
Having demonstrated the inappropriateness
of Davies's index for testing the J-curve theory
in the context of the urban riots, we attempted
a refinement of the operationalization. Given
relative deprivation theorists' reliance on objec-
tive data, we were interested in whether chang-
ing the unit of analysis, from aggregate to
individual, and refining the measurement might
yield something similar to a J-curve or other
theoretically parsimonious configuration. Also,
we were interested in observing if more so-
phisticated operationalizations of the objective
but individual-level data would correspond to
our perceptual measures (see below).
To accomplish this refinement in measure-
ment, two objective measures based on the
regression of income on education, after parti-
Table 2. Individual-Level Objective Data Measures of
Davies's Operationalization of Black Economic Satisfaction'
Average of Each Head's Income Score
Divided by Each Head's Years of Schooling Scoreb
Black Satisfaction Ratio of
Column 2 to Column 1
Year
Total
White
North
White
South
White
Total
Black
North
Black
South
Black
Total
Black
North
Black
South
Black
1956
1.68
1.72
1.55
1.72
1.99
1.47
1.02
1.15
.95
1958
1.62
1.62
1.61
1.66
1.60
1.74
1.02
.98
1.08
1960
.55
.55
.51
.47
.54
.40
.85
.99
.71
1962
1.84
1.91
1.70
2.01
2.20
1.88
1.09
1.15
1.10
1964
.59
.59
.59
.53
.56
.52
.90
.93
.88
1966
.54
.55
.52
.55
.59
.46
1.01
1.08
.88
1968
.70
.70
.70
.58
.63
.52
.83
.90
.75
'Scores are not translatable into dollar terms as they reflect the relationship between measurements on coding
categories. Codes are not always comparable from year to year, but this is of no consequence since we are inter-
ested in the differences in any one year between whites and blacks and how these differences vary longitudinally.
Column 3, the nonwhite satisfaction index, multiplied by 100 is similar to the nonwhite satisfaction index in
Table 1.
bOnly heads of household were used.
Approved For Release 2005/07/01 : CIA-RDP86B00985R000100070018-6
 Aft
Approved For Release 2005/07/01 : CIA-RDP86B00985RM0100070018-6
970 The American Political Science Review
All
Educatiopal
Groups
Grade
School
High
School
College
and
Above
Total white
1.84c
3.10
1.53
0.86
North white
1.91
3.13
1.57
0.8.5
South white
1.70
3.00
1.43
0.88
Total black
2.01
2.67
1.08
0.85
North black
2.20
3.77
1.11
0.98
South black
1.87
2.17
1.05
0.55
aData are from 1962 only since the issue is a cross-sectional one and not a longitudinal one. The year was
chosen at random.
bHeads of household only.
cScore is the average of each head's income score divided by each head's year of schooling score.
tioning race and region, were created. One of
these is called the "differ variable," and it
computes the difference between predicted
income for blacks subtracted from predicted
income for whites, when income is regressed
against education. The second and more
elaborate measure is called the "discrim vari-
able," and it is constructed by subtracting a
black individual's actual income from the in-
come he would expect if he were white.20 This
20This latter operationalizations was suggested, in
part, by Otis D. Duncan, "Discrimination Against
TOTAL BLACKS
NORTHERN BLACKS
SOUTHERN BLACKS
Figure 4.
Differ Operationalization of
Black Economic Satisfaction
creates two objective individual-level measures
of black relative deprivation. The results of
these operationalizations are displayed below in
Table 4 and Figures 4 and 5.
Regardless of operational differences, the
trends for Figures 4 and 5 demonstrate the
same time-series pattern. In neither case do we
observe a J-curve. It is also important to note
the divergences between the black population
in each of the two regions. The national
configuration is often an averaging of radically
different trends between northern blacks and
southern blacks. (This demonstrates the im-
portance of looking at specific groups rather
than treating the society as a monolith, or
failing to make important distinctions within
Negroes," in Social Intelligence for America's Future,
ed. Bertram M. Gross (Boston: Allyn and Bacon,
1969). Discrimination is measured by taking the slope
and intercept of the regression equation for whites and
plugging in the observed independent variable (educa-
tion) scores for blacks. The resultant predicted depen-
dent variable (income) scores for blacks are subtracted
from the blacks' actual score on the dependent
variable. The residual [Yp (if respondent were white)
- Ya] is the measure of deprivation.
A wholly individual measure, based on differences
between actual and predicted income, was desired for
the "differ" operationalization. Such a measure would
have been very similar to the one used for the
"discrim" operationalization. In the calculation of
measures of association, the type of measure is
feasible. In calculating summary measures for the
purpose of obtaining averages for blacks, this is not
possible. In the case of any individual black., the
"differ" is the residual between predicted income and
actual income. The average of the residuals
[E (Yi-Ya)/N ] however, always equals zero.. The
operationalization is meaningful for an individual
black, but it aggregates to zero when cumulated for
the group. To provide an analogous measure for the
purpose of calculating an average, the average dif-
ference of predicted income for whites and blacks is
calculated, given each groups' respective regression
equation. Thus the average score is based on 2 Yi/N
for blacks subtracted from 2; Yi/N for whites.
Approved For Release 2005/07/01 : CIA-RDP86B00985R000100070018-6
 z
A
2 ~
U'Q
IVi
Approved Forr,elease 2005/07/01 : CIA-RDP86B00985R000100070018-6
The J-Curve Theory and the Black Urban Riots 971
/ l1i 1
1 1
1
from 1964 to 1966 there is another rise
followed by a decline in 1966 through 1968.
The J-curve interpretation is rejected because
the rising need-fulfillment pattern is only of
two year's duration, hardly commensurate with
the long-term increases required by the theory;
and the overall pattern, despite segments of rise
and decline, is not commensurate with the
J-curve.
What the data do reveal most dramatically is
a series of extreme fluctuations and changes.
There is neither a constant trend of improve-
ment, deterioration, nor stability but a series of
positive surges and negative declines. This pat-
tern of instability may be causally related to
the urban riots. Certainly, a wide variety of
research21 has argued that political violence is a
function of the ambiguity and dissonance asso-
ciated with social change, even when the
pattern of change is positive. Before pursuing in
any detail these implications from the objective
data, it is necessary to first determine if these
patterns can be validated by perceptual data -
especially given the methodological and con-
ceptual issues which were raised with regard to
the use of even individual-level, objective data.
It is to this issue that we now turn.
1.751 1
1
a
0
2.00
y r
? ua 2.15
1956 58 60
TOTAL BLACKS
NORTHERN BLACKS
SOUTHERN BLACKS
Figure S.
Discrim Operationalization of
Black Economic Satisfaction
groups, a failure often necessitated by aggregate
analysis.)
Being concerned with the urban riots, a
northern phenomenon, we must focus specifi-
cally on the trends for northern blacks. Al-
though the data (see Figures 4 and 5) do not
follow the J-?curve pattern, from 1958-1960
there is some indication of rising need fulfill-
ment, and from 1960-1964 there is some
indication of the loss of previous gains. But
The Correspondence Between
Objective and Perceptual Indicators of
Relative Deprivation
The differ and discrim operationalizations
are a marked improvement over the kinds of
21Murray Edelman, Politics as Symbolic Action:
Mass Arousal and Quiescence (Chicago: Markham,
1971); Grofman and Muller, "The Strange Case of
Relative Gratification and Potential for Political Vi-
olence"; Feierabend et at., "Social Change"; Seymour
Martin Lipset, Political Man (Garden City: Anchor
Books, 1963), esp., pp. 54-55; Michael Aiken et al.,
Table 4. Individual-Level, Regression-Based Objective Measures of Black Economic Satisfaction
Year
Total
Blacks
North
Blacks
South
Blacks
Total
Blacks
North
Blacks
South
Blacks
1956
2.12a
1.50
237
1.35b
0.87
1.40
1958
1.95
1.40
2.47
1.26
1.07
1.31
1960
1.77
0.43
2.95
1.10
0.36
2.12
1962
1.53
0.81
2.03
0.96
0.53
1.10
1964
2.15
1.46
2.40
1.21
0.91
1.31
1966
1.25
0.72
1.99
0.55
0.29
0.91
1968
2.50
1.35
3.40
1.31
0.94
1.47
aThe differ operationalization measures the difference of white predicted income minus black predicted
income. The differ is computed by taking E YiJN for whites minus E Yi/N for blacks. The operationalization is
modified when based on noncumulated individual scores. See Footnote 20.
bThe discrim operationalization is measured by taking the average of the residual between the predicted in-
come for blacks (assuming they are white, by taking the predictions based on the intercept and slope of the
white regression equation) minus black actual income.
Approved For Release 2005/07/01 : CIA-RDP86B00985R000100070018-6
 Approved For Release 2005/07/01 : CIA-RDP86B00985R@00100070018-6
972 The American Political Science Review Vol. 71
information and measures generally available to
relative deprivation researchers, since these are
based on individual rather than aggregate mea-
sures. There still, however, remains the very
important question of whether or not even
individual-based objective measures can be sub-
stituted for perceptual measures.
This issue is far more complex than it
initially appears and must be broken down into
two component issues. First, there is the
question of whether valid substitutions can be
made on a data-point-by-data-point basis.
Second, it is conceivable that the general
pattern resulting from different types of data
might be quite similar, although the points
themselves might be unrelated for any given
year. (One explanation for this phenomenon is
the time gap between an event and its psycho-
Economic Failure Alienation and Extremism (Ann
Arbor: The University of Michigan Press, 1968);
Barrington Moore, Jr., "Revolution in America," New
York Review of Books, 12 (January 30, 1969), 6-12.
logical recognition.) In such instances the gen-
eral theoretical 'interpretation provided by the
two data sets might be the same, despite the
specific temporal differences.
The preceptual indicators are based on three
items, dealing with financial need satisfactions,
from the 1956 through 1968 Michigan election
studies. These items are: (1) whether the
respondent (R) perceived his financial situation
as being better or worse than the year before;
(2) R's expectation of his future financial
prospects; and (3) R's perception of his finan-
cial satisfaction. Each of these perceptual in-
dicators was correlated with the three preceding
individual-level, objective measures of relative
deprivation: (1) the reoperationalized Davies
index; (2) the differ operationalization and (3)
the discrim operationalization. The correlation
matrix for the six items - three perceptual and
three objective - is displayed in Table 5 below.
Some of the data in the above matrix are
presented solely for informational purposes, for
we did not anticipate that all the interrelation-
Table S. Intercorrelation Matrix of Objective and Perceptual Data
R's Perception of the
Trend of Finances
R's Perception of
Future Financial Prospects
R's Perception of
Financial Satisfaction
Year
Total
Black
North
Black
South
Black
Total
Black
North
Black
South
Black
Total
Black
North
Black
South
Black
Davies's Operationalization
1968
-.053a
-.098
-.024
-.109
-.086
-.161
.148c
.219
.01 fib
1966
.095
.021
.256
.070
-.223
.449
1964
.112
.214
.051
.068
-.008
.106
.136
.235
.081.
1962
.412
.384
.432
.025
-.284
.294
1960
.374
.374
.282
.233
.204
.326
.401
.327
.479
1958
.091
.163
.018
.038
-.054
.176
.112
.254
-.093
1956
.023
-.132
-.019
-.140
-.318
-.012
.011
.021
.002
Differ Operationalizationd
1968
-.053
-.097
.012
-.123
-.094
-.143
.131
.216
-.008
1966
.149
.203
.101
.018
-.085
.096
1964
.274
.359
.231
.116
-.076
.247
1962
.492
.636
.355
.196
-.009
.383
1960
.665
.743
.405
.259
.431
.231
.410
.414
.369
1958
.438
.623
.293
.007
-.275
.193
.276
.561
.052.
1956
.281
.436
.195
.026
-.079
.053
.048
.098
.043
Discrim Operationalization
1968
-.073
-.131
-.051
-.154 -.144
-.200
1966
.144
.174
.127
-.014 .085
.096
1964
.232
.302
.180
.079 .058
.159
1962
.492
.631
.378
.184 -.030
.353
1960
.649
.735
.389
.264 .431
.254
.421
.417
.489
1958
.323
.496
.181
-.002 -.197
.126
.256
.486
.031
1956
.201
.388
.022
-.004 -.080
-.013
.028
.063
.022
a Entries are Pearson's product moment correlations.
bLife satisfaction was used in 1968 due to change in protocol.
cPerceptual variable unavailable for 1966 and 1962.
dSee Footnote 20 for a discussion of this operationalization.
Approved For Release 2005/07/01 : CIA-RDP86B00985R000100070018-6
 Approved For R
(ease 2005/07/01 : CIA-RDP86B0098 8000100070018-6
1977 The J-Curve Theory and. the Black Urban Riots 973
ships would constitute an accurate test of the
correspondence between objective and subjec-
tive indicators. The Davies's measure has been
shown to have little value in assessing relative
deprivation. Also future financial expectations
are significantly different in conceptual mean-
ing from current assessments of discrimination,
and it was anticipated that future financial
expectations would not correlate well with the
objective measure. The main areas of interest
within the matrix revolve around the correspon-
dence between the differ and discrim operation-
alizations and R's perceived trend of finances
and R's financial satisfaction.
In evaluating the matrix, the correlation
coefficient of .25 was chosen as the minimum
level for accepting a relationship between in-
dicators.22 Even at this obviously low level, the
objective operationalizations, on a point-by-
point comparison, failed to demonstrate a
relationship with the perceptual indicators. In
the case of the relationship between the differ
and discrim operationalizations and financial
satisfaction, the .25 hurdle is crossed only 40
per cent of the time for northern blacks and 20
per cent of the time for southern blacks. The
relationship between the discrim variable and
the perceived trend of one's finances shows
little overall consistency for the total black
population, despite good. results generally for
northern blacks. In this case the .25 level is
exceeded 71 per cent of the time.
The relationships between the differ and
discrim operationalizations on the one hand
and the perceived trend of one's finances on the
other are the best of the group, although still
not very encouraging. These correlations exceed
the .25 mark 71 per cent of the time for the
northern black population but only 43 per cent
(for differ) and 28 per cent (for discrim) of the
time in the southern black population. Interest-
ingly, the inappropriateness of the Davies's
operationalization seems further underscored
by its weak correspondence with the perceptual
22Correlations of .25 in social science data are
generally viewed as quite favorable. In this case,
however, we have chosen to view .25 as a minimal
level of acceptance because a criterion for the estab-
lishment of a valid substitute measure for different
levels of analysis is being considered. Without engaging
in a lengthy discourse on the philosophical issues
involved in the level-of-analysis question, it is clear
that such substitutions do involve an issue of validity,
i.e., whether one measure is a valid substitute for
another. Correlational requirements for measures of
reliability and validity are far greater than for mere
assertions of substantive importance. See for example,
Fred N. Kerlinger, Foundations of Behavioral Re-
search (New York: Holt, Rinehart and Winston, Inc.,
1965), 444-462.
measures, being the most poorly related of the
three.
The strength of the correspondences be-
tween the differ and discrim operationalizations
and the trend of finances variable is at times
quite good. Note, for example, the relationship
for northern blacks from 1958 through 1962.
We are not sure why this correspondence exists
and then radically tapers off, reaching a nega-
tive relationship in 1968. And we are reluctant
to invoke any substantive explanations since
the overall pattern of the data is not encourag-
ing from the perspective of point-by-point
substitutions. Interpreting the few cases that
have correspondences might well tend to force
systematic meaning where the overall picture
seems to indicate that there is none.
Whether this lack of correspondence is also
true of the trends of the subjective and objec-
tive data sets (as distinct from point-by-point
comparisons) will be considered in the next
section. We also wish to note that the analysis
has not assessed the possibility that in more
dramatic circumstances, such as full-scale re-
volutions, the convergence and salience of
events might be so overwhelming as to be
reflected in various kinds of assessments. Al-
though our analysis does not preclude this
possibility, at present, this is still an article of
faith which has not been subjected to empirical
verification.
A View of the Urban Riots
from Perceptual Indicators
If Davies's argument is correct and if our
objective data are too remote from the struc-
ture of the theory to test it adequately, our
empirical observations of perceptual data
should demonstrate the following results.
(1) From 1956 to 1964 the proportion of
northern blacks perceiving positive changes in
the trend of their finances should increase
steadily.
(2) After 1964 the proportion of northern
blacks perceiving positive changes in the trend
of their finances should begin to decline con-
tinually and substantially.
The above two points express succinctly and
in empirically relevant terms the essence of the
J-curve as applied to the requisite conditions of
the urban riots. In addition to this depiction of
the rise and decline scenario, there is another
equally important consideration. This relates to
the untested assumption of relative deprivation
theory generally and the J-curve specifically
that one's current trend of need fulfillment is
positively and strongly related to future finan-
cial expectations. In the context of the J-curve
Approved For Release 2005/07/01 : CIA-RDP86B00985R000100070018-6
 Approved For Release 2005/07/01 : CIA-RDP86B0098514b00100070018-6
974 The American Political Science Review
this of course occurs during the period of rising
gratifications. Its occurrence, however, is one of
the most vital aspects of the theory. For if
there is no correspondence between current
levels of need fulfillment and future expecta-
tions of need fulfillment, the theory is wholly
untenable. Without such correspondences an
intolerable gap can not occur (see Figure 1).
These observations lead to two additional con-
siderations:
(la) From 1956 to 1964 the trend of
finances for northern blacks should be positive-
ly and strongly related to future financial
expectations.
(2a) After 1964, the pre-1964 relationship
between the trend of finances for northern
blacks and their future financial expectations
should reverse itself.
The rejection of any one of the above four
statements would result in a rejection of the
J-curve theory as an explanation of the black
urban riots. Nonetheless, we will review the
results as applied to all four statements because
we are not only interested in testing the theory,
but we are also concerned with the presence or
absence of empirical justification for the
theory's basic assumptions. Moreover, special
concern surrounds statements 1 a and 2a be-
cause they are also fundamental to the assump-
tions of relative deprivation theory generally.
Before examining the data, there is an item
concerning the demarcation of time frames for
rising and declining need fulfillment that merits
attention and commentary. This concerns the
appropriateness of looking at the post-1964
period as the likely area of decline and hinges
on the interpretation of the Watts riot which
occurred in August of 1965. If this event is
perceived as having initiated the riot phase,
then the period before 1964, it has been
suggested, is the more appropriate place to look
for financial reversals. Watts would then be
considered as the point of the "intolerable
gap," where the disparity between continually
rising expectations and falling actual need
fulfillment culminates in civil disorder.
Such issues might have been readily resolved
if the J-curve theory had been empirically more
rigorous in specifying the operationalization of
the "intolerable gap." One of the difficulties
with all forms of post hoc theory is that they
manifest an empirical prowess greater in ap-
pearance than in fact. The requirements of a
theory tested through prediction would of
necessity mandate an operationalization of crit-
ical concepts. Where theory is constructed
after instead of before prediction, such em-
pirical precision is too easily overlooked. This
lack of specification further complicates analy-
sis because all forms of civil disorder, even coup
d'etats, are often a series of events rather than a
single event. These events can occur in close
temporal proximity, but, sometimes, as in the
case of the riots, there is substantial temporal
space between them. As to which single event
among the set of events coincides with the
crucial intolerable gap is an issue that is
generally ignored.
Further specification by the original theory
could have provided some assistance in properly
assessing the role of Watts in this scenario.
However, even if Watts is considered the initi-
ator of this phase of riot activity, the two-year
gap between Watts and the preponderance of
riot activity strongly suggests that the down-
ward cycle of perceived need fulfillment began
about 1964, was in evidence by August, 1965,
but it did not approach the intolerable gap
phase until 1967.
An interpretation of Watts as coming at the
bottom of the need fulfillment spiral would
have to focus on the sharp decline from 1960
to 1962 in Table 6 and Figure 6. (Interpreta-
tion is restricted to the perceptual data because
they constitute the best indicators.) But this
analysis would be undermined by the sharp rise
in need fulfillment occurring from 1962 to
1964. It would also be forced to ignore! the
important cycle of fluctuation and change
throughout the data set, not to mention. the
two-year gap between Watts and the other riots.
For these reasons, we have chosen to maintain
1964 as the point of demarcation between
rising and declining need fulfillment in the
hypotheses above.
Turning now to the data and consideration
of increased need fulfillment, Table 6 and
Figure 6 below demonstrate that blacks did not
experience increased need fulfillment from
1956 to 1964. Instead of a trend of continual
improvement, there exists a series of sharp
fluctuations, not at all unlike the results ob-
tained with the objective data. If the northern
blacks are compared to northern whites, the
contrast in this regard is rather marked. Across
time relatively stable percentages of northern
whites perceive their finances as improving,
while percentages of northern blacks experienc-
ing the same perception fluctuate dramatically.
Except for the tremendous upsurge in 1964,
characteristic of all groups, northern whites
show a pattern of marked stability. Excluding
1964, the range of responses varies from a low
of 34 per cent to a high of 39 per cent. Given
considerations of sampling error, it is not
unreasonable to assume that a straight line
would, with the exclusion of 1964, accurately
characterize these data. Southern whites also
Approved For Release 2005/07/01 : CIA-RDP86B00985R000100070018-6
 Approved For release 2005/07/01 : CIA-RDP86B0098000100070018-6
1977 The J-Curve Theory and the Black Urban Riots 975
show a fairly stable pattern, but the fluctuation
that does exist ensues from the close temporal
proximity between the highest and lowest
points in the longitudinal distribution. Finan-
cial optimism triumphed in 1964 but crashed
disastrously by 1966. Outside of that sharp
downward spiral the pattern is seemingly con-
stant. Even with due consideration of the sharp
drop from 1964 to 1966, the range for south-
ern whites runs from a low of 29 per cent to a
high of 41 per cent, with a difference of only
12 per cent.
The argument for the stability of white
perceptions is best appreciated by looking at
the corresponding data for the blacks. Northern
blacks run a gamut of alternate high to low and
low to high fluctuations, with a range from 21
per cent to 46 per cent, or five times that of
northern whites. For southern blacks the pat-
tern is similar except there is somewhat more
stability, each low ebb is not necessarily fol-
lowed by another high point, and the period
from 1964 on is virtually constant. The range is
23 per cent, or nearly twice that of southern
whites.
Interestingly, if the same longitudinal rela-
tionships are examined from the perspective of
the respondents' conception of their financial
satisfaction or their expectation of their future
financial satisfaction, the same phenomenon of
white stability and black instability is repeated
(see Tables 7 and 8 below).
The data clearly indicate that it is the
northern black population that experiences the
greatest instability. Even incorporating the im-
pact of 1964 - a year which appears to have
produced sharp financial optimism among all
groups - the variation among whites in both
the North and South is nowhere nearly as sharp
as it is for blacks; and it is again the northern
blacks, not the southern blacks, who illustrate
the greatest instability.
Is it likely that this series of fluctuations has
emerged as a consequence of systematic dif-
ferences in sampling design between off-year
and presidential election year studies? Or is the
instability, perhaps, due to the differences in
sample size between blacks and whites, with
blacks demonstrating more instability simply
because of smaller NO The pattern of fluctua-
tion and change observed in Figure 6 and Table
6 can not be dismissed as mere aberrations in
the data resulting from systematic changes in
the sampling frame between presidential and
nonpresidential election studies. The Survey
Research Center did not use different sampling
frames in the off-year and presidential year
studies. From 1956 through 1960, the identical
Northern white
Better
39%
34%
36%
35%
48%
37%
34%
Same
43
45
47
46
38
37
45
Worse
19
22
18
20
14
26
21
(Total)
101%a
101%a
101%a
101%a
100%
100%
100%
(N)
(1168)
(1165)
(1179)
(777)
(3017)
(797)
(1644)
Northern black
Better
33%
21%
46%
21%
44%
33%
32%
Same
43
49
30
52
40
41
44
Worse
24
30
25
27
16
27
25
(Total)
100%
100%
101%a
100%
100%
101%a
101%a
(N)
(63)
(84)
(77)
(52)
(184)
(78)
(110)
Southern white
Better
40%
32%
35%
38%
41%
29%
33%
Same
44
46
49
47
44
45
50
Worse
17
22
17
15
16
26
17
(Total)
101%a
100%
101%a
100%
101%a
100%
100%
(N)
(425)
(464)
(544)
(382)
(1146)
(322)
(690)
Southern black
Better
28%
32%
19%
17%
40%
40%
38%
Same
51
32
42
50
43
35
48
Worse
22
37
40
33
17
25
14
(Total)
101%a
101%a
101%a
100%
100%
100%
100%
(N)
(83)
(76)
(91)
(58)
(234)
(57)
(132)
Approved For Release 2005/07/01 : CIA-RDP86B00985R000100070018-6
 Approved For Release 2005/07/01 : CIA-RDP86B00985b00100070018-6
The American Political Science Review Vol. 71
NORTHERN WHITES
NORTHERN BLACKS
SOUTHERN WHITES
SOUTHERN BLACKS
Figure 6.
Percentage of Respondents Perceiving the Trend
of their Financial Situation as Improving
sampling frame was used. The stability of this
design was reinforced by the presence of a large
panel study for this time period. Consequently,
the sharp fluctuations and changes observed in
Figure 6, for the 1956 through 1960 period,
can hardly be attributed to changes in the
sampling frame. In 1962 the sampling design
was updated and the resulting frame was used
through and including 1968, making it impos-
sible to attribute changes observed in this
period to changes in the sampling design. Even
the changes between 1960 and 1962 can not be
attributed to the change in sampling frames, for
the large metropolitan self-representing PSUs,
where some two-thirds of the northern blacks
reside, remained as self-representing units.23
How about the question of sample size? Can
the changes be attributed to differences in
sample size for blacks and whites? Do the
blacks display more volatile patterns simply
because their sample size is smaller? This
argument is untenable because it overlooks the
23We wish to thank Leslie Kish and Irene Hess of
the sampling section of the University of Michigan's
Institute for Social Research for explaining the sampl-
ing design to us and clarifying this issue.
crucial implications of the Law of Large Num-
bers (Bernoulli's theorem).24 The critical ques-
tion is not whether the black sample is smaller
or larger than the white sample. The critical
question is whether the black sample, both in
the North and South, is sufficiently large to
meet the requirements of Bernoulli's Theorem.
As Table 6 demonstrates, these well-known cri-
teria for statistical inference are more than
amply met by the sample Ns.
Although the above commentary would
seem to lay this issue to rest, we decided to
pursue this matter one step further. A ten per
cent sample of white respondents was selected
by means of a random number generator
applied to each discrete data point, with region
controlled. If the volatility for blacks relative to
whites is purely a function of differences in
sample size, then a sample of whites severely
reduced in size should demonstrate the same
volatile pattern as that observed for the blacks.
Yet, the ten per cent sample was virtually
indistinguishable from the larger sample in both
the case of northern and southern whites. This
finding plus the information on the sampling
frames would appear to eliminate the assump-
tion that the results could be attributed in any
meaningful way to the sampling design or the
differences in sample size.
Now that we have demonstrated that the
observed results cculd not be meaningfully
attributed to sample size or design, and that
rising expectations did not characterize the
trend of need fulfillment for northern blacks
prior to 1964, is there any indication the need
fulfillment declined sharply after 1964? Again,
if we look at Table 6 and Fiture 6, we note a
drop from 44 per cent in 1964 who perceived
the trend of their finances as improving, to 33
per cent in 1966. This is, however, a smaller
drop than took pace in the two previous
periods of decline, and the relationship sta-
bilized in 1968. Consequently, there is no
confirmation of a continual decline after 1964,
as the theory would predict.
Turning now to the all-important assump-
tion of relative derivation theory generally
that current need fulfillment is related to
expectations of future need fulfillment, this is
shown to be generally unsubstantiated (see
Table 9). The assumption could be demon-
24The theorem is represented as follows: probabili-
ty (If/n- PIKE)- O as N-0o; where f is the frequen-
cy of occurrence of event X among N trials, p is the
probability of X, E is some arbitrarily small positive
number. It is generally assumed that good results are
produced with N greater than 30. It also is at this
point that the normality assumption of the Central
Limit Theorem can be relaxed.
Approved For Release 2005/07/01 : CIA-RDP86B00985R000100070018-6
 Approved For Release 2005/07/01 : CIA-RDP86B00985000100070018-6
1977 The J-Curve Theory and the Black Urban Riots 977
strated as warranted for northern whites, and to
a far lesser extent for southern whites; however,
in the case of northern blacks it is most
untenable. (This result provides another indica-
tion of the importance of performing the
analysis within specific subgroups as opposed to
treating the society or the group as a monolith,
a problem inherent in aggregate data analysis.)
Consequently, this assumption which is virtual-
ly axiomatic to all forms of relative deprivation
theory is not as universal as its strategic role
demands. This relationship obviously varies
from group to group and from circumstance to
circumstance. All we can say is that sometimes,
and for some groups, future expectations are a
function of current need fulfillment, and at
other times, and for other groups, this relation-
ship does not hold .2 5 Indeed, this is hardly the
kind of axiomatic statement on which a science
of revolution is to be built.
25Ted Robert Gun informs us (Correspondence
12/27/74) that although he agrees that the linkage
between future expectations and current need fulfill-
ment is a fundamental assumption of relative depriva-
tion theory, the relationship may vary within the
social structure. In the course of yet uncompleted
research (with Raymond Duvall) Gurr has theorized
An important implication of these results is
that for blacks, as opposed to whites, percep-
tions of past experience are not instrumental in
shaping perceptions of the future. This would
suggest that although whites can rely, to some
degree, on past experience as a means of
interpreting the future, blacks cannot. It is
probably not unwarranted to infer that this is
an additional indication of the ambiguous and
unstable environment which blacks inhabit.
The data further indicate that instead of the
increasingly strong relationship between trend
that individuals in the middle range of attained values
(need fulfillment) appear to expect more gains in the
future than those at the extremes of the distributions.
If Gurr is correct, this would require some critical
restatement of the underlying assumptions of relative
deprivation theory. Some preliminary analyses that we
have undertaken demonstrate that for blacks, those at
the lowest end of the continuum have the highest
expectations. For example, preliminary runs from
SRC's Campbell-Schuman data set indicate that young
(18-35) northern, black males are most dissatisfied
with their current situation and most optimistic about
the future. Needless to say, the issue Professor Gurr
raised is an important one that could conceivably have
a major impact on relative deprivation theory.
North (white)
(Pretty Well) Satisfied
41%
38%
36%
a
46%
a
26%
(More-or-Less) Satisfied
42
42
42
40
65
Not Satisfied (At All)
17
20
22
14
9
100%
100%
100%
100%
100%
N=
(1168)
(1163)
(1179)
(3012)
(1684)
North (black)
(Pretty Welt) Satisfied
29%b
16%
22%
31%
15%
(More-or-Less) Satisfied
32
26
26
39
65
Not Satisfied (At All)
40
58
52
30
20
101%c
100%
100%
100%
100%
N=
(63)
(86)
(76)
(185)
(109)
South (white)
(Pretty Well) Satisfied
49%
50%
45%
46%
24%
(More-or-Less) Satisfied
34
34
35
42
66
Not Satisfied (At All)
17
16
20
13
10
100%
100%
100%
101%a
100%
N=
(429)
(465)
(553)
(1146)
(696)
South (black)
(Pretty Well) Satisfied
24%
30%
17%
32%
12%
(More-or-Less) Satisfied
37
37
42
34
74
Not Satisfied (At All)
39
33
42
35
14
100%
100%
101%C
101%c
100%
(83)
(76)
(91)
(235)
(131)
aData unavailable for 1962 and 1966.
bin 1968, for "Life Satisfaction" the categories are: Completely Satisfying, Pretty Satisfying, Not Very
Satisfying.
cColumn adds to more than 100 per cent because of rounding.
Approved For Release 2005/07/01 : CIA-RDP86B00985R000100070018-6
 Approved For Release 2005/07/01 : CIA-RDP86B0098000100070018-6
Table 8. R's Perception of Future Financial Prospects by Race and Region
(1956-1968)
1956
1958
1960
1962
North (white)
Get better
43%
41%
39%
39%
45%
37%
34%
Stay the same
47
46
50
55
45
51
55
Get worse
10
13
11
6
10
12
10
100%
100%
100%
100%
100`70
100%
99?oa
N= (1094)
(1130)
(1107)
(686)
(2884)
(700)
(1644)
North (black)
Get better
60%
68%
54%
36%
72%
37%
27%
Stay the same
29
26
39
55
24
57
58
Get worse
10
6
8
10
4
6
15
99%a
100%
101%a
101%11
100%
100%
100%
N =
(58)
(84)
(67)
(42)
(173)
(63)
(110)
South (white)
Get better
44%
41%
38%
38%
40%
28%
35%
Stay the same
48
46
54
57
51
59
55
Get worse
8
13
8
5
9
13
10
100%
100%
100%
100%
100%
100%
10050
N =
(396)
(450)
(499)
(358)
(1098)
(287)
(690)
South (black)
Get better
49%
53%
61%
21%
60%
51%
31%
Stay the same
40
33
31
71
35
36
59
Get worse
10
13
9
8
5
13
11
99%a
99%a
101%a
100%
100%
100%
101%a
N =
(77)
(75)
(81)
(48)
(224)
(45)
(132)
aDoes not sum to 100% because of rounding error.
Table 9. Correlation Between R's Perceived Trend of Financial Fulfillment and
R's Future Financial Expectations by Race and Region
1962
1964
1966
1968
Northern whites
.39
.39
.39
.41
.50
.44
.44
Southern whites
.22
.45
.29
.37
.28
.54
.42
Northern blacks
.05
-.24
.71
.02
.34
.44
.51
Southern blacks
-.01
.24
.44
.33
.42
.24
.25
of finances and financial expectations, as was
predicted by the J-curve theory for the period
between 1956 and 1964, northern blacks ex-
hibit an erratic pattern, with correlation co-
efficients ranging from a low of -.24 in 1958
to a high of +.71 in 1960. Moreover, instead of
the coefficients declining after 1964, they
actually increase (see Table 9).
The perceptual data indicate that in the case
of each of the four hypotheses relating the
J-curve to the black urban riots of the 1960s,
the theory was rejected. Equally important is
the observation that one of the basic axiomatic
components of the J-curve theory and relative
deprivation theory generally was found to be
unconfirmed - that future expectations of
need fulfillment are a function of current
perceptions of need fulfillment.
Approved For Release 2005/07/01
It has been suggested that although the
J-curve theory as disconfirmed by the data as
an explanation of the urban riots, this is not
sufficient reason for implying a general lack of
confidence in the theory. This argument is
based on the consideration that Davies may
have erred in attempting to extend his theory
beyond the parameters for which it was de-
signed. Davies's theory, it has sometimes been
asserted, was directed at revolutionary events,
and the urban riots, whatever else they may be
called, were not revolutionary events. The lack
of confirmation in the instance of the urban
riots, it has been sometimes further suggested,
is simply the result of attempting to explain
diverse phenomena with a single dynamic. This
argument is worthy of attention for two rea-
sons: (1) It is consonant with arguments of-
 Approved For ;please 2005/07/01 : CIA-RDP86B00985R000100070018-6
1977 The J-Curve Theory and the Black Urban Riots 979
fered in behalf of middle range theories-2 6 (2)
In a recent communications exchange between
Davies and David Snyder writing with Charles
Tilly, Davies appears to have made a similar
point.27
Snyder and Tilly28 assert that a longitudinal
evaluation of political violence in France from
1830 to 1960 does not fit the J-curve. In
response, Davies29 argues that a long, whole-
country series analyzing various events of civil
disorder is not an adequate test of his theory.
Davies implies that his theory need not be
expected to account for nonrevolutionary
events. In light of his own work, this is a rather
astounding argument. Even if one were to view
Davies's application of the J-curve to the urban
riots as an over-extension of the theory, such a
position would hardly be commensurate with
Davies's original formulation of the theory or
his own previous work. As Snyder and Tilly
note, ". . . Davies himself had applied his
scheme to student protests, draft riots, the
protests of American blacks in the 1960s not to
mention the Pullman strike, the Dorr Rebellion,
plus many other conflicts of all shapes and
sizes."30
If Davies is to be consistent in his contention
that Snyder and Tilly's work is irrelevant to his
arguments, then it is incumbent upon him to
propose a reformulation of the parameters of
his own theory. Certainly, the case can be made
in terms of middle-range theory that a variety
of historical experiences cannot, at this stage of
social science, be explained in the context of a
single dynamic. In Davies's case, this recon-
struction is at best difficult, for the develop-
ment of the theory itself is based on a diversity
of types of civil disorder. Davies's development
and previous applications of the theory indicate
that the attempt to explain the urban riots is
wholly in accord with the theory's general
formulation and application. It could not be
construed as an attempt to apply the theory
outside the parameters for which it was in-
tended.
26For a discussion of this position see Robert K.
Merton, Social Theory and Social Structure (Glencoe,
Ill.: The Free Press, 1957), pp. 5-11.
27Davies, "Comments E,on Snyder and Tilly],"
American Sociological Review, 39 (August, 1974),
607-612.
28David Snyder and Charles Tilly, "Hardship and
Collective Violence in France, 1830-1960," American
Sociological Review, 37 (October, 1972), 520-532.
29Davies, "Comments," pp. 607-610.
30Snyder and Tilly, "Comments [On Davies],"
American Sociological Review, 39 (August, 1974),
611.
The Black Urban Riots:
Some Suggestive Theoretical Issues
In the period preceding and during the riots,
the black community did not experience a
sustained improvement in its economic situa-
tion followed by a sharp reversal. Instead of
this J-curve scenario, the black community
generally and the northern black community
specifically displayed dramatic ambiguity and
instability in describing the trend of their
finances, their financial satisfaction, and their
expectations of future financial improvement.
These observations were borne out by both the
objective and subjective data. The ebbs and
flows of the points in any one year may vary
from objective data to subjective data. None-
theless the correspondence in the overall trend
remains.
Were these perceptions of instability more
marked in the 1960s than in any other previous
contemporary period of black history? This we
do not know. Even so, the very pattern of the
data suggests an alternate explanation of the
riots. Here, however, we must tread with
caution. Relative deprivation studies typically
observe a pattern of data, often aggregate data,
together with a concomitant episode of col-
lective or political violence. The two are then
interpreted as being related, with some form of
relative deprivation, represented by the data
configuration, being assumed. as the cause of
the ensuing violence. In terms of classical
considerations of causality and experimental
design, this type of deductive reasoning leaves
much to be desired.31 It is, however, the basis
for many of the theoretical interpretations of
revolution and the underlying causal mechan-
ism of virtually all relative deprivation studies.
Sears and McConahay32 note this type of logic
is also the basis for the relative deprivation
approach to the study of the black urban riots.
While acknowledging the limitations of such
procedures, we must also note the difficulty in
obtaining the type of data that would directly
link the assumed preconditions of violence with
the acts of violence. This is especially true of
historical studies of violence, but this problem
is also prevalent in studies of contemporary
episodes of violence. The research of Gurr,33
31For a discussion of causal inference in classical
experimental design, see Donald T. Campbell and
Julian C. Stanley, Experimental and Quasi-Experi-
mental Design for Research (Chicago: Rand McNally,
1963).
32David O. Sears and John 13. McConahay, The
.Politics of Violence (Boston: Houghton Mifflin,
1973), pp. 90-91, footnote 1.
33Gurr, Why Men Rebel.
Approved For Release 2005/07/01 : CIA-RDP86B00985R000100070018-6
 Approved For Release 2005/07/01 : CIA-RDP86B009800100070018-6
980 The American Political Science Review
Davies,34 and the Feierabends,35 to name but
a few of the most regarded scholars working in
this area, have manifested these inherent and
often unavoidable difficulties.
These limitations should not preclude in-
cisions into the problems of collective and
political violence. They must, however, caution
us about the limitations of our inferences, and
make us aware that we are perhaps more
engaged in an exploration enterprise than in
"`confirming" hypotheses.
With such caveats in mind, we are struck by
the suggestiveness of our own data. The data
demonstrate that the black community general-
ly, and the northern black community par-
ticularly, experienced extreme fluctuation and
ambiguity in its perceptions of the trend of its
finances, its financial satisfaction, and its expec-
tations of financial improvement. These ex-
treme fluctuations and ambiguities of black
perceptions might have led to the urban riots of
the 1960s.
The pattern of data fits both the theoretical
writing by Feierabend et al.36 based on obser-
vations from aggregate data, and the work by
Grofman and Muller37 based on individual
data. Feierabend et al. see constant flux and its
attendant uncertainties as producing frustra-
tion. As they note, "uncertainty is a special
quality of expectations. Ambiguity as to
whether the future will bring disaster or salva-
tion should be considered a distressful experi-
ence, adding to the present sense of frustration.
Only in the case of disaster is certainty more
likely to be judged as more frustrating than
uncertainty."38 Frustration, of course, leads to
aggression; and aggression, under the proper
circumstances, will lead to political violence.
Similar observations have been made by
Grofman and Muller. They found that "the
greatest potential for political violence is man-
ifested both by individuals who perceive nega-
tive change ... and by individuals who perceive
positive change. . . , while those who perceive
no change manifest the least potential for
political violence."39
The theoretical and empirical foundations
for these observations are anchored in a long
tradition of social science research. In his
34Davies, "The J-curve."
3 1Feierabend et al., "Social Change."
36lbid.
37Grofman and Muller, "The Strange Case of
Relative Gratification."
38Feierabend et al., "Social Change," p. 637.
39Grofman and Muller, p. 514.
famous study of suicide, Emile Durkheim40
noted that anomie could be produced by social
disruption irrespective of whether it resulted
from economic disaster or an abrupt transfor-
mation leading to power and wealth.
Ambiguity, fluctuation, and uncertainty as
conditions which contribute to one's suscep-
tibility to be mobilized for revolt have been
addressed in the works of Barrington Moore,
Jr.,41 Michael Aiken,42 Seymour Martin Lip-
set,43 Norman Cohn,44 and Murray Edel-
man.45 In a speculative essay, Edelman has
applied this interpretation directly to the black
urban riots. He argues that man's expectations
of the distant future influence the cognitions
that explain his current situation. The im-
portant expectations of status and security
come from governmental acts. The inconsistent
fluctuation between responsiveness and nonre-
sponsiveness to black needs by the government
produced alienation, fear, and anger. In this
climate, Edelman notes, a precipitating incident
can easily touch off violence. Moreover, he goes
on to argue, the greatest ambiguity existed. for
northern blacks.46
The interpretation suggested by our percep-
tual data is highly congruent with these theore-
tical statements that fluctuation, change, and
ambiguity create situations highly conducive to
violence. Edelman's suggestion that ambiguity
is greatest among the northern blacks is also
confirmed by our perceptual data.
What can be said with some confidence is
that the black community generally and the
northern black community specifically experi-
enced a great deal of perceptual uncertainty
with regard to their financial situation and their
expectations of their future financial circum-
stances. We can also argue that there is at least a
substantial set of theoretical and empirical
work to indicate that such conditions have a
high probability of precipitating political vi-
olence. It is at least plausible to note that the
pattern of fluctuation and change and the
attendant perceptual ambiguities may be com-
ponents of the causal nexus that culminated in
40Fmile Durkheim, "Anomie and Suicide," in
Sociological Theory, ed. Lewis A. Coser and Bernard
Rosenberg (New York: MacMillian, 1970), p. 529.
41Moore, "Revolution in America."
42Aiken et al., Economic Failure, Alienation and
Extremism.
43Lipset, Political Man.
44Norman Cohn, The Pursuit of the Millennium
(New York: Harper Torchbooks, paperback, 1961), p.
319.
45Edelman, Politics as Symbolic Action.
46lbid., pp. 20-21.
Approved For Release 2005/07/01 : CIA-RDP86B00985R000100070018-6
 Approved For lease 2005/07/01 : CIA-RDP861300985500100070018-6
1977 The J-Curve Theory and the Black Urban Riots 981
the black urban riots. Certainly, this argument
is no less plausible nor less buttressed by
empirical evidence than most of the current
explanations of the riots. The difficulty resides
in the linkage between the pattern of the data
on the one hand and the violence on the other.
Until it is possible to bridge this chasm in a
causally meaningful fashion, our results must
remain suggestive avenues for further research.
Summary
The purpose of this research was to test the
J-curve theory in the context of the black
urban riots by using both perceptual and
objective indicators. The results demonstrate
that the J-curve theory is not a valid explana-
tion of the riots. Several empirical, method-
ological, and conceptual problems were also
noted. These basically revolved about the re-
grettable tendency of :researchers to test a
theory based on individual perception with
aggregate and objective data.
The research demonstrated that: (1) Aggre-
gate indicators can be highly misleading substi-
tutes for individual indicators. The J-curve's use
of an aggregate index to measure black frustra-
tion was found, upon examination of the same
measure based on individual data, to be highly
misleading. Moreover, the inappropriateness of
the operationalization was concealed by the
process of aggregation. (2) Objective indicators
cannot be used as direct substitutes for percep-
tual indicators, even when both are based on
individual data. There is, however, some indi-
cation from our analysis that while sub-
stitutions are unwarranted on a point-by-point
basis, substitutions may be justified when based
on the overall trend of the data. The generality
of this conclusion, however, awaits further
confirmation. (3) Regardless of the nature of
the data configuration, societies cannot be
viewed as monoliths. It is imperative that
specifications and distinctions be made for
various groups. As our data show, trends for the
total black population often obscured crucial
divergencies between northern and southern
blacks. Relative deprivation-based theories that
treat societies as monoliths may well conceal as
much as if not more than they reveal. If
statements such as "revolutions are most likely
to occur when things get better" are to have
any theoretical meaning, then it must be
demonstrated for whom things have gotten
better. Is it indeed the revolutionary class?
When unspecified data demonstrate that things
improved for the society as a whole, can we be
sure that things did not simultaneously get
worse for the revolutionary class? Is it the rising
expectations resulting from overall improve-
ment that caused the revolution? Or is it an
increasing gap between the revolutionary class
which experiences no improvement, and the
rest of the society that experiences increased
need satisfaction? The monolithic approach
cannot assist the student of revolution in
resolving such questions and may well lead to
deceptively simple answers. (4) The critical
assumption of the J-curve theory and relative
deprivation theory generally that future expec-
tations of need fulfillment are a function of
current need fulfillment trends was shown to be
largely unsubstantiated. The data did indicate
modest correlations for these two variables for
northern whites, and to a lesser extent for
southern whites and southern blacks. There was
no basis for this assumption when applied to
the crucial northern black community. Since
this assumption is one of the axioms of the
J-curve, its consequent lack of verification
should at least raise the question of how the
axioms of the J-curve might be restructured to
accommodate this information.
To students of revolution who have used
relative deprivation type theories, this research
advises caution. Such caution ensues from a
firmer recognition of the problems of analysis
of historic revolutionary events that derive
from data which are of necessity based on
aggregate, objective indicators. Such analysis
might be thoroughly misleading, as this research
has demonstrated. This factor has even caused
some researchers to forsake relative deprivation
theory as an explanatory device. Such rejection
is not based on theoretical limitations but on
the inability to fit relative deprivation theory to
appropriate data. At the same time, if our
observation is correct that even in the absence
of point-by-point correlations between objec-
tive and subjective indicators, there is con-
sistency between the overall trends, relative
deprivation theory might still have great
promise. It remains for research to be directed
at this problem through simultaneous analysis
at both the individual and aggregate levels.
Finally, we propose some avenues for fur-
ther research on the substantive issue of the
etiology of the black urban riots. The cycles of
fluctuation and change that emerged from our
analysis of perceptual data fit an extensive
theoretical literature on the causes of civil
disorder and revolution. Although we are reluc-
tant to follow many of our colleagues across
the inferential chasm that depicts a causal
relationship when a diagrammatic configuration
of relative deprivation exists in the same tem-
poral space as an act of violence, the con-
gruence of our findings with other theoretical
Approved For Release 2005/07/01 : CIA-RDP86B00985R000100070018-6
 Approved For Release 2005/07/01 : CIA-RDP86B00985`000100070018-6
statements suggests that the fluctuation-change
hypothesis might well be a fruitful area for
further research into the causes of the black
urban riots.
Approved For Release 2005/07/01 : CIA-RDP86B00985R000100070018-6