CODING INSTRUCTIONS FOR VIOLENT EVENTS: PREPARED BY JOHN M. RICHARDSON JR. WITH THE ASSISTANCE OF THE PROJECT STAFF
DATE OF THIS REVISION: JULY 9, 1990
SUMMARY OF TOPICS
BACKGROUND
MEASURING VIOLENCE
WHAT IS A VIOLENT EVENT?
RECORDING VIOLENT EVENTS
CODING VIOLENT EVENTS
VIOLENCE CODING: QUICK REFERENCE GUIDE
VIOLENCE CODING: CATEGORIES AND AMPLIFYING COMMENTS
DURATION
LOCATION
INTENSITY OF VIOLENCE
SIZE
BACKGROUND
The work that you are doing is part of a ten year project that I initiated in 1984, focusing on a comparative study of the political economy of violence. By the political economy of violence, I mean, the way in which economic phenomena contribute to political violence and the way that political violence then "feeds back" to cause changes in economic phenomena.
The project has two objectives:
1. Develop a comprehensive theory that will explain patterns of violence, both anti-government violence and "official violence" (repression) in third world nations.
2. Answer the question: "How can nations meet the basic needs of their people and develop, economically and socially, with a minimum of violence?"
The present research is intended to answer four major questions:
1. Why does violence occur?
2. When and where is violence most likely to occur; with what duration and intensity?
3. What are the costs and benefits of violent versus non- violent alternatives to (a) seeking and maintaining power and (b) managing development?
4. What can be done to minimize use of violence as a tool of political action; when violence is used, what can be done to limit its duration and intensity?
To provide additional information and background, you have been provided with a copy of a recent article that appeared in the December issue of Futures, "Violence and Repression: Unexamined Factors in Development Planning." This article describes the theory underlying the work and applications to Argentina and Mexico.
MEASURING VIOLENCE (I now use the concept, violent political conflict. JMR )
A very important part of the research, and the work in which you will be involved is the quantitative measurement of violence. It is important to measure violence quantitatively, in order to compare levels of violence in different historical periods and in different countries. While quantitative analysis has its limitations and must be supported by careful historical analysis, it is, nonetheless, quite useful.
The quantitative index of violence we will be using was developed by a Russian-emigre sociologist who taught at Harvard University in the 1930s, Pitrim Sorokin. I selected Sorokin's index after surveying a number of different alternatives, because Sorokin's was also interested in long-term trends and in studying problems for which data was not readily available. (For example he studied cycles of violence in the Roman Empire and in Europe from the birth of Christ through the 19th century.)
Sorokin's index is developed by coding violent events. A violent event is a historical occurrence, that involves a demonstration or protest with political overtones (it could be either pro or anti-government) and, possibly, physical destruction of property, injury of persons or death. Violent events are identified by reading historical literature or, in the case of your research, by reading newspaper reports. Each violent event is coded according to its:
Ô DURATION: the length of the event
Ô LOCATION: the size and significance of the place where the event occurred
Ô INTENSITY: the amount of death and destruction that occurred; the overall impact of the violence
Ô SIZE: the number of people participating in the violent event
WHAT IS A VIOLENT EVENT?
A violent event is a politically significant occurrence, that falls outside the established legal procedures of political life and usually involves violence. Usually violent events are expression some sort of protest against the government or the established order of things. For example:
Ô Assassinations of government or party officials
Ô Attacks on destruction of the symbols of government (attacks on police stations, post offices or Kacheries [government district offices], burning of trains or SLTB [Sri Lanka Transport Board] busses, etc.)
Ô Politically motivated strikes.
Ô Riots or demonstrations with political overtones
Sometimes violent events may not be specifically anti- government. For example the anti Tamil riots in the summer of 1983 would be classified as a violent event, even though not explicitly anti-government. Attacks on opposition party officials that frequently occur during election campaigns would be counted, even though they are "pro government."
Strikes are an example of an event about which there may be ambiguity. A strike would not be included if it were directed against a private corporation and was only motivated by economic objectives (eg. higher wages, better working conditions, etc.) However in Sri Lanka, most strikes do have political overtones because (a) the organizations struck are government agencies or affiliated with the government (SLTB, hospitals, universities, etc.) and/or (b) the labour unions have political affiliations and political objectives. If you are uncertain about whether or not to include a strike, fill out a coding sheet for it and explain the ambiguity in a "comment" entry. I will make the decision as to whether to include or exclude it.
Occasionally, what is classified as a "violent event" may involve a protest or demonstration, but no violence. Sorokin's coding for intensity of violence provides a "non violent" category. Examples would be non violent work stoppages (with political overtones), boycotts and demonstrations, so long as they did not involve violence. A "satyagraha" [demonstration involving non violent civil disobedience in the Gandhian tradition] that did not deteriorate into violence would receive such classification. The massive funeral of Vijaya Kumaratunga in Colombo, which, fortunately, resulted in no violence, would be an example of such an event. [Kumaratunga, the movie star politician husband of Sri Lanka's current President, Chandrika Bandaranaike Kumaratunga, was gunned down by Buddhist Sinhalese extremists in February, 1988. More than 500,000 persons attended his funeral.]
RECORDING VIOLENT EVENTS
The process of measuring violence can be divided into two parts. First each violent event is recorded on a violence coding sheet. Second, each event is given a numerical code for duration, location, intensity and size. This section discusses how coding sheets are used to record violent events. The next section discusses coding procedures.
Remember that your coding sheets will be transcribed by someone else. Your entries must be clear and easy to read. And remember that what will be transcribed is exactly what you write.
The procedure for recording violent events will be described by providing instructions for filling in each blank in the CODING SHEET FOR VIOLENCE EVENTS.
DATE:
In this blank write the date that you filled out the coding sheet. Example 12/5/88 for 12th May, 1988.
RESEARCHER:
Write your name. Since we have only a small number of researchers, your last name or even a clearly identifiable abbreviation of your last name will be sufficient.
Be sure you write your name on each coding sheet. To save time you might want to fill out a number of blank coding sheets with your name when you start your work.
EVENT DATE START(MO/DAY/YR)
Write the month, day and year the event began. Sometimes the newspaper account will be ambiguous about this or there will be conflicting reports. Getting the precise day is not essential, however it is quite important to get the correct month. If no day is specified, assume the event occurred on the day before the date the article describing it appeared. If the accounts of an event differ in two papers, choose the account that appears to be more reliable and note the discrepancy as a COMMENT.
I will always check every sheet and I will pay particular attention to COMMENTS. So if there is a problem, I will either consult with you or resolve it myself.
EVENT DATE END(MO/DAY/YR)
Write the month, day and year the event ended. In most cases the events will be relatively short, lasting a only a day or so. However occasionally, events may be longer. For example a hartal (strike) might last for several days or even several weeks. The 1983 riots in Colombo and several other cities lasted for several days. If the event remains the same, i.e. a strike that starts in one place and spreads, it should be coded as a single event, even though it may be described in several news reports over a period of time.
In our experience of coding, so far, there have been only a few long events, so you should feel free to discuss any ambiguities in detail with me. Kandy researchers should discuss the matter with Mr. Jayewardene who will then consult me on one of his regular visits to Colombo.
One type of an "event" that has caused a bit of a problem is campaigns that appear to be coordinated but involve a number of specific disturbances. For example, at the time of the provincial council elections in the Central and Southern part of Sri Lanka, the JVP initiated a series of bus burnings, rail track sabotages, train shootings attacks on police stations and so forth. While these were part of a coordinated campaign, the mathematics of the coding scheme make it necessary that each separate incident be recorded separately.
Another ambiguity arises when an event is initiated by the security forces. In general such events are treated as "official violence" and are not recorded as violent events. (Official violence is the subject of a separate coding manual.) An example of such an event would be a sweep initiated by the IPKF against LTTE strongholds in the jungle or the Spring 1987 offensive by the Sri Lankan security forces against the LTTE. If the police break up a legitimate political meeting (by a baton charge, for example) this would be an official violence event unless it was reported that the police intervention had been provoked. In the latter case a judgement would have to be made as to the source. A report of "provocation" in a government controlled newspaper (Daily News) would be given less credibility than a similar report in an independent newspaper.
(Island;Sun;Times.)
DURATION
Write the length of the event. For less than a day, write "LT day." In coding length, an event is coded as "momentary" if it is less than a week in length, so you don't have to worry about carefully distinguishing between, say two or three or five days. Check the coding instructions below for DURATION if you are unclear about whether some ambiguity in the length of an event will be significant for the numerical coding.
LOCATION: PROVINCE
Write the name of the PROVINCE in which the event occurs.
For coding location, both province and city/village/area, the next category. You should have a detailed map of Sri Lanka, that also lists and helps you to locate villages, cities and towns.
For the purposes of this study, the Eastern and Northern provinces will be treated as separate, even though they have been amalgamated (at least temporarily) by the Peace Accord.
LOCATION: CITY/VILLAGE/AREA
Write the name of the village, city or area, plus additional information that will help me to locate it on a map, should I choose to do so. Ideally, the entry should include (a) village name and (b) district. Additional information can also be included. For example:
Urutrupurrum, Jaffna District, near Killinochche.
If the location is near a middle sized city, coded five or higher, it is not necessary to include the district, for example:
Panditerrippu, near Jaffna.
Be careful not to code "suburbs" that are part of a larger city as separate units. You may want to identify them separately, however you should use the connective, "in" rather than "near". For example: Wellawatta, would be described:
Wellawatta, in Colombo
because it is within the city limits of Colombo. On the other hand, Dehiwala-Mt. Lavinia would be described:
Dehiwala-Mt. Lavinia, near Colombo
because it is outside the city limits.
DESCRIPTION OF EVENT
Write a very brief description of the event. When describing an event, remember that it is going to be coded on a computerized "spreadsheet" along with a lot of other events. Also remember that it is going to be transcribed by a data processor, who is going to be instructed to transcribe EXACTLY what you have written, without interpretation.
Here is an example of an event description that is a bit too long:
The terrorists kidnapped an American couple from their Gurunagar home threatening to kill them unless Rs 50 million in gold was delivered to the Tamil Nadu Government.
A more useful description of the same event would be the following:
Terrorists kidnaped American couple. Threatened to kill them unless Rs 50 million paid to Tamil Nadu govt.
Note that you don't have to include the location of the kidnapping because that has already been described in the LOCATION entry. The fact that the ransom was requested in gold is more detailed information than is needed.
Here are some examples of good event descriptions:
PC attached to Kalmunai police station shot dead by terrorists.
Deputy leader of PLOTE, R. Jayachandren, shot dead by security forces.
Terrorists attempted attack on Sinhalese school and Buddhist temple.
Some ambiguity is created by the fact that newspapers generally refer to the LTTE [Liberation Tigers of Tamil Eelam] , and sometimes other Tamil separatist groups, as "terrorists" while the JVP [Janata Vimukti Peramuna - People's Liberation Front in the Sinhala language] (and perhaps other southern terrorist groups) are described as "subversives". If it is clear from the context of the article which terrorist group is being described (e.g. LTTE, PLOTE, EROS, EPRLF, etc.) or which subversive group (JVP or other) that you should use the specific name. If it is not specific, then use the more generic term "terrorist" or "subversive" used by the article.
SOURCES
Write the name, date and page number of the newspaper from which you recorded the event. The name of the paper can be abbreviated; CDN for Ceylon Daily News, IS for Island. A typical entry would be:
CDN, 5 May/84, p.5.
In Sri Lanka, the weekend papers have a different name from the daily papers but are essentially the same paper. If your source is a weekend paper, record the title and put the name of the corresponding daily paper in parentheses. For example:
Observer (CDN), 8 May/82, p. 14
Whenever possible, you should be checking the event in more than one paper. Some events are listed in all papers, some in only one or two. When the event appears in two or more sources, you should be sure to list both. For example:
IS, 3 Jun/87, p. 4; SUN 4 Jun/87, p. 1.
COMMENTS
COMMENTS entries are for two purposes, (a) brief amplifying or contextual discussions of a violent event or (b) descriptions of problems or ambiguities in recording or coding the event that I need to address and/or resolve. Comments will not, in general be recorded on the computer so you have more flexibility in length and format. The principal requirement is that you communicate clearly.
Here is an example of the first category, amplifying or contextual discussion:
The DESCRIPTION OF EVENT was as follows:
Four trade union leaders stage "fast unto death" protesting against the refusal of their employer, the Port Commission to allow them to address union members during lunch hour.
The COMMENT was:
Fast called off after assurance given by Minister.
Here is an example of the second category of COMMENT, description of an ambiguity:
The DESCRIPTION OF EVENT was as follows:
11 persons injured and 5 houses burnt in clash between two castes on Sept. 16. One of the injured dies in hospital. Further clashes erupt time and again. More injuries reported. Army deployed to maintain peace. Curfew imposed. Last incident reported on Nov. 11.
The COMMENT was:
Caste problem. Is this relevant?
The event is relevant, even though the initial political overtones may seem minimal because it was a major disturbance, involving violence and destruction of property and eventually army intervention.
Incidently, this also provides an example of an event with a LENGTH of three months. The coder noted that, while there were periodic outbreaks, they were all part of the same basic event. Be alert for longer-term events that have this character.
CODING VIOLENT EVENTS
Each violent event is numerically coded according to its DURATION, LOCATION, (intensity of) VIOLENCE, and size. For easy reference, I have summarized coding instructions for each category on the next page. You may want to tear out this page and keep in front of you as a ready reference when you are coding. Amplifying comments for the respective categories and codes are provided on subsequent pages.
VIOLENCE CODING: QUICK REFERENCE GUIDE
DURATION*
1 Momentary, up to one week, where only one short-time shock is noticeable
3 A longer disturbance, one week to more than one month, but less than two months
5 Two months up to eleven months (less than one year)
10 One year to 23 months (one year to almost two years)
15 24 months to 35 months (two years to almost three years)
20 36 months to 47 months (three years to almost four)
25 48 months to 59 months (four years to almost five)
30 60 months (5 years to almost six)
LOCATION
1 A settlement with population of 5,000 or below, or an area of density less than 1 person per 1 square kilometer
3 Settlements between 5,001 people up to 25,000
5 From one to five medium town(s), population of 25,001 to 250,000.
10 Jaffna or more than five medium towns or one province
20 Jaffna and more than five medium towns and/or part of capital city and/or at least two provinces
40 All of capital city or at least four provinces.
60 All of capital city and spread over at least three other provinces
80 A disturbance where almost the whole country is involved (at least 70% of the population)
100 The whole country
(INTENSITY OF) VIOLENCE
1 Non-violent. Events such as demonstration, work stoppages, boycotts -- so long as they did NOT result in deaths, shootings or destruction of property.
3 Slight, dependent upon the phrasing. For example a "riot," "robbery" or "hijacking" would be considered a 3 as long as there were no deaths, shootings or destruction of property.
5 Violent, involving destruction of property--fights, lootings, sackings. Kidnappings and hostage taking.
6 Violent, involving deaths of people
SIZE
1 A few individuals -- ie. murder plots
3 A small group -- 10 or more, up to 100 individuals.
4 A large group, but not a social class -- 101 to 5,000 people
5 A large social class -- ie. a major political party, extensive occupational, economic, racial, national groups, a religious denomination, or any group of 5,001 individuals or more if they share a common grievance
10 all adults
VIOLENCE CODING: CATEGORIES AND AMPLIFYING COMMENTS
DURATION*
1 Momentary, up to one week, where only one short-time shock is noticeable
3 A longer disturbance, one week to more than one month, but less than two months
5 Two months up to eleven months (less than one year)
10 One year to 23 months (one year to almost two years)
15 24 months to 35 months (two years to almost three years)
20 36 months to 47 months (three years to almost four)
25 48 months to 59 months (four years to almost five)
30 60 months (5 years to almost six)
When Sorokin conceived this index, he was interested in covering a much longer historical period. His references were chronicles and history books, not newspapers. From that vantage point some "events" were viewed as much longer duration than they appear from our vantage point and sources.
But do be alert for events that are reported over several days or weeks and, thus should be given a longer duration code than one. This may not be immediately apparent from a news story describing one event. Possibly a longer "background" piece may reveal the connection. If you have any questions, be sure to make a COMMENT entry and call the matter to my attention when we meet.
If an event has a duration of longer than one month, the event should be recorded for each of the months in which it occurs. Consider, for example, a strike that begins on 28th January 1985, continues throughout the month of February and then is settled on March 12. The strike would be recorded as event for January, February and March. In January, the duration code would be 1 and the event would be recorded as running from January 28 through January 31. The duration code would be 1. In February the event would be recorded as beginning on the first day of the month and running for the entire month. The duration code would be 3. In March the event would be recorded as beginning on the first day of the month and concluding on the 12th. Since the event was of duration longer than one week in March, a duration code of 3 would again be assigned.
When an event runs over several months this should be noted in the event description (not as a comment.) The reason is that event descriptions are recorded in the computerized data base while "comments" are not. In the above example, the event description for January might read:
Elephant house workers on strike over firing of workers involved in political demonstration. (Strike continues until March 12.
LOCATION
Coding location has posed some of the most difficult problems in this work. Manjula de Silva has now developed an extensive list of towns and cities with corresponding codes. While these are based on 1970 census material, they should be more or less applicable for later periods as well. Those of you coding the post 1980 period should check this list against 1980 census data to see if there are any medium towns that should be bumped to an "important city" on the basis of their population. Jaffna is classified as an important city despite the fact that its population is less than 250,000.
When coding LOCATION, if you're not sure into which one of the categories the event should be placed, bump it down to the next smaller category.
1 A settlement with population of 5,000 or below, or an area of density less than 1 person per 1 square kilometer
Locations that do not fall into the larger categories will always be placed in this category. Events that occur on the open sea, unless in the harbor of a major town will also be classified as a 1. The following are some examples of small towns.
Agalawatte Akkaraipattu
Andigama Bandarawela
Dambulla Dehiwatte
Habarama Hali-Ela
Haputale Kadugannawa
Kalpitiya Kayts
Kekirawa Kolikai
Koslanda Mahihangana
Maho Mankulam
Mawathagama Nedawachchiya
Miragama Moneragala
Nattandiya Nilaveli
Padaviva Passara
Palmate Rakwana
Uruwela Weyangoda
Warakapola Welimada
Wellawaya Wennapuwa
Yakkala
3 - Settlements between 5,001 people up to 25,000.
The list that follows includes most of the settlements that fall within this category.
Ahangama Akuressa
Aluthgama Ambalangoda
Ambalantota Amparai
Avissawella Balangoda
Balabitiya Bandaragama
Bentota Beruwela
Chankannai Chavacachcheri
Chilaw Chunnakam
Darga town Dehiowita
Deniyaya Devinuwara
Dodanduva Elpitiya
Eravur Gampaha
Gampola Hambantota
Hatton-Dickoya Hingurakgoda
Homagama Horana
Ja-Ela Kalmunai
Kandana Kankesanthurai
Kantalai Kaththankudi
Kegalle Kilinochchi
Kinniya Kochchikade
Kuliyapitiya Madampe
Manipay Mannar
Matugama Mawanella
Minuwangoda Mullativu
Mulleriyawa Muttur
Nawalapitiya Neliyadi
Nuwara Eliya Pandatharippu
Pelmadulla Pilyandala
Point Pedro Polgahawela
Polonnaruwa Puttalam
Ragama Rambukkana
Ruwanwella Samanthurai
Seeduwa-Katunayake Tangalle
Urimpirai Velvettiturai
Vavuniya Wadduwa
Wattala-Mabola Watugedera
Weligama Welisara
5 - Medium towns, population of 25,001 to 250,000.
The following towns/cities fall within this category:
Anuradhapura Badulla
Battaramulla Batticaloa
Dalugama Dehiwala - Mt. Lavinia
Galle Hendala
Kalutara Kandy
Kelaniya Keselwatte
Kotte (except that, beginning in 1983 (after completion of the parliamentary complex) events occurring in Kotte should be classified as 20 (part of capital city).
Kurunegala Maharagama
Matale Matara
Moratuwa Negombo
Panadura Peliyagoda
Ratnapura Trincomalee
10- Jaffna or five or more medium towns or one province
Events occurring in Jaffna, throughout any entire province or in ten medium towns should be classified as 10.
Anuradhapura, Kandy, Galle, Dehiwala-Mt. Lavinia, Moratuwa, Kelaniya, Batticaloa, Trincomalee, Matara.
20 - Jaffna and more than five medium towns and/or part of capital city and/or at least two provinces
Events occurring in Colombo that are not city-wide fall within this category.
40 - All of capital city or at least four provinces.
There is some ambiguity about use of the word "all". However a good test is the imposition of a city wide curfew (e.g. following the implementation of the Gandhi Jayewardene accord and the bombing of the Colombo bus station) or a total breakdown of law and order throughout the city (e.g. during the summer, 1983, riots.)
To be classified as "at least four provinces", there should a set of clearly related events scattered throughout rural areas and a number of the towns of four provinces.
60 - All of capital city and spread over at least three other provinces
80 - A disturbance where almost the whole country is involved (at least 70% of the population)
100 -The whole country
I am not aware of any events that would fall in either of the last two categories. Although I have not read the accounts carefully, possibly the demonstrations following the peace accord might be classified as a 60.
(INTENSITY OF) VIOLENCE
In general, the key to coding the intensity of violence is to read the description of the event carefully looking for key words that differentiate one level of intensity from another. Be sure to include these words in your description of the event so that your coding can be checked. So far, coders do not seem to have had much difficulty with this classification.
1 Non-violent. Events such as demonstration, work stoppages, boycotts -- so long as they did NOT result in deaths, shootings or destruction of property.
3 Slight, dependent upon the phrasing. For example a "riot," "robbery" or "hijacking" would be considered a 3 as long as there were no deaths, shootings or destruction of property.
5 Violent, involving destruction of property--fights, lootings, sackings. Kidnappings and hostage taking.
The destruction of property can be minor so long as destruction is reported. For example any "break in" would be coded as a 5.
6 Violent, involving deaths of people
If a death occurs after the event, but as a consequence of it, it would still be classified as a 6. For example, someone might be injured and taken to the hospital. If a later article reported that he died of his injuries, that would count as a death associated with the event.
SIZE
1 A few individuals -- ie. murder plots
3 A small group -- 10 or more, up to 100 individuals.
Distinguishing between 1 and 3 events sometimes creates some problems where it appears to be a group effort, but the size of the group is not reported. An event should always be reported as a three if it involves a clash between terrorists and the security forces, policy or IPKF. Remember to count the people affected by event as well as those instigating the event.
4 A large group, but not a social class -- 101 to 5,000 people
5 A large social class -- ie. a major political party, extensive occupational, economic, racial, national groups, a religious denomination, or any group of 5,001 individuals or more if they share a common grievance
One of the questions that has come up is the coding of hartals [general strikes] that may affect an entire town or area. In general, these should be classified as a 5 rather than a 10, even though most of the populace may be affected. The reason is that the hartals, in the North at least, tend to be imposed somewhat arbitrarily by one group or another, without mass support. A hartal with wide mass support would be classified as a 10.
10 All adults
The meaning of "all" is somewhat ambiguous. However an event should receive a 10 if a very large proportion of the population is touched by it. An event leading to a city-wide curfew, for example, would be classified as a 10.
REMEMBER THAT AN QUESTIONS RELATING TO YOUR CODING OR
AMBIGUITIES IN YOUR CODING SHOULD BE NOTED TO MY ATTENTION IN A
COMMENT ENTRY. ALSO, FEEL FREE TO CALL ME IN THE EVENINGS IF I
CAN RESOLVE A QUESTION RELATING TO YOUR CODING.