It all started with the observation that all data concerning public contentious interactions between protestors and police were problematic:
Drawing on the work of Clark McPhail, we thought we would try something.
- memory is biased and partial
- newspapers don't cover everything
- actors lie to put forward their best image
Drawing on the work of Clark McPhail, we thought we would try something.
On May 17, 2012 we were given on opportunity to try out our idea. At this time in Chicago, Illinois there was going to be held a meeting of the G8/NATO as well as an announced protest of the event. We decided that we were going to cover the event comprehensively. We thought we would scrape all tweets, news reports and where possible acquire the raw footage taken by the media (they would not give this up however) as well as news stories both published and those put on the wire. In addition to this, we grabbed all press releases and lawsuits concerning the event (e.g., discussions about where the marchers could/could not go). We tried to conduct a survey of residents in Chicago following the event but could not obtain funding. In order to have something to compare all this information to, we decided to actually go to the event and video tape grids (like that discussed above). Every 30 minutes, individuals would do a 360 turn with their video camera in their designated area. In addition to this, we obtained footage from people that we met at the event (protestors and those videotaping them are not the most trusting people and thus we only obtained footage from two people).
Objective set, we selected 15 students from the illinois and indiana areas as well as laid out where our people would go/stay during the event. As conceived, there were to be two protest sites as well as a march in between them. The grid for the protest sites was relatively easy to work out. This is shown below.
Objective set, we selected 15 students from the illinois and indiana areas as well as laid out where our people would go/stay during the event. As conceived, there were to be two protest sites as well as a march in between them. The grid for the protest sites was relatively easy to work out. This is shown below.
The grid for the march turned out to be a bit more complex. Essentially, we stretch out our team into groups of two for the length of the march. Unfortunately, this presumed that the people in the selected grid space would maintain their distance from the grids in front of/in back of/to the side of them, which was not the case. We ended up moving all over the place as we tried to follow the group of people that were in our grid.
From our effort, we discovered many things:
- Contention starts earlier than the event in question
- Contention lasts longer than the event in question as the battle ensues for documentation/memory as well as getting individuals released from jail:
- Protests and marches are not singular events but rather they are highly variable:
What did we end up with?
- A lot of tweets, news accounts and video footage
- A clear understanding that what actually happened, what was reported and what was remembered were three different things with some overlaps
- We realized that just covering the occurrence of events was perhaps one of the easiest but the least interesting. We needed to get at the variation in coverage, depictions of winners/losers as well as sense of who was believed to have escalated the conflict.
- But, mostly, we ended up with a lot of video footage and a large number of questions regarding what should be done with it. Returning to Clark McPhail and David Schweingruber proved a good jumping off point but we still needed more. This is where we went next.
A first effort to think through coding video
Coding Scheme Proposal for the Davenport and Moore G8/NATO Protest and Policing Project
Patrick Scott
For Dr. Will Moore Florida State University Friday, May 3, 2013
Introduction
The following is a coding scheme proposal for the Davenport and Moore G8/NATO Protest and Policing Project. In the initial stage of this project, researchers participated in the protest that occurred in Chicago, IL, in 2012 for the G8 and NATO summits that occurred there. The researchers, spread throughout the protest, recorded 17 different panoramas every fifteen minutes for three hours. This coding scheme represents a method by which we can take this quasi-sequential data and code into a format that can be used to answer various research questions. When completed, this data will be of great use for researchers in understanding the micro-relationships in protest-police interactions and media interpretations of violent clashes.
This scheme will proceed as follows. The first part will briefly present the research questions that I aim to code the data for. Whenever a coding scheme is made, understanding the questions that we are trying to answer is critical to make sure that the method we employ will allow us the best leverage on our questions. In the second part of this scheme, I will look at what behaviors and interactions in particular we want to capture from these videos. After that, in the third section, I provide a set of instructions for prospective coders, followed by criteria that a coder may use to discern what information to document and the best way to do so. I will then conclude with a few remarks regarding the importance and application of this data.
Objectives
There are two main objectives for this project. The first is gather data to model and deepen our understanding of the relationship and interaction between protesters and police forces that are tasked to manage them. The second objective is to investigate the factors that determine which events, in protests with protest-police clashes, are reported in news media and the impact that media has in framing perceptions of protests.
The interactions between protester and police are not fully understood. Protesters and police forces both have incentives to provoke the other to take aggressive action against them first, because the use of violence by the other side robs them of their legitimacy and credibility. Violent repression on nonviolent movements can increase the legitimacy of the movements, and signal the resolve of the movement (Chenoweth and Stephan 2011). In order to appropriately understand this interaction, how police might goad protesters into attacking them, and vice versa, requires detailed event data that can shed light on the steps that both sides take to get themselves close to a conflict with being the one to initiate it. The difficulty in obtaining this data, however, is that researchers who seek to study a protest’s interactions are part of a protest that could become violent and dangerous. The first step in this project, obtaining this dangerously acquired data, has been accomplished.
The second objective of this project is to understand the determinants of variations between actual events in a protest and the reported events. A protest’s efficacy as a tool of social change can be heavily influenced by the type of media coverage it gets (McLeod and Detenber 1999), but the factors that affect what is reported versus what happens in protests is less certain. Moreover, the recent ubiquity of social media means that information about the goings on in protests could become less exclusive to those in attendance; footage from the protests in Chicago revealed a large number of participants recording photos and video of the events. The opportunity provided by this protest allows a comparison between actual events, recorded by on-the-scene researchers, social media records of events, and news media reports. Understanding the evolving link between protest events and, eventually, public perceptions shaped through new or social media reports, could change the way we view the efficacy of protests as a medium of social awareness and change.
What Do We Want to Capture?
The videos taken at the Chicago protests provide a valuable source of information, but the manner in which they are coded can profoundly affect the usefulness of end-product dataset. The data must be coded in such a way as to provide us the maximum leverage possible over the research questions we have presented. Toward this end, I outline three primary criteria that should be coded in this scheme: tracking the movements and behaviors of protesters groups over time and locations, the behavior of police both inside and without the protest, and the presence of media personnel and social media reporting by protestors themselves.
The first criterion is coding the data in such a way as to capture the protest events both temporally and spatially. This means having systematic way of recording behaviors over time as the protest moved along its route from the northern start point to its end in ( ***) park. Toward this end, the coding scheme requires input that allows the observations to be geocoded. The coder will be instructed to use Google Maps Street View to locate approximately where along the protest route the video capture took place, allowing the coder to document side of street, and the nearest intersection.
From this information, these observations can be geotagged, and can allow researchers to control and track events along the routes (e.g. control for a particularly rowdy group of protesters as they move along the route).
I use Schweingruber and McPhail (1999) as a template for coding the protest behaviors, because this coding scheme will allow us to document a broad range of salient actions that will help us understand what events took place and how. This data can be used in conjunction with the information collected from social media accounts to gauge how quickly knowledge of the events that occurred in the protest spread, and then compared with what news media reported, so we can understand the discrepancies. Although the observations occur in fifteen minute increments, and a true sequential dataset is beyond our grasp, (Bakeman and Gottman 1996, 51), we can still use this behaviors template for tracking how disruptive or energetic sections of the protest impacted other section’s behavior (action diffusion), and see how various behaviors such as mass video recording impacted peoples’ willingness to act more or less violent.
Omitted from my coding scheme that appears in the S and M (1999) framework are numerous distinctions between behaviors that we are uninterested in recording. For example, handing out leaflets is irrelevant to our coding scheme because it does contribute to the study of our research questions. Also, onlookers, passersby and “others” are omitted due to their lack of relevance. However, our research questions necessitate generating new categories that will be more salient and allow us to measure the interactions we seek.
Specifically, what we hope to expand our understanding on is the second criterion, the behavior of the police and their role in the development of the protest. The S and M (1999) framework considers all law enforcement personnel the same, whereas I create a meaningful distinction between law enforcement beyond the boundaries of the acceptable protest lines, and those who are inside the protest itself, monitoring activities and providing eyes and information to the police command center. Police within the crowd could have different effects of the likelihood of violent agitation, as opposed to those outside of the protest, looking in. Conversely, these law enforcement agents could also be acting as agent provocateurs, stirring up trouble to provide the police the excuse they need to end the protest as defenders of the city’s law and order. I therefore hope to capture these effects to promote the study of the role of law enforcement in large protests, and the interaction between the protesters and police inside and without the crowd.
The presence of media personnel and the use and ubiquity of video recordings are the third criteria captured in this coding scheme, providing us with the data needed to look at the effect of media presence on behaviors as opposed to social media pressures. Is law enforcement less likely to use violence if there are 30 cell phone cameras recording his behavior, as opposed to a traditional news crew? Will protesters consider the same pressure in the same way? Also, the presence of traditional news media crews might impacts the news that makes its way into reports that air, but it is often difficult to know that for sure. This coding scheme allows us to get at both of those questions, opening up the potential for studying and understanding the effects of new and social media on social movements’ perceptions and effectiveness.
Instructions for Coding
4 observations per hour * 17 observers = 68 total observations per hour * 3 hours = 204 observations
The following are coding instructions for using the coding sheet. These instructions should allow any prospective coder the process for coding the videos collected.
Step one
Record the time of the video (to the nearest 15 min interval), and the number and position of the observer. Row=(1-8, or R for the roving team). These fields are found at the top of the coder sheet.
Step Two
Watch the video, and focus on the location. Use Google Street View of the route to determine the location of the footage. At the top of the coder sheet, report the street the video is taken on, and the N, S, E, or W side of the street. Also record the nearest intersecting cross street. This will be used to geocode the location.
Step Three
Watch the video, and this time, focus on the crowd. You will need to document the behavior of the people that you see. There are four types of people we are interested in observing: the protesters, the in-crowd police, the barricaded police (positioned behind barriers or maintaining a barricade line of some sort), and media personnel. We will ask you to note the behavior of each group, one at a time. In this step, we focus on the protesters.
1. Record your estimate of how many you can see in each video clip. While you do not need to count every person, try to be as precise as you can. Then record this number by writing the corresponding number (1-8) in the proper column at the bottom of the sheet. For example, if you can only see 100 people, you would mark the number 4 in the bottom, in the row marked “Total actors” and the column marked “Protest”
- Answer the density question by considering how difficult it looks to travel through the crowd. Circle the closest response from the options provided.
- Now go through the list on the left. There are 60 possible behaviors that we are interested in. Your task is to go through each behavior, and determine the percentage of actors that are participating in that activity. Once you determine the percentage of actors participating in a behavior, mark the appropriate number (e.g. 2 for between 10 and 20 percent) in the column that labeled “Protest.”
- If any activity of a violent nature (anything beyond vocal) occurs, mark at the bottom of the sheet who the targets of the violence were (e.g. if the protesters throw rocks at the police, you would record the percent of visible participants un the “Melee Weapon - Throwing” category, and then at the bottom, under targets in the “protest” column, write in ‘police.’).
Step Four
Record the behaviors from the police.
- Re-watch the video and count how many police are in the crowd, and, separately, how many are
outside the protest behind the barricades or other barriers. Record this number, like done with
the protest crowd, at the bottom in the appropriate column. - Now go through the list on the left. There are ## possible behaviors that we are interested in.
Your task is to go through each behavior, and determine the percentage of police that are participating in that activity. Once you determine the percentage of police participating in a behavior, mark the appropriate number (e.g. 2 for between 10 and 20 percent) in the column that labeled “Police (Crowd).” Repeat this step for “Police (Bar.)” [Barricaded Police]. Re-watch as many times as needed to get your counts as accurate as possible.
Step Five
Re-watch video once again, but this time focus on the behavior of the media people – journalists, writers, and social scientists. Record he number of media personnel visible at the bottom of the “Media” column. Then go through each behavior, and determine the percentage of media personnel that are participating in that activity. Once you determine the percentage of people participating in a behavior, mark the appropriate number (e.g. 2 for between 10 and 20 percent) in the column that labeled “Media.” Re-watch as many times as needed to get your counts as accurate as possible.
Step Six
Once you are done coding the behavior as you saw them, go to the section on the bottom right of the coding sheet. This is where you mark your overall confidence in the outcomes you are reporting on the sheet. This section should be used to validate or express concern with the quality of the video in allowing you to document the behaviors sought in this design.
Note: We have provided these instructions in the ‘watch-code, watch-code’ format intentionally. This is done to avoid your reliance on your memory for the information you are coding. We want to explicitly discourage you from trying to remember how many people you saw doing one thing or another, and encourage you to draw all of your coding from the videos as they are seen.
How to Code the Behaviors (What We Are Looking For)
There are five main categories of behaviors: vocal, movement, actions, emotional, and media. I will outline each one, and set forth how they are to be ideally interpreted. These explanations are guidelines to follow, and a certain amount of discretion will need to be used. However, using these categories, the videos should afford you the opportunity to document the behaviors in the protest as faithfully as possible.
Vocal
The first behavior category is the vocal. We include three sub categories, and eleven vocal behaviors. The three sub-categories, speech, non-speech and silence, are relatively straightforward. People speaking recognizable words to each other, on the phone, to police, or to any others are talking.Singing is defined as a single or collective melodic production. Chanting is considered a common chorus spoken by more than 5 people simultaneously. Taunting is broken down into general, which means as a derisive or angry speech directed at no one entity, and “at opp”, meaning at opposition, means that the taunting is directed at a party in particular whom the taunt-or is engaged. This can mean protesters taunting cops or vice versa. Shouting is speech with elevated volume.
The non-speech elements are sounds that do not involve word formation. Cheers, boos, and oohs and ahhs all fit into their own categories. The quiet elements are included to capture both attentiveprotestors to speeches around them (recorded as few participants shouting or talking, and many participants listening), and law enforcements who are paying attention to the commotion around them. Of course, actors may be quiet because they are simply not paying attention to anything, which, if apparent, is considered uninvolved.
Movement
The movement categories involve the level of movement by the actors. Actors can be stationary andstanding, sitting, which in a march is unlikely, or walking or running. Running is considered movement with the feet where only one foot is in contact with the ground at a time. Law enforcement has additional means by which they may be mobile, including horseback and vehicles. The scale of movement is unnecessary with mounted police (they are unlikely to be galloping through the crowd), but the direction of their movements is what interests use here. Cops on horseback may move with the crowd or against it, but vehicles, not likely to be inside the crowd, will move toward the crowd outside the barriers or away from them. The behavior of the vehicle or mounted police should be so coded.
Actions
Actions refer to behaviors involving the use of hands. In this coding scheme, there are several behaviors with hands, and then with objects which can occupy those hands. I will start with empty-hand behaviors. Hands can be doing nothing in general, fidgeting, or making gestures. Gestures have explicit meaning, whereas waving and restlessness or fidgeting convoy a different state from the actors. Restlessness or fidgeting includes empty moving hands and arms, but for no purpose, and as a result of a general state of agitation of the actor. Holding hands involves hold other human’s hands, holding another humanincludes grasping at some other part of another’s body besides their hands, to stay together, or to prevent harm or a strike. The context should be evident. Lastly, hands can strike, push or pull. These are considered violent actions. Targets of any violent behavior (striking, throwing, discharging, pushing, or pulling) should be included at the bottom of the list; these should be considered by the coder as primary targets, and as protesters, police inside the crowd, police outside the crowd, or possibly media. If there are more than one target, it should be noted on the back of the coding sheet.
Hands can grasp a variety of objects; we are interested in signage, melee weapons, ranged weapons, shields. Anything not included there may be included as other. A note on the back of the coding sheet should be made for this situation. Most of these objects can be held, waved, thrown or used for
striking. Signage is any sort of carried symbol or materials with characters on them representing an understood message. Melee weapons include rocks, sticks, clubs, batons, blackjacks, knives, swords, and any other held item whose primary purpose of possession is to inflict bodily harm, but only within an arm’ reach. If a melee weapon is thrown, it should be recorded as such. Ranged weapons are weapons that not require a throw to do harm beyond the immediate arm’s reach of an actor. These include firearms, grenade launchers or any sort, bows, crossbows, slings, slingshots, and spear throwers. One may strike another with a ranged weapon within arm’s reach, or an actor may discharge one.Shields are any device used for the purpose of protection from any attack of any nature. Riot shields, planks, trays, and boards can all be included here. These or course can be used offensively, for pushing or striking, and should be recorded as such.
Emotional
The emotional behaviors will naturally be more subjective, the intention is to capture an increased degree of mood or attitudes for explicit behaviors. Whether or not they are actually used in any data analyses or not will be seen later, but the hope and the present reason for their inclusion is to capture general changes in emotional tone of relevant actors in the protest. Since we cannot capture a continuous feed, some behaviors might seem to come from nowhere. If the emotional feelings of the crowd can acts as predictors for future aggression, then emotions will provide us important leverage over the onset and diffusion of feelings and behaviors through the protest.
Facial expressions will follow a template that will be provided at a later time, but are intuitively straightforward. Pleased is a content or non-dissatisfied look, sad is indicated by sorrow or lament, and crying is coded with any visible evidence of such. Aggression is kept a general category, and any angry or determined face will be considered aggressive. Indifference is coded as the absence of any of the other facial expressions. Tone will be the most subjective component of the behaviors; consequently, it is kept as simple as possible, as they will capture the growing or waning level of aggression that is expressed through speech and non-speech communication. Unapparent is the baseline category here, and if there is doubt regarding the general tone, then this category should be used. Pleased should only be coded when the communication sounds distinctively jovial, fun, or excited. If the communications sound markedly aggressive, angry or hateful, they should be thus coded.
Media
The last category is media behaviors, including phone or camera. The phone category includes checking, typing and video recording. Checking amounts to more than a three second glance, reading should be apparent. Typing of any kind should be obvious and included where seen. We will assume that holding the phone in one hand in a motion that aims the built-in camera on their phone as a particular direction indicates that an actor is using their phone to record video. We will also assume that legitimateinterviews will involve more than a camera phone; an actual video recorder will distinguish these. Other video recording using a purpose built camera should be coded as camera video recording.
Conclusion
I have outlined about a coding scheme which can be used to take video recordings from an important protest and code them as observations that can be used to expand our understanding of protest behavior. This scheme effectively captures the behaviors to inform both of our research questions, and advance our understanding of social protests. This work is the first step towards building the data needed to testing the relationships and interactions between protesters and law enforcement personnel. Toward this end, I hope to see the use or small drones by researchers to monitor violent clashes, which should expand our ability in the future to model protesters and police behavior by allowing researcher to observe clashes directly, while protecting them from immediate danger. The biggest area to benefit from this project will be the public perceptions of social movements. As the availability of social media and its use expands, we could one day hope to gather a data from a thousand phones documenting events from all angles, helping us understand reporting selection in protests, and the impacts this has on public perception.
References
- Bakeman, Roger, and John M. Gottman. 1997. Observing interaction: An introduction to sequential analysis. Cambridge University Press.
- Chenoweth, Erica & Maria J. Stephan. 2011. Why Civil Resistance Works: The Strategic Logic of Nonviolent Conflict, New York: Columbia University Press.
- McLeod, Douglas M., and Benjamin H. Detenber. 1999. "Framing effects of television news coverage of social protest." Journal of Communication 49, no. 3 (1999): 3-23.
- Schweingruber, David, and Clark McPhail. 1999. "A method for systematically observing and recording collective action." Sociological methods & research 27, no. 4 (1999): 451-498.
A Coding Sheet?!
A more detailed consideration
Protester/Police Interaction Video Data Project: Coding Rules
Casey Delehanty Eryn Jones Scott Meachum Ryan Welch
Political Violence Class Project Political Science Florida State University
02 May 2013 Version 1.00
Contents
1 Introduction
1.1 ImportanceoftheDataCollection
2 General Guidelines for Coders
2.1 Assignments
2.2 CodingProcesses
2.3 ExcelCodingSheet
2.4 CoderNotes
3 Data
3.1 ActionType
3.2 ActionSeverity
3.3 Participant(s)Type
3.4 NumberofParticipants1
3.5 NumberofParticipants2
3.6 GeneralMoodofProtesters
3.7 GeneralMoodofPolice
3.8 TimeStamp
3.9 FileName
4 Appendix
1 Introduction
The Protester/Police Interaction Video Data Project aims to provide scholars with quanti- tative data to test hypotheses about the actions of protestors, state agents, their interactions and what consequences follow from them. This codebook describes the processes associated with coding the video data. The Protester/Police Data contains information from videos shot on site in real time. The goal of the data collected here is to obtain information on different types and severities of actions taken by both protesters and state agents i.e. police. Both actions directed at each other and directed otherwise are of interest in order to answer questions regarding escalation at events that begin as non-violent in nature.
The Unit of Observation in the study is the contentious action. This is defined as any action taken by an involved actor(s) that is more contentious than the defined baseline. The baseline is the modal non-contentious activity being taken by the actors during the protest. While this particular study focuses on the World Trade Organization (WTO) protest that occurred in May 2012, the principles used here can be generalized to many other protests. The baseline for this project are as follows. For the protesters it is a slow-paced march with either no sign or a non-threatening sign. The baseline for the police is standing and observing without any weapons drawn and with no riot gear.1
1.1 Importance of the Data Collection
Citizens use protest to enact public policy changes in countries (Opp 1994). It is especially important for citizens with the (actual or perceived) inability to directly influence policy (Machado et al. 2011). In fact, some claim dissent is crucial for legitimizing sovereignty and the state (Habermas 1985). [T]he protection of non-violent dissent within the bounds of the law is supposed to guarantee that the law reflects the will of the people, which is a source of legitimacy (Kohn 2013, 102). Given the state’s monopoly of the use of force within its territory (Weber 1910), the change from non-violent to violent deligitimizes protest in the eyes of the state. This allows the state the claim needed for the legitimate use of force; which is an outcome that has been observed. (Moore 1998, 2000). Non-violent protest turned violent often leads to more violence. Studying the factors that lead to the violent transition may help researchers and policy-makers understand how to reduce overall levels of violence within states. Given that protest is used to influence policy, understanding the factors that influence initially non-violent dissent to turn violent may also be important for activists. According to Chenoweth and Stephan (2011), non-violent protest is more successful than its violent counterpart.
The study of non-violent dissent turned violent is important to scholars, policy-makers, and activists. Given its importance as a non-formal political strategy, it is not surprising that scholars have been studying dissent from Hobbes until the present (e.g. Hobbes 1985; Klandermans 1983, 1984; Machiavelli 1985; Murdie and Bhasin 2011; Opp 1982, 1988, 1994, 1998). Unfortunately systematic research of this process is plagued by data complications. One strategy is to use survey data (e.g. Opp 1994; Machado et al. 2011). While this approach allows the researcher to test micro-level mechanisms, it leaves out an important part of the theoretical process - the realization of the violent protest. In other words, researchers have been content to test their theories using survey participants’ word that that they protested rather than the actual action. This is troubling, as is explicitly admitted by Opp (1994, 109): “Asking questions about past attitudes, behaviors, or circumstances may yield particular problems of validityInterviewees may even lie to interviewers because they do not want to reveal behavior that is now regarded as undesirable.” That is, when a protest has succeeded, people may answer that they participated (whether due to mistaken memory or outright dishonesty). It is also the case, that if a protest has failed, the regime is still in place or more willing to crack down on dissenters, and people will be reluctant to admit to demonstrating against that regime.
Another strategy is to use observational data from media sources. This is problematic because the majority of protests go unreported, and those that do are the larger or more destructive episodes (Schweingruber and McPhail 1999). The selection bias of the reporting may cause incorrect inferences with many of the statistical models used in systematic studies.
This data collection project is designed with these obstacles in mind. Observers were trained to walk with the protestors, spaced at 25 foot intervals, and every 5 minutes would turn on the video camera and turn around 360 degrees. This provides a roughly 15 second video clip for coders. By observing and coding video of protest events, data regarding protester and police actions that contribute to variation in the (de)escalation of protests can be used in statistical models to answer the questions such as: Why do some protests escalate to violence, while others do not? What protester actions cause changes in the violence level of the protest? What police actions cause changes in the violence level of the protest?
2 General Guidelines for Coders
Each clip that you watch is approximately 15 seconds in length. In order to capture all of the information you will need to watch each clip multiple times. Any questions you have should be emailed to the project administrator.
2.1 Assignments
You will receive instructions in how to access the video clip library from the project admin- istrator. When you log in and begin to work on a clip, check the appropriate box so that other coders know which clips have/have not been viewed. You can view the clip from the library. You do not have to download each clip that you code.2
2.2 Coding Processes
There are multiple pieces of information that you are looking for in each clip. You are most interested in the interactions between protestors and police, but you will also record such things as the general mood of the crowd and the police, specific actions taken by either side, the number of protestors in the crowd, etc. Specific information on these variables can be found in Section 3 and in the attached appendix. All information you collect from a video clip will be recorded in an Excel spreadsheet. It is expected that you will watch each video clip multiple times to capture as much information as possible.
2.3 Excel Coding Sheet
You will be recording data on an Excel Coding sheet. Before you watch any clip you must fill in the following information in the appropriate fields: your name, the date on which you coded the clip, the time frame that is being captured in the video, the clip num- ber, and afterwards the duration of coding this clip. These fields are found in Row 1 of the Excel template. Immediately save your file in the following format “LastNameof- Coder ClipNO DDMonYYYY” (i.e. Meachum 123 01May2013). Once you have done this, you can begin to code your clip. You will turn in your code sheets on a weekly basis to the project administrator.
2.4 Coder Notes
Along with your Coding Sheet, you should also compile a Word document of Coder Notes that contains any information that you were unable to fit into the Coding Sheet or with any questions/observations/problems. An example of Coder Notes will be distributed at the beginning of the project. The format should be followed in your own work, and the file should be saved according to the following format: LastNameofCoder ClipNoNotes DDMonYYYY (i.e., Meachum 123Notes 01May2013).
3 Data
This section includes a description of each of the variables that we code at the Contentious Action unit of observation. The variables are listed in the same order as the columns on the coding sheet. Details on the coding scheme of each variable follow a definition of the concept.
3.1 Action Type
Definition: This variable records the type of action that is being taken by the participant(s).3
Coding Scheme: Assign values according to the following nominal scale.
• 1 = Vocalization: Any sounds made with the mouth
• 2 = Locomotion: Any movements made with the body
• 3 = Manual Manipulation: Any change of one’s surroundings using one’s hands
3.2 Action Severity
Definition: This is how contentious the action taken is. In other words, this records how violent or prone to violence the action is.
Coding Scheme: Assign values according to the following ordinal scale, increasing in severity as values get larger. This variable is mutually exclusive.
• 1 = Just above baseline activity
• 2 = Actions that could be perceived as threatening, but are not overt.• 3 = Overtly threatening violence
• 4 = Violence short of death
• 5 = Violence resulting in death
3.3 Participant(s) Type
Definition: This records what type of participant is involved in the activity recordedCoding Scheme: Assign values according to the following nominal scale. This variable is mutually exclusive.
• 1 = protester
• 2 = other civilian (i.e. bystander)
• 3 = police
• 4 = other member of the state apparatus (i.e. military)• 5 = other
• 6 = unknown
3.4 Number of Participants 1
Definition: This is a rough guess at how many people are participating in the activity recorded.
Coding Scheme: This is an integer value ¿= 1
3.5 Number of Participants 2
Definition: This is a rough guess at how many people are participating in the activity recorded.
Coding Scheme: Assign values according to the following ordinal scale. This variable is mutually exclusive.
• 1=1–10
• 2 = 11–50
• 3 = 51–200• 4 = 201–500• 5=>500
3.6 General Mood of Protesters
Definition: This is an estimation of how contentious the group of protesters as a whole are when the activity is recorded.
Coding Scheme: Assign values according to the following ordinal scale. This variable is mutually exclusive.
• 0 = Baseline
• 1 = Just above baseline; slightly agitated
• 2 = Actions that could be perceived as threatening, but are not overt• 3 = Participating in overtly threatening action; generally destructive• 4 = Participating in violence with police
3.7 General Mood of Police
Definition: This is an estimation of how contentious the group of police as a whole are when the activity is recorded.
Coding Scheme: Assign values according to the following ordinal scale. This variable is mutually exclusive.
• 0 = Baseline
• 1 = Just above baseline; slightly agitated
• 2 = Actions that could be perceived as threatening, but are not overt• 3 = Participating in overtly threatening action; generally destructive• 4 = Participating in violence with protesters
3.8 Time Stamp
Definition: This variable records the time of the recorded activity on the video file.Coding Scheme: Assign a time value in minutes and seconds e.g. MM:SS.
3.9 File Name
Definition: This variable records the name of the file that is being coded.Coding Scheme: Assign a string value that is the name of the file.
4 Appendix
The following appendix further elucidates some of the logic behind the coding scheme pre- sented above. We present a number of possible scenarios and how to code them in order to assist coders in making tough decisions. We acknowledge that scenarios will occur which are not currently anticipated by the rules delineated in this code book. As more information becomes known during the coding process, better cut points can be established.
Activity Type
As noted in the Codebook, this variable is mutually exclusive, meaning an actor can be engaging in multiple contentious actions simultaneously. The main goal as a coder is to capture each action and record it separately. We make no assumptions between types of activity as to which is more or less contentious as varying severity levels capture that aspect of the action. The following hypothetical scenarios will help illustrate
Scenario 1: In a video clip, among the group of protestors, we observe one protestor with a sign that reads “Fuck the Police” and protestor with a sign that reads “Down with the WTO.” The first protestor is marching silently while the second is flipping off police officers that stand along the parade route. From what we can see, the police officers are all standing still and none act/react.
Coding: Purely from the Action Type variable, we would code one entry for the first protestor as a 3, which corresponds to manual manipulation. A separate entry is created for the second protestor and is coded as 2, which corresponds to locomotion and captures his gesture toward the police. As the officers don’t take action over baseline, no entries are noted/coded for them.4
Scenario 2: In a video clip, we see observe a protestor pick up a bottle from the street and throw it toward the police while at the same time yelling “Fuck the Police”. At the same time, we are able to hear an officer yelling at the crowd that they will tase the next protestor that steps out of line.
Coding: There are three entries to code in this scenario. First, there is the protestor who throws the bottle. This is coded as 3, a manual manipulation. The second entry applies to the language used by the protestor which is potentially threatening in nature and, again, against the tone of the march. Finally, a third contentious action is coded for the police officer for his comment about tasing the next protestor that strays from the parade route.
Scenario 3: In a video clip, we observe a large group of protestors engaging in contentious actions with the police. A number of protestors are chanting against the police. A smaller sub-group of protestors are gesturing with their middle fingers toward police. Another group begins to throw objects at the police line. One police officer tells the protestors to step back in line and move with the march. Down the line, two other officers move in toward the group throwing objects and begin to hit them with batons.
Coding: For this scenario five observations will be coded: three protestor actions and two police actions. The first is the group of protestors chanting, which would be coded as a 1 for vocalization. The second observation is the group of gesturing protestors. This falls in the category of locomotion and is coded as 2. The final protestor observation is for the group throwing objects at the police line. This is manual manipulation and is coded as 3. The police officer who tells the protestors to get back in line is coded as 1.5 The police officers who use their batons against the protestors are coded as 3 for manual manipulation.
Scenario 4 (video provided): In the sample video clip we have provided, three protesters are engaging in the same action: gesturing towards police, holding up two fingers in the traditional “peace” sign. Two of the three appear to be together, while the third gesturing protester is alone. Otherwise, the rest of the crowd (police and protesters) appears to be engaging in baseline activity.
Coding: There are a total of two entries to code from this video clip. First, two women in the clip appearing from 0:08-0:10 (one wearing a floppy hat, the other with a “No to NATO and G8 Warmakers” sign) are showing the “peace” hand gesture to a line of officers. Since the two women appear to engaging in the same activity together, this is coded as one event with two participants. While the woman in the light blue tank top at 0:15 is engaging in the same action (giving officers the peace sign), she does not appear to be engaging in this activity with anyone else. As such, this video clip contains two contentious actions, one involving two protesters and one involving one protester. Both of these actions would be coded as a 2, since they are making bodily movements that do not actively change their surroundings.
Action Severity
The coding of this particular variable is subjective in many ways. With the exception of category 1, coders must decide for themselves how severe an action is, keeping in mind that the same action may be seen differently in separate contexts. Take for instance the first protestor from Scenario 1 that had a sign that read “Fuck the Police” but was otherwise marching along and not doing anything versus the second protestor who verbally stated “Fuck the Police” in the context of other activity. The first protestor should likely be coded as 1, maybe a 2. The second protestor’s shout is at least a 3. We should note here that his action of throwing the bottle, which has been recorded separately, is clearly a 4. It is important as a coder to view each contentious action, the unit of analysis, as its own action and not to conflate multiple actions.
For example, in Scenario 3, both the chanting and gesturing on behalf of the protestors may be perceived as threatening actions, but because they are not overt actions, both would be coded as a 2 for action severity. Throwing objects can be categorized as overly threatening violence and would thus be coded as a 3. The police officer who tells protestors to get back in line is not necessarily a threatening action, however it is above baseline activity and should therefore be coded as a 1. The police officers who engage in violence using their batons would be coded as a 4 for violence short of death.
Referring back to Scenario 4, the protesters giving peace signs to police officers would be coded as a 1, since their actions could hardly be taken as threatening, and just barely exceed baseline behavior.
Participant(s) Type
The coding of this variable should be very straightforward. For each contentious action observed, simply record the identity of the individual involved. As long as the identity of the actor can be determined, you should code the actor in one of the first five categories. Obviously, the fifth category is a catchall and the sixth could come into play if an actor shouts off camera or some other similar scenario exists.
Number of Participants
In some circumstances, it might be difficult to count the number of participants in a given action. Going back to the previous scenarios discussed above, it is clear that for each entry, one participant was engaged in each contentious activity. Thus for 3.4, the value of 1 is recorded and for 3.5, the value of 1 is recorded as well. The utility of having a specific and a broader measure comes when multiple participants, perhaps even a mob, are engage in a behavior. While we prefer the specific measure, the categorical measure will likely be useful in some circumstances.
Remember to take care in assigning activities to groups. As in Scenario 4, multiple individuals can engage in a similar action without necessarily being part of the same group.
General Mood of Protesters/Police
These two variables are mirror images of one another and attempt to give future researchers some idea of how the temporal dynamics of a protest might change over time. For instance, after the principal investigator establishes the baseline behavior for each of the respective parties (protestors, state agents), it is useful to know whether the tone of each group changes across contentious actions. That is, this variable is an average at the time of each contentious action (i.e. a snapshot). In the event of escalation, this could be very useful information.
Casey Delehanty Eryn Jones Scott Meachum Ryan Welch
Political Violence Class Project Political Science Florida State University
02 May 2013 Version 1.00
Contents
1 Introduction
1.1 ImportanceoftheDataCollection
2 General Guidelines for Coders
2.1 Assignments
2.2 CodingProcesses
2.3 ExcelCodingSheet
2.4 CoderNotes
3 Data
3.1 ActionType
3.2 ActionSeverity
3.3 Participant(s)Type
3.4 NumberofParticipants1
3.5 NumberofParticipants2
3.6 GeneralMoodofProtesters
3.7 GeneralMoodofPolice
3.8 TimeStamp
3.9 FileName
4 Appendix
1 Introduction
The Protester/Police Interaction Video Data Project aims to provide scholars with quanti- tative data to test hypotheses about the actions of protestors, state agents, their interactions and what consequences follow from them. This codebook describes the processes associated with coding the video data. The Protester/Police Data contains information from videos shot on site in real time. The goal of the data collected here is to obtain information on different types and severities of actions taken by both protesters and state agents i.e. police. Both actions directed at each other and directed otherwise are of interest in order to answer questions regarding escalation at events that begin as non-violent in nature.
The Unit of Observation in the study is the contentious action. This is defined as any action taken by an involved actor(s) that is more contentious than the defined baseline. The baseline is the modal non-contentious activity being taken by the actors during the protest. While this particular study focuses on the World Trade Organization (WTO) protest that occurred in May 2012, the principles used here can be generalized to many other protests. The baseline for this project are as follows. For the protesters it is a slow-paced march with either no sign or a non-threatening sign. The baseline for the police is standing and observing without any weapons drawn and with no riot gear.1
1.1 Importance of the Data Collection
Citizens use protest to enact public policy changes in countries (Opp 1994). It is especially important for citizens with the (actual or perceived) inability to directly influence policy (Machado et al. 2011). In fact, some claim dissent is crucial for legitimizing sovereignty and the state (Habermas 1985). [T]he protection of non-violent dissent within the bounds of the law is supposed to guarantee that the law reflects the will of the people, which is a source of legitimacy (Kohn 2013, 102). Given the state’s monopoly of the use of force within its territory (Weber 1910), the change from non-violent to violent deligitimizes protest in the eyes of the state. This allows the state the claim needed for the legitimate use of force; which is an outcome that has been observed. (Moore 1998, 2000). Non-violent protest turned violent often leads to more violence. Studying the factors that lead to the violent transition may help researchers and policy-makers understand how to reduce overall levels of violence within states. Given that protest is used to influence policy, understanding the factors that influence initially non-violent dissent to turn violent may also be important for activists. According to Chenoweth and Stephan (2011), non-violent protest is more successful than its violent counterpart.
The study of non-violent dissent turned violent is important to scholars, policy-makers, and activists. Given its importance as a non-formal political strategy, it is not surprising that scholars have been studying dissent from Hobbes until the present (e.g. Hobbes 1985; Klandermans 1983, 1984; Machiavelli 1985; Murdie and Bhasin 2011; Opp 1982, 1988, 1994, 1998). Unfortunately systematic research of this process is plagued by data complications. One strategy is to use survey data (e.g. Opp 1994; Machado et al. 2011). While this approach allows the researcher to test micro-level mechanisms, it leaves out an important part of the theoretical process - the realization of the violent protest. In other words, researchers have been content to test their theories using survey participants’ word that that they protested rather than the actual action. This is troubling, as is explicitly admitted by Opp (1994, 109): “Asking questions about past attitudes, behaviors, or circumstances may yield particular problems of validityInterviewees may even lie to interviewers because they do not want to reveal behavior that is now regarded as undesirable.” That is, when a protest has succeeded, people may answer that they participated (whether due to mistaken memory or outright dishonesty). It is also the case, that if a protest has failed, the regime is still in place or more willing to crack down on dissenters, and people will be reluctant to admit to demonstrating against that regime.
Another strategy is to use observational data from media sources. This is problematic because the majority of protests go unreported, and those that do are the larger or more destructive episodes (Schweingruber and McPhail 1999). The selection bias of the reporting may cause incorrect inferences with many of the statistical models used in systematic studies.
This data collection project is designed with these obstacles in mind. Observers were trained to walk with the protestors, spaced at 25 foot intervals, and every 5 minutes would turn on the video camera and turn around 360 degrees. This provides a roughly 15 second video clip for coders. By observing and coding video of protest events, data regarding protester and police actions that contribute to variation in the (de)escalation of protests can be used in statistical models to answer the questions such as: Why do some protests escalate to violence, while others do not? What protester actions cause changes in the violence level of the protest? What police actions cause changes in the violence level of the protest?
2 General Guidelines for Coders
Each clip that you watch is approximately 15 seconds in length. In order to capture all of the information you will need to watch each clip multiple times. Any questions you have should be emailed to the project administrator.
2.1 Assignments
You will receive instructions in how to access the video clip library from the project admin- istrator. When you log in and begin to work on a clip, check the appropriate box so that other coders know which clips have/have not been viewed. You can view the clip from the library. You do not have to download each clip that you code.2
2.2 Coding Processes
There are multiple pieces of information that you are looking for in each clip. You are most interested in the interactions between protestors and police, but you will also record such things as the general mood of the crowd and the police, specific actions taken by either side, the number of protestors in the crowd, etc. Specific information on these variables can be found in Section 3 and in the attached appendix. All information you collect from a video clip will be recorded in an Excel spreadsheet. It is expected that you will watch each video clip multiple times to capture as much information as possible.
2.3 Excel Coding Sheet
You will be recording data on an Excel Coding sheet. Before you watch any clip you must fill in the following information in the appropriate fields: your name, the date on which you coded the clip, the time frame that is being captured in the video, the clip num- ber, and afterwards the duration of coding this clip. These fields are found in Row 1 of the Excel template. Immediately save your file in the following format “LastNameof- Coder ClipNO DDMonYYYY” (i.e. Meachum 123 01May2013). Once you have done this, you can begin to code your clip. You will turn in your code sheets on a weekly basis to the project administrator.
2.4 Coder Notes
Along with your Coding Sheet, you should also compile a Word document of Coder Notes that contains any information that you were unable to fit into the Coding Sheet or with any questions/observations/problems. An example of Coder Notes will be distributed at the beginning of the project. The format should be followed in your own work, and the file should be saved according to the following format: LastNameofCoder ClipNoNotes DDMonYYYY (i.e., Meachum 123Notes 01May2013).
3 Data
This section includes a description of each of the variables that we code at the Contentious Action unit of observation. The variables are listed in the same order as the columns on the coding sheet. Details on the coding scheme of each variable follow a definition of the concept.
3.1 Action Type
Definition: This variable records the type of action that is being taken by the participant(s).3
Coding Scheme: Assign values according to the following nominal scale.
• 1 = Vocalization: Any sounds made with the mouth
• 2 = Locomotion: Any movements made with the body
• 3 = Manual Manipulation: Any change of one’s surroundings using one’s hands
3.2 Action Severity
Definition: This is how contentious the action taken is. In other words, this records how violent or prone to violence the action is.
Coding Scheme: Assign values according to the following ordinal scale, increasing in severity as values get larger. This variable is mutually exclusive.
• 1 = Just above baseline activity
• 2 = Actions that could be perceived as threatening, but are not overt.• 3 = Overtly threatening violence
• 4 = Violence short of death
• 5 = Violence resulting in death
3.3 Participant(s) Type
Definition: This records what type of participant is involved in the activity recordedCoding Scheme: Assign values according to the following nominal scale. This variable is mutually exclusive.
• 1 = protester
• 2 = other civilian (i.e. bystander)
• 3 = police
• 4 = other member of the state apparatus (i.e. military)• 5 = other
• 6 = unknown
3.4 Number of Participants 1
Definition: This is a rough guess at how many people are participating in the activity recorded.
Coding Scheme: This is an integer value ¿= 1
3.5 Number of Participants 2
Definition: This is a rough guess at how many people are participating in the activity recorded.
Coding Scheme: Assign values according to the following ordinal scale. This variable is mutually exclusive.
• 1=1–10
• 2 = 11–50
• 3 = 51–200• 4 = 201–500• 5=>500
3.6 General Mood of Protesters
Definition: This is an estimation of how contentious the group of protesters as a whole are when the activity is recorded.
Coding Scheme: Assign values according to the following ordinal scale. This variable is mutually exclusive.
• 0 = Baseline
• 1 = Just above baseline; slightly agitated
• 2 = Actions that could be perceived as threatening, but are not overt• 3 = Participating in overtly threatening action; generally destructive• 4 = Participating in violence with police
3.7 General Mood of Police
Definition: This is an estimation of how contentious the group of police as a whole are when the activity is recorded.
Coding Scheme: Assign values according to the following ordinal scale. This variable is mutually exclusive.
• 0 = Baseline
• 1 = Just above baseline; slightly agitated
• 2 = Actions that could be perceived as threatening, but are not overt• 3 = Participating in overtly threatening action; generally destructive• 4 = Participating in violence with protesters
3.8 Time Stamp
Definition: This variable records the time of the recorded activity on the video file.Coding Scheme: Assign a time value in minutes and seconds e.g. MM:SS.
3.9 File Name
Definition: This variable records the name of the file that is being coded.Coding Scheme: Assign a string value that is the name of the file.
4 Appendix
The following appendix further elucidates some of the logic behind the coding scheme pre- sented above. We present a number of possible scenarios and how to code them in order to assist coders in making tough decisions. We acknowledge that scenarios will occur which are not currently anticipated by the rules delineated in this code book. As more information becomes known during the coding process, better cut points can be established.
Activity Type
As noted in the Codebook, this variable is mutually exclusive, meaning an actor can be engaging in multiple contentious actions simultaneously. The main goal as a coder is to capture each action and record it separately. We make no assumptions between types of activity as to which is more or less contentious as varying severity levels capture that aspect of the action. The following hypothetical scenarios will help illustrate
Scenario 1: In a video clip, among the group of protestors, we observe one protestor with a sign that reads “Fuck the Police” and protestor with a sign that reads “Down with the WTO.” The first protestor is marching silently while the second is flipping off police officers that stand along the parade route. From what we can see, the police officers are all standing still and none act/react.
Coding: Purely from the Action Type variable, we would code one entry for the first protestor as a 3, which corresponds to manual manipulation. A separate entry is created for the second protestor and is coded as 2, which corresponds to locomotion and captures his gesture toward the police. As the officers don’t take action over baseline, no entries are noted/coded for them.4
Scenario 2: In a video clip, we see observe a protestor pick up a bottle from the street and throw it toward the police while at the same time yelling “Fuck the Police”. At the same time, we are able to hear an officer yelling at the crowd that they will tase the next protestor that steps out of line.
Coding: There are three entries to code in this scenario. First, there is the protestor who throws the bottle. This is coded as 3, a manual manipulation. The second entry applies to the language used by the protestor which is potentially threatening in nature and, again, against the tone of the march. Finally, a third contentious action is coded for the police officer for his comment about tasing the next protestor that strays from the parade route.
Scenario 3: In a video clip, we observe a large group of protestors engaging in contentious actions with the police. A number of protestors are chanting against the police. A smaller sub-group of protestors are gesturing with their middle fingers toward police. Another group begins to throw objects at the police line. One police officer tells the protestors to step back in line and move with the march. Down the line, two other officers move in toward the group throwing objects and begin to hit them with batons.
Coding: For this scenario five observations will be coded: three protestor actions and two police actions. The first is the group of protestors chanting, which would be coded as a 1 for vocalization. The second observation is the group of gesturing protestors. This falls in the category of locomotion and is coded as 2. The final protestor observation is for the group throwing objects at the police line. This is manual manipulation and is coded as 3. The police officer who tells the protestors to get back in line is coded as 1.5 The police officers who use their batons against the protestors are coded as 3 for manual manipulation.
Scenario 4 (video provided): In the sample video clip we have provided, three protesters are engaging in the same action: gesturing towards police, holding up two fingers in the traditional “peace” sign. Two of the three appear to be together, while the third gesturing protester is alone. Otherwise, the rest of the crowd (police and protesters) appears to be engaging in baseline activity.
Coding: There are a total of two entries to code from this video clip. First, two women in the clip appearing from 0:08-0:10 (one wearing a floppy hat, the other with a “No to NATO and G8 Warmakers” sign) are showing the “peace” hand gesture to a line of officers. Since the two women appear to engaging in the same activity together, this is coded as one event with two participants. While the woman in the light blue tank top at 0:15 is engaging in the same action (giving officers the peace sign), she does not appear to be engaging in this activity with anyone else. As such, this video clip contains two contentious actions, one involving two protesters and one involving one protester. Both of these actions would be coded as a 2, since they are making bodily movements that do not actively change their surroundings.
Action Severity
The coding of this particular variable is subjective in many ways. With the exception of category 1, coders must decide for themselves how severe an action is, keeping in mind that the same action may be seen differently in separate contexts. Take for instance the first protestor from Scenario 1 that had a sign that read “Fuck the Police” but was otherwise marching along and not doing anything versus the second protestor who verbally stated “Fuck the Police” in the context of other activity. The first protestor should likely be coded as 1, maybe a 2. The second protestor’s shout is at least a 3. We should note here that his action of throwing the bottle, which has been recorded separately, is clearly a 4. It is important as a coder to view each contentious action, the unit of analysis, as its own action and not to conflate multiple actions.
For example, in Scenario 3, both the chanting and gesturing on behalf of the protestors may be perceived as threatening actions, but because they are not overt actions, both would be coded as a 2 for action severity. Throwing objects can be categorized as overly threatening violence and would thus be coded as a 3. The police officer who tells protestors to get back in line is not necessarily a threatening action, however it is above baseline activity and should therefore be coded as a 1. The police officers who engage in violence using their batons would be coded as a 4 for violence short of death.
Referring back to Scenario 4, the protesters giving peace signs to police officers would be coded as a 1, since their actions could hardly be taken as threatening, and just barely exceed baseline behavior.
Participant(s) Type
The coding of this variable should be very straightforward. For each contentious action observed, simply record the identity of the individual involved. As long as the identity of the actor can be determined, you should code the actor in one of the first five categories. Obviously, the fifth category is a catchall and the sixth could come into play if an actor shouts off camera or some other similar scenario exists.
Number of Participants
In some circumstances, it might be difficult to count the number of participants in a given action. Going back to the previous scenarios discussed above, it is clear that for each entry, one participant was engaged in each contentious activity. Thus for 3.4, the value of 1 is recorded and for 3.5, the value of 1 is recorded as well. The utility of having a specific and a broader measure comes when multiple participants, perhaps even a mob, are engage in a behavior. While we prefer the specific measure, the categorical measure will likely be useful in some circumstances.
Remember to take care in assigning activities to groups. As in Scenario 4, multiple individuals can engage in a similar action without necessarily being part of the same group.
General Mood of Protesters/Police
These two variables are mirror images of one another and attempt to give future researchers some idea of how the temporal dynamics of a protest might change over time. For instance, after the principal investigator establishes the baseline behavior for each of the respective parties (protestors, state agents), it is useful to know whether the tone of each group changes across contentious actions. That is, this variable is an average at the time of each contentious action (i.e. a snapshot). In the event of escalation, this could be very useful information.
Current Status
We began coding but stopped because of Prof. Moore's passing. We are back on this now. Additionally, as we upload videos to Youtube we will provide links.