RAND Corporation. Social Science for Counterterrorism
Подождите немного. Документ загружается.
Analytic Measures for Counterterrorism and Counterinsurgency 465
Table B.1 (continued)
Concept Measure Group
Measure
Type Comments
Grievances Antigovernment graffiti MOE Moderate data collection
difficulty; reliable and
responsive; connection with
end-state
Mobilizing
structures
Hazardous social group
membership
MOE/
MOP
Very difficult data
collection; reliability is
questionable; possible
direct connection with end-
state; must be considered
alongside other social
ferment indicators
Mobilizing
structures
Population’s willingness to
talk about insurgency
MOE Moderate data-collection
difficulty; reliable
and responsive; direct
connection with end-state
government action and population response, they may be valid MOEs
for regime legitimacy.
Measures related to social ferment may also be directly connected
to the end-state, because social movements may indicate that substate
actors seek to undermine the regime. e frequency or intensity of
public demonstrations, riots, and desertions from the military may all
be measures of shifts in the social ferment, depending on local condi-
tions (Tanham, 1988). ese measures are reliable and valid for gauging
the end-state and constitute MOEs, in that they indicate a population
response. Further, polls indicating that perceptions of the legitimacy
of the regime are moderating may also be valid and reliable as MOPs,
although they do not indicate that the populace is necessarily willing
to take action. Measures of the perceived illegitimacy of the regime
should be considered in tandem with measures of repression, because
direct action by the regime may affect the population’s willingness or
ability to act.
Repression
Repression can have many manifestations, which suggests the need
for multiple measures. Measures may include incidents or intensity of
466 Social Science for Counterterrorism: Putting the Pieces Together
indiscriminate military or police actions against particular religious
or political groups, disproportionate flows of government economic
support to unpopular or minority institutions, estimated changes in
the numbers of disenfranchised voters, or changes in access to key
resources for certain subpopulations. Repression measures should be
considered as MOPs because of their lack of a direct connection to the
end-state and lack of responsiveness. ese MOPs feed into perceptions
of regime illegitimacy, and thus the population responses mentioned
above should be closely monitored as MOEs that react to repressive
processes.
Demography
Demographics interacts with disruptive aspects of modernization and
resultant changes in the social ferment. Census or survey data show-
ing increases in the number of youth in the population may track the
potential for terrorism or insurgency when combined with other fac-
tors. Detailed surveys of at-risk subpopulations may be of interest in
monitoring reactions to certain policies or actions. Although demo-
graphics cannot be influenced, demographic trends should be moni-
tored in conjunction with other metrics, such as selection measures of
the social ferment.
Grievances
Grievances provide a trigger for violent action. In situations such as
Iraq, where the policy of the counterinsurgent is considered to be a
communal grievance, changes in attitudes and activities of the popu-
lace should be considered as important indicators of CT/COIN effec-
tiveness. Polling, monitoring local or regional media sources, and anal-
ysis of information gathered in the process of conducting information
operations from power brokers or elites may be informative about shifts
in beliefs or attitudes about the counterinsurgent force. However, these
indicators may provide only MOPs, because they are not directly linked
to the end-state despite having high responsiveness. Other indicative
measures, such as the incidence of anticounterinsurgent graffiti, may
be more closely tied to the end-state, because it combines opinion with
some form of antigovernment action.
Analytic Measures for Counterterrorism and Counterinsurgency 467
Mobilizing Structures
Mobilizing structures are clearly an important conduit for recruitment,
and monitoring the nature of these structures will be of interest to CT/
COIN officials. e relative levels of personal ties that individuals have
with known terrorists are clearly proxy measures for terrorism, because
there is a high correlation between the ratio of number of personal ties
with group members to other personal ties and the individual’s pro-
pensity to join the group. In this sense, measuring personal ties is of
interest to the counterinsurgent/counterterrorist; monitoring areas or
groups where the aforementioned ratio is high may afford some abil-
ity to warn, and decreasing this ratio in at-risk populations may be an
important MOE. Numerous experts on COIN have noted the warn-
ing power of the general populace’s willingness to talk about possible
insurgent activity with counterinsurgent forces (Hosmer and Crane,
1962; Sweetland, 1968; and Tanham, 1985), and this measure may
also capture the strength of insurgent mobilization once it has become
powerful enough to coerce the populace.
Radicalization Factors
Because radicalization is a multicausal process and a multiple-fields
approach is necessary to understand the phenomenon, the mea-
sures used to monitor and assess radicalization must be drawn from
the individual, group, and environmental levels that are all critical
to the process of radicalization (see the chapter by Helmus). Develop-
ing measures for radicalization is less an exercise in trying to predict
or warn of terrorist events or an incipient insurgency than an effort
to understand at-risk subpopulations and the factors that influence
the radicalization process of individuals within a given subpopula-
tion. e measures for radicalization may be most useful for moni-
toring terrorist groups in stronger states where full-scale insurgency
seems unlikely, although they will also have utility for possible insur-
gent scenarios as well. For radicalization, the desired end-state is the
prevention or avoidance of the radicalization of at-risk subpopula-
468 Social Science for Counterterrorism: Putting the Pieces Together
tions. e elements that I consider in subsequent paragraphs are
social groups, a desire for systemic change, and duty. (See Table B.2.)
Social Groups
Taking into account that social groups are a critical factor in the radi-
calization process, identifying and tracking them is important to suc-
cessfully predict and disrupt threats. Existing research supports the
efficacy of monitoring those social and familial groups with known
connections to terrorists, as well as groups of foreigners who may feel
culturally alienated. e monitoring of networks is inherently difficult
and controversial in liberal democracies, because it requires extensive
Table B.2
Radicalization Measures
Concept Being
Measured
Measure
Specification
Measure
Type Comments
Social groups Number of social
groups or individuals
with ties to known
terrorists
MOE Very difficult data collection;
reliability is questionable; direct
connection with end-state
Desire for
systemic change
Polled dissatisfaction
on systemic issues
MOP Difficult data collection;
reliability is questionable; direct
connection with end-state
Desire for
systemic change
Polled desire for Sharia
law
MOE Difficult data collection;
reliability is questionable; direct
connection with end-state
Desire for
systemic change
Number of protests or
riots
MOE/
MOO
Easy data collection; reliability
is questionable; connection
with end-state may depend on
political system
Duty Number of killed or
wounded civilians
MOP Moderate data-collection
difficulty; validity is
questionable; direct connection
with end-state
Duty Internet traffic from
specific geographic
areas on extremist Web
sites
MOE/
MOP
Difficult data collection; validity
is questionable; possible
connection with end-state
Analytic Measures for Counterterrorism and Counterinsurgency 469
personal-level data and may tempt authorities to push the limits of per-
sonal privacy laws. It may be useful to aggregate group-level social net-
work data to find the number of groups with known contacts to terror-
ists or places where terrorist influences are significant, or the number of
at-risk groups that seek connections with known terrorist groups. Such
a metric would be valid for gauging the effectiveness of efforts to social-
ize potentially radical groups and for gauging the basic risk of the for-
mation of groups with intentions to commit terrorism in a given area;
there is agreement in the literature that socialization processes are a
necessary precondition for radicalization. Measures of at-risk socializa-
tion may also be directly linked to perceived duty and perceived or real
rewards in the radicalization process. ese concepts can be measured
only by personal interviews or intelligence and should be understood in
light of the fact that rewards may differ even within a single group. But
increasing levels of perceived or real rewards combined with increasing
numbers of at-risk individuals or groups may indicate that radicaliza-
tion is on the rise.
Desire for Systemic Change
Factors related to widespread desire for systemic change are widely rec-
ognized as being important to driving radicalization and terrorism, but
it is difficult to directly measure any form of personal desire. Polls of
at-risk populations, structured interviews with group opinion leaders,
or interrogation reports taken from suspected terrorists may illuminate
common points of desire for change among radicals. Monitoring the
relevant issues and the strength of these desires may be valid MOEs for
shifts in government policy aimed at the radicalization process or CT/
COIN psychological operations (PSYOPs). Polls that show a desire for
Sharia law and reveal the strength of Muslim identity may also be inte-
gral measures linked to the radicalization process when combined with
measures of group processes and rewards. Qualitative indicators of a
desire for Sharia law may also be relevant; notable absence of cigarettes
or alcohol in stores may actually be a superior measure of radicalization,
although only where open insurgency is not occurring, because it could
also measure the ability of insurgents to coerce the population. Also,
the numbers of riots, rallies, and protests may be powerful measures of
470 Social Science for Counterterrorism: Putting the Pieces Together
a desire for political change (Tanham, 1988), although they may con-
stitute only MOPs, because they do not directly link to the end-state of
terrorism and available avenues of open protest may in fact decrease the
risk of terrorism. It has been shown that the desire for systemic change
is a potent factor leading to terrorism when combined with group pro-
cesses and rewards, so measures of the desire for change should be con-
sidered in tandem with measures of these two associated concepts. But
group processes and rewards may also be analytically powerful when
combined with perceived duty, which is discussed below.
Duty
Duty in this context is conceptually both a desire to exact revenge and
a perceived need for a collective defense of Islam, so measures of these
concepts are likely to be of interest to counterterrorism policymakers.
ose who have experienced personal attacks against loved ones, those
who are likely to have post-traumatic stress disorder (PTSD), and those
who have been convinced of a duty to defend Islam against threats
are clearly more likely to engage in terrorism when other critical fac-
tors are present. It is difficult to conceive of operable measures to cap-
ture changes in these factors, even if specific populations are targeted.
However, ideal proxy measures that would be relevant to these con-
cepts would include the number of civilians killed or wounded during
combat and changes in Internet traffic to extremist Islamic Web sites.
Although it is difficult to develop measures to monitor radicaliza-
tion processes, these factors are critical to pinpoint changes in an indi-
vidual’s propensity to become a terrorist within certain populations
and thus are most critical in domestic campaigns against homegrown
terrorism or for foreign governments whose populations emigrate to
join terrorist groups elsewhere. Policies to counter radicalization will
be inherently difficult to assess in this manner, but the metrics listed
above may shed some limited light on how well government actions are
affecting movements toward extremism and terrorism.
Analytic Measures for Counterterrorism and Counterinsurgency 471
Decisionmaking Factors
Factors that influence group decisionmaking in this context should be
considered with the group’s decisionmaking unit as the unit of analy-
sis, because this level of command is relevant to the proximal deci-
sions of whether, how, and when to conduct an attack. Because a large
number of factors have been determined to be substantial contribu-
tors to group decisionmaking (see the chapter by Jackson), I will con-
centrate here on the most crucial factors that both lead a group to
certain decisions and may be measurable longitudinally. Further, the
topic of terrorist decisionmaking is fundamentally different from the
topics discussed above, because the outcome of decision processes—
behaviors—are observable, whereas the factors that contribute to deci-
sions are more difficult to measure. us, I will first consider mea-
sures of relevant outcomes that may follow decisionmaking processes
and determine their utility in understanding how terrorist or insurgent
groups operate. en I will examine critical and observable factors that
lead to decisionmaking and evaluate the potential use of these mea-
sures. For decisionmaking, the desired end-state is being able to influ-
ence the decisions of existing terrorist or insurgent groups. e ele-
ments that I consider in subsequent paragraphs are outcomes, lessons
learned with outcome measures, decisions, organizational structure,
and resource stocks. (See Table B.3.)
Outcomes
For our purposes, outcomes of interest at the operational and tacti-
cal levels primarily involve attack data; these data include attack types
(numbers of fighters, tactics, and weapons), attack locations or targets,
and numbers of attacks across time. ese simple MOOs are tracked in
quarterly reports about the war in Iraq made by the DoD to the U.S.
Congress in the forms of weekly attack trends broken down by attack
type (infrastructural attacks, improvised explosive devices, small arms
fire, indirect fire) as well as by attack levels for a given period broken
472 Social Science for Counterterrorism: Putting the Pieces Together
Table B.3
Decisionmaking Measures
Concept Being
Measured
Measure
Specification
Measure
Type Comments
Decision
outcomes
Attack data (e.g.,
types, size, targets,
numbers of attacks)
MOO Easy or moderate data
collection; should be analyzed
alongside calendar events, CT/
COIN action, and qualitative
indicators of second- or third-
order effects
Decisions Financial activity (e.g.,
revenue sources,
expenditure types)
MOO Very difficult data collection;
reliability is questionable;
other indicators of group
processes are also necessary
Organizational
structure
Organizational
structure type
MOP, MOE,
or MOO
Very difficult data collection;
reliability is questionable;
can be MOP, MOE or MOO
depending on how it is used
Resource stocks Order of battle (e.g.,
organization structure,
weapons, training,
leadership)
MOP or
MOE
Difficult or very difficult
data collection; sensitivity is
questionable; connection with
end-state may be mediated by
many other factors
down by Iraqi province (U.S. Congress 2005, 2006, 2007, 2008).
David Galula once noted that the size of guerilla engagements and the
guerillas’ ability to conduct complex attacks against counterinsurgent
convoys were foremost among important indicators of success (Hosmer
and Crane, 1962), and changes in these outcomes may signal impor-
tant shifts in insurgent strategy according to key decisionmaking fac-
tors. Subsequently, systems analysts at the DoD during the Vietnam
War were fixated on attack data. A number of commentators have
noted the disastrous use of enemy body count as an MOE of COIN
operations at the time (Army, 2006; Corson, 1968; Murray, 2001;
Long, 2006). Despite this improper use of one attack data metric, it
must be acknowledged that changes in attack types have accurately
pinpointed changes in insurgent strategy, such as with the insurgent
group Farabundo Marti para la Liberacion Nacional in El Salvador
in the early 1980s (U.S. General Accounting Office, 1991) as well as
with the North Vietnamese in the late 1960s (ayer, 1985). us,
Analytic Measures for Counterterrorism and Counterinsurgency 473
well-crafted attack metrics may shed considerable insight into group
decisionmaking as an outcome.
Lessons Learned with Outcome Measures
Despite past analytical successes with attack data, past failures may
be informative about the use of metrics by analysts and policymakers.
ayer’s treatise War Without Fronts demonstrated that the operational
environment must be considered in tandem with threat levels; he noted
the destructive lack of awareness in the U.S. Army that clear seasonal
attack trends were evident in the rainy seasons (ayer, 1985). Attack
trending based on calendar events has also been present in Operation
Iraqi Freedom; attack levels have consistently spiked during Ramadan
and during national elections (U.S. Congress, 2008). It is also intuitive
that attack data will differ depending on CT/COIN operations and
shifts in policy, such that the concept of “net assessment”—or the over-
all balance of insurgent and counterinsurgent forces—should be taken
into consideration when assessing trends in attacks. Furthermore, anal-
ysis of allied strategic operations in World War II showed that second-
ary effects should be taken into account when analyzing the effective-
ness of military operations (Roche and Watts, 1991). us, it may not
be sufficient to use enemy attacks as an analytic measure alongside
known calendar effects and CT/COIN actions; second- or third-order
effects may be present and quantifiable and should also be taken into
account. In this light, it must be acknowledged that, although attacks
reflect group processes, they are higher-order outcomes that depend
also on other processes and are useful in certain limited contexts and
lack the precision of a true MOE.
Decisions
Some group decisions and outcomes are less easily observable, beyond
simple attack planning, but may be interesting in helping us under-
stand the choices being made by the decisionmaking unit. e deci-
sonmakers in the group may have to decide whether to solicit funds
from outside the immediate group structure or to generate their own
revenues. Such decisions may be observable at certain points in time,
given that relevant intelligence is being collected and becomes avail-
474 Social Science for Counterterrorism: Putting the Pieces Together
able. If it is known that a group decisionmaking unit is accepting funds
from a higher node of the organization, it may indicate a lower risk
tolerance, because generating revenue independently may increase the
odds of capture. Other group decisions may be of interest to policy-
makers, but it is difficult to attribute changes in these outcomes to par-
ticular policies because of the complex processes at play.
Organizational Structure
With some important outcomes of decisionmaking considered, I now
examine metrics for important factors that contribute to group deci-
sionmaking processes. e first key factor to consider is that of organi-
zational structure; this feature of terrorist groups feeds into the other
decisionmaking and behavioral factors. A group’s organizational struc-
ture is difficult to gauge, and, as al-Qaeda has demonstrated, different
structural elements may exist simultaneously. Appraising group struc-
ture is difficult in CT/COIN because of the clandestinity of such orga-
nizations; the potential for denial and deception makes high-quality
multisource intelligence all the more necessary to accurately character-
ize structural forms. Developing a metric that captures group organi-
zation is difficult as a result of the artificial nature of the dichotomy
between formal and networked organizations, but composite proxy
metrics that combine aspects of group organization may be assembled
from ordinal scales. Analysis of the literature indicates that organiza-
tions should be categorized on such criteria as division of labor, levels
of decisionmaking authority within the organization, geographic speci-
ficity in operational command, and singularity of decision processes
(that is, authoritarian versus participatory). ese elements of group
organization could be scaled from 1 (lowest) to 5 (highest), such as in
the HES. Strict guidelines should be developed so that specific crite-
ria are attributed to each step in the scale, and these ratings should be
vetted extensively. Also, such a composite proxy metric would need to
be constantly reassessed and evaluated for validity as new intelligence
and analysis become available.