196 ACCESSING AND GENERATING DATA
compilation of the number of protests were able to be attained, interpreting its meaning is not
straightforward: a high (or a low) frequency of protests in a country may be due to a high (or low)
level of protest everywhere or a very high (or a very low) level in only one part of the country. The
summary number, also, says nothing about the territorial problem of counting protests in various
parts of the country. Thus, if there is not accurate information on the territorial distribution of
protest, the data set will be of little use for predicting internal conflict.
If an event is counted as a protest only when it is deemed “successful,” meaning that a signifi-
cant proportion of the target population participates or the protest remains peaceful, then how are
territorial differences in participation to be combined to determine whether a protest event can be
said to have occurred? Again a myriad of examples might be cited. The impact of a twelve-hour
bandh in the state of Manipur called by the All Manipur Students Union (AMSU) was described
in The Hindu as follows: “The strike had a total impact in the Sadar hill and Saikul sub-divisions
of Senapati district. However, there was no impact at Ukhrul and Tamenglong districts. In Chandel
district, there was partial impact.”^46 If “having an impact” is required for the event to count as
having occurred, then one is faced with the problem of deciding how to combine “no impact” in
some areas with “total impact” in other areas to determine whether the event should be counted.
Such a report would require a guess about whether the level of participation met Banks’s defini-
tional requirements and whether the target was the national government’s policies or authority.
As noted, Banks’s source of information is the New York Times, which is unlikely to report the
kind of detail contained in The Hindu noted above. So, neither would this problem occur nor
would any of this information be obtainable from Banks’s frequencies of protest.
A similar problem arises in assessing whether to count as a protest what happened when the
People’s War Group, a radical guerilla group found in the rural districts of several states, called a
statewide bandh in the state of Andhra Pradesh in April of 2003 to protest the killing of one of its
senior leaders. “Normal life” was affected in rural areas, but there was little impact in urban
areas.^47 This territorial problem is compounded when information is incomplete or in disagree-
ment and when organizational and territorial support are combined. The Hindu reported on a July
2001 action against mining in the Western-Ghats mountain area, saying that it “evoked an en-
couraging response if the overwhelming voluntary support being expressed to it by various orga-
nizations and institutions is an indication,”^48 while the Deccan Herald said it “failed to evoke
even a dismal response in Bellary, Raichur and Koppal, but seems to have in Shimoga and
Chikkamagalur districts.”^49
Information on territorial variation in protest, both within India and within parts of India, is not
provided by the summary frequencies, yet such information is critical to an attempt to understand
the nature and impact of various forms of protest. A frequent answer to the challenge presented
by territorial variation in the response to calls for protest is to say the call “evoked a mixed
response.” If the “mixed response” is counted as a protest event, information on important terri-
torial variation is excluded—as may be other important information as well. The UP bandh, re-
ferred previously, was reported to have met with “a mixed response.” More specifically, a report
on it in the Indian Express stated that the home secretary, A.P. Singh, said that “80 per cent of the
shops remained closed and life was affected in the plains but conceded that the situation was ‘just
the reverse’ in the hills, where a student-sponsored anti-reservation stir had gained momentum
prompting the ruling Samajwadi Party-Bahujan Samj Party combine [combination-ed.] to call
Tuesday’s bandh as a counter.”^50
In addition to their absence of information about the uneven territorial distribution of protest
events, Banks’s frequencies provide no clue to the impacts of protest across multiple levels of analy-
sis. Only for strikes does Banks explicitly state that the protest must be aimed at “national government