Packaging protest: media coverage of Indigenous people's collective action.

Wilkes, Rima ; Corrigall-Brown, Catherine ; Myers, Daniel J. 等

FOR SOME TIME NOW, INDIGENOUS PEOPLES in Canada have used direct action as an important means of achieving social and political justice when more conventional routes have been blocked. Indeed, since the early 1980s there have been several hundred instances of collective action, ranging from marches and demonstrations to protracted standoffs with authorities (Blomley 1996; Ramos 2006, 2008a, 2008b). The reason why most Canadians, be they Indigenous or not, know about these events is that they have received significant media attention. Yet, while it is clear that the media pay disparate attention to these events, no study has considered the causes of the differing quantity and quality of coverage they received. By featuring some events and issues more than others, the media works to characterize groups and events in a particular way. This, in turn, can impact public opinion and government policy on a range of issues related to Indigenous peoples.

We have two objectives in this paper. First, we describe the newspaper coverage of Indigenous people's collective actions from 1985 through 1995 in a new way. Most studies of media coverage of protest by Indigenous peoples focus on the framing of such events, typically the framing of a single event, such as the 1990 Oka crisis or the 1995 Gustafsen Lake standoff (Grenier 1994; Lambertus 2004; Miller 2005; Skea 1993). However, no study has attempted to compare coverage across events. How much coverage do different events receive? What types of events get front-page coverage? What types of events are accompanied by pictures? In considering these and other questions, we not only contribute to the specific understanding of media coverage of these specific events, but also to news coverage of collective action more generally. Although a large number of studies consider the quality of coverage, only a few consider any type of quantitative measure of coverage across events (except see Amenta et al. 2009; Vliegenthart, Omegema, and Klandermans 2005). These quantitative measures can allow for a systematic comparison across a much wider number of events and articles than was previously possible.

Our second aim is to identify the factors associated with the quality and quantity of media coverage. The media studies and social movements literature has shown that the coverage received by collective action events is not random (Boyle et al. 2005; Myers and Caniglia 2004; Oliver and Maney 2000). Past work has found that events that fit better with the organizational and structural demands of creating the news are more likely to be covered. How and to what extent are the demands of creating the news associated with coverage of collective action (1) by Indigenous peoples? Is it similar to or different from findings about other forms of contentious action? What are the characteristics of the events that receive more coverage than others? Is coverage determined by characteristics of the events themselves and which ones are deemed more newsworthy? Although these questions have been asked about a host of other instances of collective action ranging from the small events in Madison, WI (Oliver and Myers 1999) to massive demonstrations in Washington, DC (Smith et al. 2001), they have yet to be asked about media coverage of collective action by Indigenous peoples.

We find that although there were several hundred events, only four of these garnered almost half of all prominent forms of packaging. Given that packaging affects the visibility of specific articles, it has enormous potential to affect public opinion about the kinds of events that are representative of collective action by the members of Indigenous communities. This unevenness matters further in that that the content within these articles about these events is already shown to be extremely biased (Grenier 1994; Miller 2005; Skea 1993). Thus, the research in this paper provides new evidence on the extent and degree of biases in newspaper coverage of collective action generally and by Indigenous peoples specifically. Overall, the results clearly show that mobilizing groups and social movements could come to be defined and understood by a very limited number of nonrepresentative events.

DIRECT ACTION BY INDIGENOUS PEOPLES

Despite comprising just slightly < 4 percent of the population, Indigenous peoples (2) have nevertheless engaged in widespread mobilization in Canada. And, as our review of these actions and the scholarly literature about them will show, much of this mobilization has generated significant media interest. However, before discussing these events and the news coverage they have received, we provide a brief overview of the context within which they have taken place.

Canadian Indigenous peoples face numerous inequities in comparison with the general population. Rates of unemployment, incarceration, and poverty are dramatically higher than average, though as with any group, there is considerable intergroup variation (Armstrong and Rogers 1996; Maxim, White, and Beavon 2003). The inequities faced by Indigenous peoples have their roots in current and past colonial practices on the parts of provincial and federal governments and their representatives. Among the most egregious has been the failure to honor, or to even make, binding treaties. In some cases, the federal government has neglected to provide the basic educational, health, and housing services and resources specified in the Indian Act. (3) In other cases, disputes between specific First Nations and the provinces in which they are located have centered on land dispossession (4) or rights to natural resources, such as fish, wildlife, forestry, and minerals. Such disputes are widespread in Canada, where there are currently over 800 unresolved treaty "claims." (5) Finally, in British Columbia, few land cessation treaties were made and as a result, a majority of the province is still under dispute.

One way in which First Nations have sought justice has been through established legal channels. The courts have upheld Indigenous people's rights in several high-profile decisions. (6) However, going through the courts is costly and time consuming. The Nisga'a treaty (2000), for example, required over 100 years of negotiation. Furthermore, as scholars and activists, such as Taiaike Alfred (2001), Ward Churchill, and Kahn Tineta Horn (7) point out, there is a fundamental and systemic unfairness in the expectation that Indigenous peoples should seek justice through a legal system in which the perpetrator is both party and judge to the allegations of the plaintiff. (8) It is in this context that Indigenous peoples in Canada have used collective action to achieve their goals (though most First Nation groups have used both strategies).

This collective resistance, though dating back to the 1800s, only became fairly regular in practice during the latter part of the twentieth century (Wetzel 2009). It is widely accepted that the current phase began in 1969, when Indigenous people successfully used mass mobilization to block the passage of the White Paper, which was federal legislation aimed at eliminating "Indian Status" and the special rights this status entails (Ramos 2008a, 2008b; Sanders 1985; York 1989). In the decades to follow, the annual number of events began to rise such that by the mid-1980s and 1990s, they were a standard feature of the Canadian national political landscape, drawing much-needed public, and political attention to long-standing grievances with deep roots in colonial history (Long 1997; Ramos 2006; Wilkes 2006). They have involved active and courageous struggles for change at considerable personal risk.

A number of studies of newspaper coverage of this mobilization have considered the coverage that several prolonged standoffs received: particularly the 1990 Oka and the 1995 Ipperwash and Gustafsen Lake crises. In March 1990, the town of Oka in Quebec set in motion plans to expand a local golf course onto land known as the Pines, which Mohawks from the Kanesatake First Nation had been trying to regain for several hundred years. When their court challenge failed, the Mohawks erected a barricade (York and Pindera 1991). In response, on July 11, members of the Surete du Quebec (Quebec Provincial Police force) stormed the barricade at the behest of the town's mayor. The Mohawks fought back, resulting in the death of one police officer, Corporal Marcel Lemay. (9) Lemay's death was followed by a 78-day standoff between the Mohawks and the police, who were later replaced by Canadian Armed Forces. In support of their fellows at Oka, Mohawks from the nearby Kahnawake First Nation blockaded the busy Mercier Bridge into Montreal. (10)

On June 13, 1995 a group of 21 Indigenous people who had been holding annual Sun Dances in the northern interior of British Columbia began an occupation of ranch land at Gustafsen Lake. The argument advanced by the group members and their lawyer Bruce Clark was that Indigenous people in British Columbia had never ceded title to the land (an argument with strong merit given the lack of treaties in that province). The rancher, Lyell James, enlisted the help of the Royal Canadian Mounted Police (RCMP). What followed was a 31-day standoff with 400 police officers which became the "largest and most expensive RCMP operation in Canadian history" (Lambertus 2004:4). The standoff was eventually diffused, primarily through the work of elders. In a later trial, most of the participants were acquitted except the group leader William Jones Ignace. (11)

The 1995 Ipperwash crisis began on July 29, 1995 when protesters from the Stoney Point First Nation occupied and successfully reclaimed a military base on land confiscated by the federal government during the Second World War but never returned. When the protesters later moved into nearby Ipperwash Provincial Park in an attempt to repossess it as well, the response was brutal, and eventually resulted in the killing of an unarmed protester, Anthony Dudley George, shot by an Ontario Provincial Police officer later identified as Kenneth Deane (see Edwards 2001 for a detailed account). Government documents later confirmed the protesters' ownership of the land in question and, in 2004, after almost 10 years of pressure (and several further protest events) a federal inquiry into the shooting was finally initiated.

The research on media coverage of these standoffs has largely focused on story content as opposed to packaging. An analysis of the 208 stories published in 15 Canadian newspapers during the first week of the Oka crisis showed that almost 60 percent framed the conflict as being a law-and-order issue (e.g., Mohawks as criminals) as opposed to a political struggle with deep historical roots (Skea 1993; see also Grenier 1994; Stuart 1993). Nevertheless, the media's acceptance of Mohawk Warriors (12) as warriors also legitimated the latter's assertions of nationhood (Smith et al. 2001; see also Kalant 2007 on these juxtapositions). Studies of media coverage of both Gustafsen Lake and Ipperwash have also found that the media favored officials' version of events, not only legitimating the latter's use of force and violence but also leading to considerable anti-First Nation content (Lambertus 2004; Miller 2005). This reporting pattern was caused in large part by journalists' reliance on "official" sources, such as police, politicians, and outside First Nation leaders (Miller 2005:8). (13)

Although it is clearly important to know about the coverage that each individual standoff received, consideration should also be given to the coverage that these events received in comparison with other collective action events by Indigenous peoples. These standoffs were only a few of several hundred events that have taken place since the 1980s. While none of the other events were prolonged standoffs in the same manner as the above (14) many nevertheless involved the use of very innovative and dramatic tactics, and as a result, have also garnered significant media attention. For example, in the fall of 1985, members of the Haida First Nation on Lyell Island off the coast of British Columbia erected road blockades to prevent logging trucks from entering their traditional territory (Blomley 1996). In 1992, Cree people from James Bay, under the leadership of chief Matthew Coon-Come, canoed from Ottawa to New York City to garner support for a battle against a dam project (Niezen 1998). Two years later, proposed government legislation attempting to tax income earned off-reserve was countered with the occupation of a Revenue Canada office (Globe and Mail December 17, 1994).

The fact that this (and many other) cases of collective have also attracted much media attention, suggests that any general conclusions about media coverage of these events as a whole are premature. By focusing solely on the most dramatic events (in this case standoffs), potential variation across events is ignored. Furthermore, like studies of collective action more generally, while the focus on content is needed, existing studies of Indigenous people's collective action and the media do not consider media packaging as a feature of coverage.

MEDIA PACKAGING

The media package news stories in a variety of ways. Packaging elements can include the volume of coverage (the number of stories), page placement, headlines (both size and content), as well as the length of story and the inclusion of visuals. Events that are the subject of more news coverage are more likely to garner attention (e.g., see Alimi, Gamson, and Ryan 2006; Amenta et al. 2009; Lee and Craig 1992; McKay 1996). Ongoing coverage of an event or issue signals its importance (Carvalho 2008; Carvalho and Burgess 2005), and increases the likelihood that a broader range of readers will receive at least some information about the event, even if they are not daily consumers of newspapers or other news sources. (15) The location and accompanying visual (16) features of an article are also important (Ryan 1991). There is also general agreement that the single most effective way for a story to garner attention is through front-page placement (Davis and McLeod 2003; Tuchman 1978). While packaging is by no means the only important aspect of coverage, (17) it is one about which little is known (except see Amenta et al. 2009) and is a neglected topic in the literature (Amenta et al. 2009).

In light of the importance and lack of attention to media packaging of protest, what accounts for different quality and quantity of media packaging an event or issue will receive? Two major theories have been advanced to explain how the media gathers and selects what news is "fit to print." First, the media follow a set of "news routines," which lead them to select and feature certain types of stories and events. Second, the media seeks to present stories that are dramatic and sensational, providing high "news values." Each of these theories leads to different expectations about how Indigenous people's collective action will be covered in the media.

There is strong evidence from a host of empirical studies (McCarthy, McPhail, and Smith 1996; Myers and Caniglia 2004; Smith et al. 2001; Wilkes and Ricard 2006) demonstrating the important relationship between event characteristics and the media's organizational practices. In particular, the "news gathering" perspective with its emphasis on news routines and news values suggests not only that media attention will vary across events, but also that different events will receive different types of media attention including different media packaging.

News routines are the day-to-day organization and structural aspects of news production. Because newspapers still have to be published irrespective of whether a given day has any newsworthy events, events occurring before slow news days (usually Mondays) are more likely to be included in the newspaper than those occurring on busier news days (Myers and Caniglia 2004; Oliver and Myers 1999). A number of scholars have also pointed out that reporters have regular newsbeats--either locations (such as city hall) where they routinely go (see, e.g., Oliver and Myers 1999; Ryan 1991) or issues for which they are regularly responsible. In the case of the latter, the assignment of a "beat" reporter is a signal of an issues' perceived importance (Eliasoph 1988).

Staff time and resources are also saved (and page space used up) when editors use reports furnished by newswires (Ryan 1991). Newswire reports provide news from far-away locations as well as more generic human-interest stories. Yet, because the news is subject to its own distinct internal organizational dynamics, the competitive nature of journalism is such that newswire reports are not often featured as prominently as those written by in-house staff (Sumpter 2000). The use of newswires also reflects editors' judgment about whether or not an event merits having a journalist investigate and write an original story. Finally, different newspapers have different style formats ranging from tabloid to broadsheet, longer or shorter average story length, and more or fewer visuals. Papers also have different audience mandates that require different types of news ranging from nationally relevant to smaller and more local interest stories.

Clearly the filtering these types of routine elements provide is likely to shape the type of packaging a story receives (it is also likely that some events will receive more coverage than others). In terms of the day of publication, Indigenous people's collective action events that appear in the Monday edition of the paper should be given more prominent packaging than those that appear in stories on other days of the week. (18) An article that is written by a beat reporter should also receive more prominent packaging than should one written by an in-house reporter (as opposed to a wire). Finally, we might also expect to find some differences across papers in terms of the kinds of events and the kind of attention each receives.

In addition, because newspapers are a consumer product, they also need to feature stories that are of interest to their readers to maintain their circulation numbers. A story is said to have "news value" if it has some or all of the following characteristics: relevance, immediacy, novelty/innovation, and drama (Boyd 1994; Bridges 1989; Bridges and Bridges 1997; Shoemaker and Reese 1996). In terms of relevance, events that involve more people will appear to be more relevant and important to a broader spectrum of readers than events involving a small group (see, e.g., Mueller 1997; Neveu 2002). Immediacy refers to the "newness" and timeliness of a story. Recent "episodes" are more newsworthy than ongoing "issues." Novelty refers to anything that is seen to depart from the usual. In the case of protest, scholars have noted for some time now that marches and demonstrations are relatively routine and predictable ways to air grievances (McCarthy, McPhail, and Crist 1999; see also Tilly 1978). This normalization of protest may be a primary reason why the empirical evidence does not show a significant difference between media coverage of different event forms, such as marches, demonstration, vigils, and literature tables (e.g., Oliver and Maney 2000; Oliver and Myers 1999). (19) Finally, events are perceived as dramatic if they involve violence, death, and/or destruction--"If it bleeds it leads" (Gitlin 1980). For example, Myers and Caniglia (2004) found that riots with higher levels of violence and greater numbers of arrests were more likely to be reported in the New York Times.

Many First Nations activists are well aware of the media's need for dramatic and newsworthy events and of the importance of the media more generally. Accordingly, they have tailored their protests in ways that facilitate bold headlines and front-page placement. Lubicon Chief Bernard Ominyak and his supporters were clearly aware of this need when, in order to draw media attention to their land dispute with the Alberta government, they parodied the 1988 Calgary Olympic torch run logo "Share the Flame" by changing it to "Shame the Flame." The 1985 blockades of Lyell Island by Haida people resulted in media images of elders adorned in traditional Haida dress being arrested and hauled away from the site. The associated content appears to have been supportive, suggesting that the Haida had little choice but to engage in civil disobedience ("So what choice do the Haida have?" Vancouver Sun November 21, 1985:A5). Similarly, in 1987 widespread media interest was also generated when Peguis Chief Louis Stevenson extended a controversial invitation to South African Ambassador Glenn Babb to tour his reserve in northern Manitoba. (20)

While the above events are dramatic and contentious and should have garnered prominent packaging, they are also notable for their absence of violence. In contrast, the protracted standoffs discussed in the previous section (Oka, Mercier, Ipperwash, and Gustafsen Lake) could be labeled as contentious, though in global terms they are still marked by extremely low levels of violence. Furthermore, even though these events may have started out as direct action, even their participants ultimately ended up having some relationship with the media. (21) For example, in the case of Oka, not only did the warriors have a spokesperson (Ellen Gabriel) and in some cases refused to talk with reporters whom they found unsympathetic, but some media photographs also show Warriors reading the newspaper or watching TV while manning the barricades. We therefore investigate the extent to which the factors associated with quality and quantity of coverage received by the standoffs is similar to, or different from, other events.

DATA

We collected data on events that took place from 1985 to 1995. This time period was selected because it was a major period for mobilization by Indigenous peoples in Canada (Ramos 2006). We also discuss the coverage received by several more recent events in the discussion section of this paper. Data were collected from two major daily newspapers, the Globe and Mail and the Vancouver Sun. (22) The Globe and Mail, based in Toronto, is one of two national newspapers and has the second highest circulation of any newspaper in the country. (23) Canada's other national newspaper--the National Post--did not exist during the period considered for this study. We selected the Globe due to its national presence and the Vancouver Sun because it is located in British Columbia, the province in which the most protests by Indigenous peoples have taken place. The Vancouver Sun is the largest circulating paper in Vancouver. While there are clearly many other papers that could have been selected, and some, which likely provided more detailed coverage of particular events, we elected to go with these two papers on the grounds that they are papers that a wide number and range of Canadians read.

To find information about Indigenous people's collective action events, we searched both the titles listed on the paper index and also used a keyword search of the electronic versions of Canadian Newsstand--an index of major daily newspapers. We searched the electronic index using general keywords, such as "Native," "Aboriginal," "First Nation," "Indian," and "protest," "demonstration," "blockade." We then refined the search to locate all possible articles about each individual event using event-specific keywords (e.g., "Mohawk" and "Oka"). Finally, to ensure that we had as comprehensive a database as possible, we also searched using multiple spellings of proper names, such as "Gitksan" and "Gitxsan."

Because only a very limited number of events received coverage beyond that provided within the calendar year of the event's occurrence we restricted the number of articles collected about particular events to those published within the same calendar year that the event took place. In cases where the event lasted for more than one year, we included articles published until the final year of the protest. This limited our data set to articles with a primary focus on the given event (with a few exceptions which focused on several events at the same time). Past events mentioned in passing, years later, were therefore excluded from our sample. While this means that our data do not speak to shifts in coverage over time, from a practical point of view these shifts only apply to a small number of events. Our final data set consists of 1,801 articles about 230 events (24) reported on in the Globe and Mail and the Vancouver Sun.

Packaging

We use the following variables to indicate packaging: the number of articles, page placement (front page versus other), the length of the article (measured by the word count), and whether an image accompanied the article. (25) While this is by no means an exhaustive list of packaging elements, this group serves to demonstrate the importance of such practices. (26)

News routines are measured by variables indicating the day of the week of publication (Monday versus all others); whether or not the article was obtained from a newswire; whether or not the article was written by an "Aboriginal Beat" reporter; and a dummy variable indicating that the paper was the Globe and Mail (versus the Vancouver Sun). Because the beats of reporters were rarely indicated we classified any article written by a reporter who was the author of more than 20 articles about these events as a beat reporter. Using this definition, 10 of the 183 reporters who wrote stories about these events could be classified as beat reporters. (27) We also include a dummy variable indicator of the paper in which the article appeared, recognizing that different papers have different styles of placement and reporting.

News values are indicated by measures of event size, timing/immediacy, and tactics. (28) The size of the event is a count of the total number of people who participated in the event. (29) This ranged from a low of three people who stopped the construction of a mining access road in 1985 to a high of 2,500 for a 1986 rally in front of the Ontario legislature. Because it is unlikely that an increase of one person would yield a meaningful result we divided this measure by 100. Thus, all subsequent multivariate analyses are to be interpreted as the effect that increments of 100 participants have on the [beta]-coefficients. Immediacy is measured as the number of days between the publication of the article and the start day of the event. (30) This ranged from -5 (for an announcement that appeared five days before a 1989 blockade) to 292 days for an article that appeared months after the end of the Oka crisis. We subsequently divided this measure by 30 so that a one-unit change refers to a monthly rather than daily change.

We coded event tactics as a series of dummy variables indicating whether the event was a march/demonstration, a road blockade/closure, a rail blockade, a land occupation, or other (such as building occupation, fish/ log/hunt-in, (31) dam diversion, and hunger strike). We also coded whether an event was a standoff (a dummy variable). Because the level of violence of these events is especially, low we use this measure to indicate drama rather than the number of deaths and injuries (e.g., see Myers and Caniglia 2004). (32) A further issue is that the characteristics of the events we consider do not conform as neatly as we might have liked to standard statistical assumptions. The measure of standoffs is confounded with tactics. (33) That is, all standoffs occurred during road blockades (Mercier) or land occupations (Oka, (34) Ipperwash, Gustafsen Lake) but never occurred during marches or rail blockades. Although we considered coding this measure in a number of alternative ways (35) in the end, the confounding could not be completely eliminated. We explain how we addressed this issue in the methods section below.

METHODS

In addition to descriptive statistics, we use a range of multivariate methods to analyze the data. Because the number of articles is a count (and the mean of this variable exceeds the variance) we use negative binomial regression. The models for front page and picture variables involve dichotomous dependent variables, and we therefore use binary logistic regression. (36) However, to account for the correlation between outcomes corresponding to the same events we used logistic generalized estimating equations (GEE) with an exchangeable correlation structure. The full details of this method are provided in the Appendix.

[FIGURE 1 OMITTED]

FINDINGS

Variation in Packaging across Events

Figure 1 shows the monthly frequency counts for the number of events, the number of articles, and the number of front-page articles. It clearly illustrates that the packaging received by events is not a one to one ratio in terms of the number of events. The number of articles generated is much higher than the actual number of events and much higher at certain times than at others.

Table 1 shows the frequencies for the different packaging elements and how these varied depending on whether events were covered by the Vancouver Sun, the Globe and Mail or both. Of the 230 events, the average event generated just over seven articles and articles of about 550 words. Seventeen percent of articles appeared on the front page and almost 30 percent were accompanied by an image. The results in this table also show, as we would expect, that the coverage accorded those events covered in only one paper was substantially less than the coverage of events covered in both papers. Also of note in this table is the much greater likelihood of events covered by the Vancouver Sun to be accompanied by a picture. Finally, the results clearly show that if an event was deemed important enough to be covered by both papers, it was also deemed important enough to be given more prominent packaging as well.

Table 2 provides the differences in event characteristics for all 230 events and by newspaper. About 10 percent of events could be classified as violent, although as noted previously there is considerable variation within this category. The average event had over 100 participants and lasted about 27 days. (38) There was considerable variation in terms of the tactics used. Only one-quarter of events were marches and demonstrations. Road blockades were the most common event (at least as reported by the media) comprising over 38 percent of events, followed by land occupations (10 percent of events). The majority of events took place in British Columbia or Ontario and involved a single First Nation. Most years had in the range of 10 to 20 events, the exception being 1988, 1989, and 1990, which had 30, 40, and 60 events, respectively.

Turning to events covered by single versus multiple papers we would expect the more dramatic events to get coverage in both papers and for local papers to have an increased likelihood of covering local events. Only one of these expectations is supported by the data. Violent events are more likely to be covered by both papers. However, contrary to expectations if an event in British Columbia was covered by only one paper it was more likely that this paper would be the Globe and Mail than the Vancouver Sun. Seventy-five percent of events covered only by the Globe and Mail took place in British Columbia. Similarly, if an event in Ontario was covered by only one paper it was more likely to be by the Vancouver Sun than by the Globe and Mail. Fifty-three percent of events covered only by the Vancouver" Sun took place in Ontario. If the event was covered by both papers then it is most likely to have taken place in British Columbia (44 percent).

Table 3 provides the distribution of coverage across events. Almost half of the events were only covered by one article. These events in total only generated 5 percent of the coverage all the events received. In contrast the most covered event, the 1990 Oka crisis, was covered by over 400 articles in the two papers, generating almost one-quarter of all coverage. The ten most covered events generated about 50 percent of all coverage received. While it is not surprising that some events generate more coverage than others, the imbalance in the coverage means that some events (the most dramatic and/ or violent) are more likely to be seen as "representative" of these events in reader's minds.

Table 4 contains the frequency counts for all the packaging elements with the exception of the length of the article, which is a mean. This table contains the cross-classification of these packaging elements with other coverage characteristics as well as by event characteristics themselves. Turning to the article counts, of note is that the majority of articles were about road blockades or land occupations. In contrast, marches only generated 4 percent of articles, even though 23 percent of events were marches. The column for front-page coverage shows that 317 events, 17.6 percent, received front-page coverage. Over half of this front-page coverage was about standoffs. About one-third of articles were accompanied by a picture but again over half of the pictures appeared with stories about standoffs. Taken together the results from Tables 1 to 4 clearly establish that while the media cover a wide range of events there is far less equality in terms of the packaging these events receive.

Factors Associated with Differential Packaging

Table 5 presents the results of the GEE models showing the association between coverage characteristics, event characteristics, and media packaging. In order to ensure that the disproportional attention received by standoffs does not skew the results, we present two models for each type of coverage. In the first model all events are included and in the second model the standoffs are removed. Models 1a and 1b illustrate how event characteristics that are or are not newsworthy are related to the number of articles (beyond the first one) an event receives. Because this model was at the event rather than the article level we do not include the measures of coverage characteristics. The results show that although large events receive marginally more coverage, this is not the case for longer events. Turning to the effects of tactics, the results indicate that all other tactics generate more coverage than marches and demonstration. Model lb shows that the effect of land occupations decreases from 3.18 to 1.94 when standoffs are omitted. The fit statistics indicate that overall these models result in a significant improvement in model fit over an intercept-only model.

In models 2a and 2b we examine the factors associated with front-page coverage. Because the unit of analysis is now at the article level we replaced the measures of the length of event with a variable indicating the number of days between the start day of the event and the article publication date. (39) These models also contain the news routines variables. The Globe and Mail was no more likely than the Vancouver Sun to place articles about these events on the front page. Nor was there any difference in terms of events that appeared in Monday articles (40) or in terms of the timing of the article. However, the results show that tactics have a strong association with front-page coverage. In particular, road and rail blockades are considerably more likely than marches to appear on the front page. In contrast, land occupations including standoffs are no more likely to appear on the front page than marches and demonstrations. The fit statistics (Quasi Likelihood under the Independence model Criterion) indicate that when standoffs are included these models do not have strong explanatory power for front-page coverage. Without the standoffs the results show that tactics are a better predictor of this type of packaging than news routines.

Finally, models 3a and 3b show the coefficients from models predicting whether an article was accompanied by a picture. The presence of an image seems to have far more to do with the news routines of the media than with the characteristics of the events themselves. While articles written by beat reporters were more likely to have a visual, articles published on Mondays and in the Globe and Mail were less likely to have one, likely reflecting the space and publishing practices across days and papers. Although it is not entirely clear why the tactic has so little to do with the inclusion of pictures one reason might be that the events that involved more contentious tactics also tended to take place in more remote locations. The model fit statistics show a similar pattern to that for front-page coverage in that the models without standoffs provides a much better fit to the data.

DISCUSSION

In summary, the results clearly indicate that while tactic escalation increases the amount of coverage, only disruptive tactics are more likely to appear on the front page. Whether an article will be accompanied by a picture is largely determined by news routines as opposed to tactics. The results also show that the coverage of standoffs is much less predictable than other types of events.

The events we consider in the paper end in 1995. How might these results look had we considered a more recent set of events? In the interim, other events (we have found at least 130 additional events from 1996 to 2008), such as the ongoing land conflict at Caledonia and the 2001 blockade at Sun Peaks, BC, have taken place. We have little reason to believe that tactic escalation is less important now than it was during the late 1980s and early 1990s. However, the results for the inclusion of visuals might be quite different had we considered more recent events. The advent of on-line newsmedia has dramatically changed the nature of news production. The transfer of high-quality images is much simpler and less costly than it was in the past. News routines may therefore have less to do with whether an article is accompanied by a picture. Nevertheless, additional research is clearly needed to ascertain the extent to which these hypotheses hold true.

CONCLUSION

This paper is the first study to compare media coverage across multiple instances of collective action by Indigenous peoples. While a number of studies have considered single and high-profile events, no study has attempted to consider the range of events in their entirety. What factors distinguish the quality and quantity of coverage an event receives? Our results show that in the case of collective action by Indigenous peoples, the size and length of the event have very little association with coverage. These results stand in contrast to the results of studies of collective action by other groups (e.g., McCarthy et al. 1996; Mueller 1997; Smith et al. 2001), likely because in comparison, the protests under consideration in this study are very small.

Despite the small size of many of the events in this study, they received a large quantity of coverage. Therefore, it seems that the use of contentious tactics provides an excellent strategy for overcoming the disadvantage of small size. In addition, the fact that prolonged standoffs (recognizing that this may not have been activist's initial intent) with authorities generated prominent packaging is clearly in line with previous research linking violence to media coverage (McAdam and Su 2002). However, the findings also show that being contentious or unusual is not a guarantee of high-profile packaging. If this were the case, there would have been a strong relationship between land occupations and media packaging. The tactics that did generate prominent packaging--road and rail blockades--can be distinguished from land occupations in terms of their capacity to disrupt (41) the lives of outsiders. As long as it is an extended event, tactics such as preventing people from getting to work or home or preventing supplies from reaching their destinations, work well in this regard.

The findings underscore the need to pay more attention to differences in media attention across events. While contention is clearly associated with differential packaging the magnitude of difference in the effect sizes is worth commenting on. Standoffs generated significantly more attention across multiple packaging elements than other forms of contention. Although there were several hundred events, they generated nearly 40 percent of the articles and half of all front-page coverage. This means that the difference in packaging across events, like academic citations patterns or salaries, is exponential rather than linear (see also Biggs 2005; Taleb 2007).

We believe that this exponential packaging matters for our treatment of mobilization as both a cause and a consequence of other outcomes. In the standard model, mobilization (measured as an event count, the rate of mobilization) is caused by a combination of resources and opportunities (e.g., see Khawaja 1994; Okamoto 2003; Ramos 2008a, 2008b; Wilkes 2004). (42) Underlying this strategy is an assumption that events are roughly equivalent, irrespective of the differing media packaging one event might have compared with another. Yet, the differential packaging across events probably means that the widely used practice of simply counting collective action events or newspaper mentions of social movement organizations is more problematic than we might have assumed. The greater media attention some events receive is clearly communicating messages about the importance of the event or issue. One could argue that recent studies that model the relationship between an event's drama and policy outcomes (McAdam and Su 2002; Santoro 2002) address this issue. Yet, because drama is operationalized in terms of the level of violence rather than the level and prominence of packaging (43) the exponential differences in attention across events may not be captured. For example, while the Watts riot may have involved twice as many deaths as other riots that took place in the United States during the 1960s, its coverage was in all likelihood exponentially greater. What is showing up as the effect of multiple events may be the result of something more singular (see also Burstein and Sausner 2005). This fact needs to be incorporated into our models of the cause and effect of contentious action.

APPENDIX: DETAILS OF GEE MODELS PRESENTED IN PAPER

The models were fit using Proc Genmod with Repeated statements in SAS. The binary outcome [Y.sub.ij] (e.g., front page versus not), for the jth article on the ith event, is modeled as:

[Y.sup.fr.sub.ij] ~ Bernoulli([p.sub.ij]), cor([Y.sub.i1j1], [Y.sub.i2j2]) = [[rho].sup.fr]1{il = i2},

log([p.sub.ij]/(1 - [p.sub.ij])) = [[[beta].sup.fr.sub.0] + [[alpha].sup.fr.sub.1] [X.sub.ij1] + ... + [[alpha].sup.fr.sub.p] [X.sub.ijp] + [[beta].sup.fr.sub.1]][Z.sub.i1] + ... + [[beta].sup.fr.sub.q] [Z.sub.iq]

where [X.sub.ijk] are known article characteristics (either numerical or dummy 0,1 variables), [Z.sub.ik] are known event characteristics, [[alpha].sup.fr.sub.k] are unknown article effects, [[beta].sup.fr.sub.k] are unknown event effects, and [[rho].sup.fr] is the unknown within-event correlation to be estimated. (44)

The count of the number of articles per event is modeled as a generalized linear model with a negative binomial (45) response and logarithm link function. Because negative binomial models require zeros and our count measure begins at one, we subtracted one from all responses. This means that the count of the number of articles refers to the number of additional articles beyond the first one. There are no within-event correlation issues with this model because event data have been collapsed to one row per event. Thus, the characteristic of each event is only counted for one and not for multiple outcomes. The number of articles published on event i = 1,..., 229 is assumed to be

([Y.sup.ac.sub.i] -1) ~ NegBin([m.sub.i] , k). with

Log(E[[Y.sup.ac.sub.i] - 1]) = log([m.sub.i]) = [[beta].sup.ac.sub.0] + [[beta].sup.ac.sub.1][Z.sub.i1] + ... _ [[beta].sup.ac.sub.q] [Z.sub.iq]

where [Z.sub.ik] are observed event characteristics, k and [[beta].sup.ac.sub.k], k = 0, ..., q are unknown parameters to be estimated.

We thank Rich Carpiano, Sylvia Fuller, Tamara Ibrahim, Scan Lauer, Dina Okamoto, Pamela Oliver, and Howard Ramos for assistance and advice. We also thank three anonymous reviewers for their helpful comments. Funding for this research was provided by a grant from the Social Sciences and Humanities Research Council of Canada.

References

Alfred, T. 2001. "Deconstructing the British Columbia Treaty Process." Balayi: Culture, Law and Colonialism 3:37-66.

Alfred, T. and J. Corntassel. 2005. "Being Indigenous: Resurgences Against Contemporary Colonialism." Government and Opposition 40:597-614.

Alfred, T. and L. Lowe. 2005. "Warrior Societies in Contemporary Indigenous Communities." Report to the Ipperwash Inquiry, 65 pp.

Alimi, E., W.A. Gamson and C. Ryan. 2006. "Knowing Your Adversary: Israeli Structure of Political Opportunity and the Inception of the Palestinian Intifada." Sociological Forum 4:535-57.

Allison, P. 1999. Logistic Regression Using the SAS System. North Carolina: SAS Institute.

Amenta, E., N. Caren, S.J. Olasky and J.E. Stobaugh. 2009. "All the Movements Fit to Print: Who, What, When, Where, and Why SMO Families Appeared in the New York Times in the Twentieth Century." American Sociological Review 74:636-56.

Andrews, KT. 2008. "Making the News: Movement Organizations, Media Attention, and the Public Agenda." Unpublished manuscript obtained from author.

Armstrong, R. and T. Rogers. 1996. A First Nation Typology: Patterns of SocioEconomic Well-Being. Ottawa: Department of Indian Affairs and Northern Development.

Barakso, M. and B.F. Schaffner. 2006. "Winning Coverage: News Media Portrayals of the Women's Movement, 1969-2004." Press/Politics 11:22-44. Biggs, M. 2005. "Strikes as Forest Fires: Chicago and Paris in the Late 19th Century." American Journal of Sociology 96:144-85.

Blomley, N. 1996. "'Shut the Province Down': First Nations Blockades in British Columbia, 1984-1995." BC Studies 111:5-35.

Boyd, A. 1994. Broadcast Journalism, Techniques of Radio and TV News. Oxford: Focal.

Boyle, M.P., M.R. McCluskey, D.M. McLeod and S.E. Stein. 2005. "Newspapers and Protest: An Examination of Protest Coverage from 1960 to 1999." Journalism and Mass Communication Quarterly 82:638-53.

Bridges, J.A. 1989. "News Use on the Front Pages of the American Daily." Journalism Quarterly 66:332-37.

Bridges, J.A. and L. Bridges. 1997. "Changes in News Use on the Front Pages of the American Daily Newspaper, 1986-1993." Journalism and Mass Communication Quarterly 74:826-39.

Burstein, P. and S. Sausner. 2005. "The Incidence and Impact of Policy-Oriented Collective Action: Competing Views." Sociolagical Forum 20:403-19.

Carvalho, A. 2008. "Medialtedt Discourse and Society: Rethinking the Framework of Critical Discourse Analysis." Journalism Studies 9:161-77.

Carvalho, A. and J. Burgess. 2005. "Cultural Circuits of Climate Change in U.K. Broadsheet Newspapers, 1985-2003." Risk Analysis 25:1457-69.

Davis, H. and L.S. McLeod. 2003. "Why Humans Value Sensational News: An Evolutionary Perspective." Evolution and Human Behavior 24:208-16.

Edwards, P. 2001. One Dead Indian: The Premier, the Police and the Ipperwash Crisis. Toronto: McClelland and Stewart.

Eliasoph, N. 1988. "Routines and the Making of Oppositional News." Critical Studies in Mass Communication 5:313-34.

Frideres, J.S. and R.R. Gadacz. 2008. Aboriginal Peoples in Canada. Toronto: Pearson Education Canada.

Gitlin, T. 1980. The Whole World is Watching. Berkeley, CA: University of California Press.

Grenier, M. 1994. "Native Indians in the English-Canadian Press: The Case of the 'Oka Crisis'." Media, Culture and Society 16:313-36.

Kalant, A. 2007. National Identity and the Conflict at Oka. New York: Routledge.

Khawaja, M. 1994. "Resource Mobilization, Hardship, and Popular Collective Action in the West Bank." Social Forces 73:191-211.

Koopmans, R. and S. Olzak. 2004. "Discursive Opportunities and the Evolution of Right-Wing Violence in Germany." American Journal of Sociolagy 110:198-230.

Lambertus, S. 2004. Wartime Images, Peacetime Wounds: The Media and the Gustafsen Lake Standoff. Toronto: University of Toronto Press.

Lee, J. and R.L. Craig. 1992. "News as an Ideological Framework: Comparing US Newspapers' Coverage of Labor Strikes in South Korea and Poland." Discourse and Society 3:341-63.

Long, D. 1997. "Culture, Ideology, and Militancy: The Movement of Native Indians in Canada, 1969-91." Pp. 118-34 in Organizing Dissent, edited by W. Caroll. Toronto: Garamond Press.

Maxim, P., J. White and D. Beavon. 2003. "Dispersion and Polarization of Incomes among Aboriginal and Non-Aboriginal Canadians." Pp. 222-47 in Aboriginal Conditions: Research as a Foundation for Public Policy, edited by J. White, P. Maxim and D. Beavon. Vancouver: University of British Columbia Press.

McAdam, D. 1996. "The Framing Function of Movement Tactics: Strategic Dramaturgy in the American Civil Rights Movement." Pp. 338-56 in Comparative Perspectives on Social Movements, edited by D. McAdam, JD. McCarthy and M. Zald. Cambridge, UK: Cambridge University Press.

McAdam, D. and Y. Su. 2002. "The War at Home: Antiwar Protests and Congressional Voting, 1965-1973." American Sociological Review 67:696-721.

McCarthy, J.D., C. McPhail and J. Crist. 1999. "The Emergence and Diffusion of Public Order Management Systems: Protest Cycles and Police Response." Pp. 49-69 in Globalization and Social Movements, edited by H. Kriesi, D. della Porta and D. Rucht. London: McMillan.

McCarthy, J.D., C. McPhail and J. Smith. 1996. "Images of Protest: Dimensions of Selection Bias in Media Coverage of Washington Demonstrations, 1982 and 1991." American Sociological Review 61:478-99.

McKay, J. 1996. "Reflecting the Hazard or Restating Old Views: Newspapers and Bushfires in Australia." International Journal of Mass Emergencies and Disasters 14: 305-19.

Miller, J. 2005. "Ipperwash and the Media: A Critical Analysis of How the Story was Covered." Report submitted to Aboriginal Legal Services of Toronto.

Mueller, C. 1997. "International press coverage of East German protest events, 1989." American Sociological Review 62:820-32.

Myers, D. and B. Caniglia. 2004. "National Newspaper Coverage of Civil Disorders, 1968-1969." American Sociological Review 69:519-43.

Neveu, E. 2002. "The Local Press and Farmers' Protests in Brittany: Proximity and Distance in the Local Newspaper Coverage of a Social Movement." Journalism Studies 3: 53-67.

Niezen, R. 1998. Defending the Land: Sovereignty and Forest Life in James Bay Cree Society. Boston, MA: Allyn & Bacon.

Okamoto, D. 2003. "Toward a Theory of Panethnicity: Explaining Asian American Collective Action." American Sociological Review 68:811-42.

Oliver, P. and G. Maney. 2000. "Political Processes and Local Newspaper Coverage of Protest Events: From Selection Bias to Triadic Interactions." American Journal of Sociology 106:463-505.

Oliver, P. and D. Myers. 1999. "How Events Enter the Public Sphere: Conflict, Location, and Sponsorship in Local Newspaper Coverage of Public Events." American Journal of Sociology 105:38-87.

Ramos, H. 2006. "What Causes Canadian Aboriginal Protest? Examining Resources, Opportunities and Identity, 1951-2000." Canadian Journal of Sociology 31: 211-34.

Ramos, H. 2008a. "Aboriginal Protest." Pp. 55-70 in Social Movements, edited by S. Staggenborg. Don Mills: Oxford University Press.

Ramos, H. 2008b. "Opportunity for Whom? Political Opportunity and Critical Events in Canadian Aboriginal Mobilization, 1951-2000." Social Forces 97:795-823.

Ryan, C. 1991. Prime Time Activism: Media Strategies for Grassroots Organizing. Boston, MA: South End Press.

Sanders, D.E. 1985. "The Indian Lobby and the Canadian Constitution, 1978-82." Pp. 151-89 in Indigenous Peoples and the Nation-States: Fourth World Politics in Canada, Australia and Norway, edited by N. Dyck. St. John's, NF: Institute of Social and Economic Research, Memorial University of Newfoundland.

Santoro, W.A. 2002. "The Civil Rights Movement's Struggle for Fair Employment: A "Dramatic Events-Conventional Politics" Model." Social Forces 81: 177-206.

Shoemaker, P. and S. Reese. 1996. Mediating the Message: Theories of Influence on Mass Media Content. White Plains, NY: Longman.

Skea, W.H. 1993. "The Canadian Newspaper Industry's Portrayal of the Oka Crisis." Native Studies Review 9:15-31.

Smith, J., J. McCarthy, C. McPhail and B. Augustin. 2001. "From Protest to Agenda-Building: Description Bias in Media Coverage of Protest Events in Washington, D.C." Social Forces 79:1397-423.

Snow, D.A. and C. Corrigall-Brown. 2004. "Why Frames Sometimes Fall on Deaf Ears: The Problems of Misalignment, Scope, Exhaustion, and Relevance." Pp. 222-38 in Rhyming Hope and History: Activists, Academics, and Social Movement Scholarship, edited by D. Croteau, W. Hoynes and C. Ryan. Minneapolis, MN: University of Minnesota Press.

Snow, D.A., R. Vliegenthart and C. Corrigall-Brown. 2007. "Framing the French Riots: A Comparative Study of Frame Variation." Social Forces 86:385-415.

Staggenborg, S. 1993. "Critical Events and the Mobilization of the Pro-Choice Movement." Research in Political Sociology 6:319-45.

Steele, S. 1997. "Gustafsen Lake Standoff: 15 Charged." MacLean's, June 2.

Stuart, C. 1993. "The Mohawk Crisis: A Crisis of Hegemony: An Analysis of Media Discourse." M.A. thesis, University of Ottawa.

Sumpter, R.S. 2000. "A Case Study: How to Get on or Stay off the Front Page." Newspaper Research Journal, 9 pp.

Taleb, N.N. 2007. The Black Swan: The Impact of the Highly Improbable. New York: Random House.

Tilly, C. 1978. "The Routinization of Protest in Nineteenth Century France." CRSO Working Paper 181, 12 pp.

Traugott, M. 1993. "Barricades as Repertoire: Continuities and Discontinuities in the History of French Contention." Social Science History 17:309-23.

Tuchman, G. 1978. Making the News. New York: Free Press.

Vliegenthart, R., D. Omegema and B. Klandermans. 2005. "Media Coverage and Organizational Support in the Dutch Environmental Movements." Mobilization 10:365-81.

Wetzel, C. 2009. "Theorizing Native American Land Seizure: An Analysis of Tactical Changes in the Late Twentieth Century." Social Movement Studies 8:15-32.

Wilkes, R. 2004. "First Nation Politics: Deprivation, Resources, and Participation in Collective Action." Sociological Inquiry 74:570-89.

Wilkes, R. 2006. "The Protest Actions of Indigenous Peoples: A Canadian-U.S. Comparison of Social Movement Emergence." American Behavioral Scientist 50:510-25.

Wilkes, R. and D. Ricard. 2006. "How Does Newspaper Coverage of Collective Action Vary?: Protest by Indigenous People in Canada." The Social Science Journal 44:231-51.

Wolfsfeld, G. 1997. Media and Political Conflict." News from the Middle East. Cambridge, UK: Cambrdige University Press.

York, G. 1989. The Dispossessed: Life and Death in Native Canada. Toronto: McArthur.

York, G. and L. Pindera. 1991. People of the Pines. Toronto: Little Brown and Company.

Young, C. 2009. "Model Uncertainty in Sociological Research: An Application to Religion and Economic Growth." American Sociological Review 74:380-97.

Court Cases Cited

Calder v. British Columbia (Attorney General) [1973] S.C.R. 313, [1973[ 4 W.W.R. 1 Delgamuukw v. British Columbia [1997] 3 S.C.R. 1010

R. v. Marshall (No. 1) [1999] 3 S.C.R. 456

RIMA WILKES University of British Columbia

CATHERINE CORRIGALL-BROWN University of British Columbia

DANIEL J. MYERS University of Notre Dame

(1) We use the term collective action in order to recognize that seine of the events under consideration are protest events while others are better categorized as direct action (see Blomley 1996) or resistance.

(2) The Canadian federal government officially recognizes three groups of indigenons peoples (as Aboriginal peoples) Indian (both status and nonstatus). Metis, and Inuit (for a critical discussion of the term "Aboriginal" see Alfred and Corntassel 2005). There are currently over 600 bands/First Nations in Canada (for a detailed overview see Frideres and Gadacz 2008). The figure for population size holds for the current year as well as for the time period under investigation 1985 to 1995. It is based on Canadian Census reports of the number of people who reported some Aboriginal ancestry. The percentage would decrease if a different definition, such as the registered Indian population, were used.

(3) The Indian Act is federal legislation that spells out some of the rights of Indigenous peoples as well as the duties of the federal governments in this regard.

(4) For example, in British Columbia, the provincial government has given companies logging rights on First Nations' lands that have never been ceded.

(5) The Canadian government favors the use of the term claim. However, this term is biased in the government's favor as to say someone made a "claim" is to suggest doubt as to its veracity. There are two types of land claims in Canada--specific and comprehensive. Specific claims concern the administration of existing treaties and laws. In contrast, comprehensive claims concern land and title rights that have not already been settled through existing treaties.

(6) Examples include Calder v. the Attorney General of British Columbia (1973), Delgamuukw v. British Columbia (1997), and R. v. Marshall (1999).

(7) First author heard Ms. Horn speak on this matter on a radio show, though no source information is available, Churchill's comments were made in a public lecture at the University of British Columbia.

(8) The same critique has been made of government designed treaty negotiation programs.

(9.) No one was ultimately charged with Corporal Lemay's death.

(10.) The Oka crisis and Mercier Bridge occupation set oft what has since become known as the "Indian Summer" of 1990 (Ramos 2008a, 2008b), in which Indigenous people from across the country, sometimes with the aid of outsiders, displayed their solidarity with the Mohawks by engaging in marches, demonstrations, road blockades, and rail blockades. Among the most comprehensive accounting of the standoff(s) is provided by documentary filmmaker Alanis Obomsawin (who was with the Mohawks at Oka at the time) in a series of films: Kanehsatake: 270 Years of Resistance (1993); My Name is Kahentiiosta (1995); Spudwrench: Kahnawake Man (1997); and Rocks at Whiskey Trench (2000). Rocks at Whiskey Trench provides footage of residents and commuters from Chateauguay, in one of the most disturbing incidents of contemporary Canadian racism, stoning a convoy of vehicles (containing women, children, and elders) that were fleeing the Kahuawake reserve.

(11.) The trial judge refused to apply international law to the case or to consider arguments made about genocide (Steele 1997).

(12) For an excellent overview of Warrior societies see Alfred and Lowe (2005).

(13) In a later turn of events leaked police film footage from the Ipperwash scene showed its own officers making racist remarks about Native Peoples. The ongoing use journalists made of police sources clearly affected the tone of their reports. Although in most cases, outside First Nation leaders are supportive of the protesters, this was not the case for Ipperwash where there was a split between the protesters and some elected band council members (Miller 2005).

(14) There were some events, such as the attempts by Peigan Lonefighters to divert water from the Oldman dam project, that also involved confrontation with authorities. However, the confrontation tended to be extremely short, typically one day.

(15.) Studies have documented variation in the number of' news stories garnered by women's movement organizations in the United States (Barakso and Schaffner 2006), environmental social movement organizations in the United States (Andrews 2008), and Palestinian and right-wing social movement groups in Israel [Wolfsfeld 1997). Wolfsfeld also documents the difference in the number of media stories the Palestinians received compared with subnational challengers in Ireland, Spain, Iraq, and Indonesia.

(16.) Visuals, such as photographs, not only draw attention to a story but the selection of a preferred visual framing can work to counter stereotypes about the group (Ryan 1991). One of the reasons why Martin Luther King was such an effective leader was that his promotion of the use of nonviolent tactics ensured that media stories about civil rights movements created the visual of peaceful black protesters being violently attacked by whites (McAdam 1996).

(17.) There is also selection (e.g., Myers and Caniglia 2004) (whether an event gets covered or not) and there is content (this would encompass the various studies on description bias) (e.g., Smith et al. 2001; Snow, Vliegenthart, and Corrigall-Brown 2007) and resonance and legitimacy (e.g., Koopmans and Olzak 2004; Snow and Corrigall-Brown 2004).

(18.) While the convention in the literature is to consider" the day the event took place this conceptualization only works for single day events.

(19.) A literature is a table with pamphlets and other information about an issue.

(20.) A Vancouver Sun article published about the incident clearly accepted Chief Stevenson's articulation of the parallel between the First Nations in Canada and blacks in South Africa: "The Peguis houses, scattered around the reserve, were a strong reminder of the circumstances in the (South African) townships and homelands, ranging from rare, nicer homes to abject poverty. Same props, it seemed, different case" ("Babb Visit to Reserve Gets Big Play." Vancouver Sun March 11, 1987:B1).

(21.) Thus, while some, such as Traugott (1993), suggest that a given tactic, such as barricade, can be treated as a repertoire (e.g., its form and aim shifted over time) the case of Oka suggests that this change can occur even over the course of a single event.

(22.) We justify our focus on newspapers (as opposed to other media such as the Internet and television) on the following grounds: first, while there is clearly a growing importance of the Internet as a news source many of the events that scholars of collective action study (including the ones considered here) occurred before it came into widespread use. While we can not discount the importance of television, it covers fewer events than newspapers (McCarthy et ah 1996). Furthermore, the costs and resources associated with analyzing the packaging of multiple television news segments for the same number of events would be prohibitive in comparison with newspapers.

(23.) The Toronto Star has the highest circulation.

(24.) Because our data come from two newspapers, two types of bias are at issue. Newspapers do not cover all events that actually take place and the coverage rates vary across papers. With respect to the first issue, we offer the qualifying statement that our results refer to coverage in the press as opposed to the entirety of all political events during the period. In order to assess the extent of missed coverage that resulted from the use of our two newspapers and not others, we collected additional information from the other papers available in Canadian newsstand on all other events during the period. We found that an additional 59 events were covered by other Canadian newspapers but not by the Globe or the Sun. Most of these events were local and covered in a single article. Unfortunately, we can not include these events in our final data set because it would mean collecting hundreds (or more) additional newspapers articles on all the events already in out" data set.

(25.) Most images were photographs, although they could also be a map or cartoon.

(26.) Other packaging elements that we do not examine include headline size and placement of the article on its page.

(27.) The 10 reporters are Stewart Bell, Mike Crawley, 'Ferry Glavin, Ross Howard, Mark Hume, Rudy Platiel, Patricia Poirier, Scott Simpson, Deborah Wilson, and Geoffrey York.

(28.) While documenting these characteristics, we found that in some cases information about the details of specific events was missing. This was most common for the longer events and events located in more isolated areas. In some cases, an article might state that a protest started last week but not give the exact date, and in other cases, a series of articles in one paper might mention that an event had begun but then provide no record of the event concluding. Therefore, in order to have as comprehensive a set of event characteristics as possible, we used information provided by both papers as well as by other newspapers and sources not examined in the present study to fill in this missing data.

(29.) The size figures were missing for 54 events (137 articles). Because most of these appeared to be smaller events, deleting them from the analysis would have skewed our results. Thus, we imputed a size of 10 for each of these events. We then created a dummy variable indicating this imputation and ran all models with the dummy included (effect of size did not change with the dummy measure). However, we ultimately dropped the dummy from the analysis because it was confounded with the violence measure (e.g., no events that were missing on size were violent).

(30.) There was one event (a blockade by the Toosey First Nation) where the timing was missing because the exact start of the event could not be identified. This event (and two associated articles) was dropped from our list of cases.

(31.) Fish/log/hunt-ins are events where people fish. cut down trees, or hunt in defiance of federal regulations as a means of protest or civil disobedience. The purpose of these events, on top of the direct use of natural resources, is to assert treaty rights. In some cases, these events were staged specifically to draw attention to the issue or they were general resistance to outside control.

(32.) While standoffs have many of the features of critical encounters (Staggenborg 1993) in that they involve direct confrontation with authorities, based on our knowledge of these cases, only Oka can be linked to any kind of real increase in the rate of collective action (e.g., see Ramos 2006). For this reason, and because the effect of the event (which we are measuring) can not tell us about the event itself, we do not use this term for these events.

(33.) This was also an issue for other measures, such as urban location or type of mobilizing group, that we might have liked to examine but ultimately can not include in our models.

(34.) Oka initially started out as a blockade. However, because the majority of the event took the form of land occupation we coded it as the latter.

(35.) These alternative ways included the presence and level of violence. However, compared with many other cases of contentious action, these events are extremely peaceful. When violence of any form has been used, it has only taken place in response to attacks by officials on Native lands (Alfred and Lowe 2005). Over the 10-year period there were two deaths, one police officer and one protester (Oka and Ipperwash). There were also a number of events, which did not appear to involve much violence but where collective actors were described as wearing military fatigues. While the wearing of clothing is not itself an act of violence, fatigues have military symbolism and hence communicate a message, symbolic or otherwise, about a willingness and/or ability to use violence. The wearing of fatigues also communicates a political meaning about nationhood status. Nevertheless the wearing of fatigues was also only associated with certain types of tactics.

(36.) We also estimated models for the length of the article (coded as the number of words and alternatively as a 0/1 measure indicated whether the article was above or below average in length). However, because the results were not robust to different specifications (Young 2009) we do not present them.

(37.) For example, one event can have multiple front-page outcomes (some articles are on the front page and some are not). The same event characteristics will count for each outcome and therefore there may be within-event error that needs to be taken into account.

(38.) This average is quite long due to the presence of several very long events. Half of the events were one-day events.

(39.) The length of the event can not be used to predict coverage at the article level because the end of some events occurs after any given article might have been published. We use this new measure in this and all subsequent tables. Fer front-page coverage the cross-classified cell counts for beat reporter and Canadian Press were too small to include these variables in the models. We do, however, include them in all subsequent tables.

(40.) Because of confounding with other variables, we could not include the measures for Canadian Press and beat reporter in the models for front-page coverage.

(41.) Our use of the term disruption contrasts with McAdam and Su (2002) who define it in terms of violence or property damage.

(42.) We use these examples not to criticize the particular studies but rather to illustrate how widespread the practice is (including by first author in a previous study).

(43.) For a notable exception see Koopmans and Olzak (2004) who model media visibility of right-wing violence in terms of percentage of front-page coverage that it receives (and find that it matters).

(44) Some of the measures had very small ceil counts (e.g., no Canadian Press articles written by a beat reporter were published on a Monday). In order to ensure the stability of the model estimates we adopted Oliver and Maney's (2000) approach and either collapsed some of the categories on our independent measures or left the measure out of the model.

(45) There are many modeling techniques that can be used for count data. Because our data are over-dispersed negative binomial is most appropriate (see Allison 1999).

Rima Wilkes, Department of Sociology, University of British Columbia, 6303 NW Marine Drive, Vancouver, BC, Canada V6T 1Z1. E-mail: wilkesr@interchange.ubc.ca Table 1 Variations in the Kinds of Media Packaging Events Received All 230 72 events in events in Globe and all papers Mail only N/mean % N/mean % Total number of articles 1,801 104 Number of front-page articles 317 17.60 13 12.50 Number of words (mean) per 554.13 427.03 article Number of articles with image 523 29.04 10 9.62 included Number of articles per event 7.87 1.61 (mean) 64 events in 94 events Vancouver in both Sun only papers N/mean % N/mean % Total number of articles 159 1,538 Number of front-page articles 13 8.18 291 18.92 Number of words (mean) per 403.20 578.32 article Number of articles with image 48 30.19 465 30.23 included Number of articles per event 2.21 16.45 (mean) Table 2 Differences in Event Characteristics across Papers All 230 events in all papers Number of events (N) % Violent 24 10.43 Size (mean) 119.67 Length of event (mean) (a) 27.13 Tactic March/demonstration 55 23.91 Road blockade/closure 89 38.70 Rail blockade 18 7.83 Building occupation 11 4.78 Land occupation/reclamation 23 10.00 Fish-in/log-in/hunt-in 12 5.22 Other 22 9.57 Location British Columbia 97 42.17 Alberta 6 2.61 Saskatchewan 3 1.30 Manitoba 15 6.52 Ontario 55 23.91 Quebec 25 10.87 Atlantic 29 12.61 Type of group First Nation/band 145 63.04 Multiple FNs 45 19.57 Urban/supra-tribal 40 17.39 Outside support 88 38.26 Years 1985 9 3.91 1986 8 3.48 1987 12 5.22 1988 30 13.04 1989 40 17.39 1990 60 26.09 1991 16 6.96 1992 9 3.91 1993 11 4.78 1994 14 6.09 1995 21 9.13 Total number of events 230 72 events covered by Globe and Mail only Number of events (N) % Violent 5 6.94 Size (mean) 58.83 Length of event (mean) (a) 12.54 Tactic March/demonstration 20 27.78 Road blockade/closure 29 40.28 Rail blockade 5 6.94 Building occupation 3 4.17 Land occupation/reclamation 4 5.56 Fish-in/log-in/hunt-in 7 9.72 Other 4 5.56 Location British Columbia 54 75.00 Alberta 2 2.78 Saskatchewan 1 1.39 Manitoba 3 4.17 Ontario 5 6.94 Quebec 3 4.17 Atlantic 4 5.56 Type of group First Nation/band 46 63.89 Multiple FNs 12 16.67 Urban/supra-tribal 14 19.44 Outside support 20 27.78 Years 1985 2 2.78 1986 1 1.39 1987 2 2.78 1988 18 25.00 1989 10 13.89 1990 21 29.17 1991 4 5.56 1992 0 0.00 1993 5 6.94 1994 4 5.56 1995 5 6.94 Total number of events 72 64 events covered by Vancouver Sun only Number of events (N) % Violent 2 3.13 Size (mean) 193.05 Length of event (mean) (a) 25.45 Tactic March/demonstration 24 37.50 Road blockade/closure 23 35.94 Rail blockade 0 0.00 Building occupation 2 3.13 Land occupation/reclamation 9 14.06 Fish-in/log-in/hunt-in 2 3.13 Other 4 6.25 Location British Columbia 2 3.13 Alberta 1 1.56 Saskatchewan 0 0.00 Manitoba 4 6.25 Ontario 34 53.13 Quebec 10 15.63 Atlantic 13 20.31 Type of group First Nation/band 34 53.13 Multiple FNs 15 23.44 Urban/supra-tribal 15 23.44 Outside support 24 37.50 Years 1985 3 4.69 1986 5 7.81 1987 3 4.69 1988 3 4.69 1989 20 31.25 1990 13 20.31 1991 3 4.69 1992 4 6.25 1993 2 3.13 1994 4 6.25 1995 4 6.25 Total number of events 64 94 events covered by both papers Number of events (N) % Violent 17 18.09 Size (mean) 116.30 Length of event (mean) (a) 39.34 Tactic March/demonstration 11 11.70 Road blockade/closure 37 39.36 Rail blockade 13 13.83 Building occupation 6 6.38 Land occupation/reclamation 10 10.64 Fish-in/log-in/hunt-in 3 3.19 Other 14 14.89 Location British Columbia 41 43.62 Alberta 3 3.19 Saskatchewan 2 2.13 Manitoba 8 8.51 Ontario 16 17.02 Quebec 12 12.77 Atlantic 12 12.77 Type of group First Nation/band 65 69.15 Multiple FNs 18 19.15 Urban/supra-tribal 11 11.70 Outside support 44 46.81 Years 1985 4 4.26 1986 2 2.13 1987 7 7.45 1988 5 5.32 1989 10 10.64 1990 30 31.91 1991 9 9.57 1992 5 5.32 1993 4 4.26 1994 6 6.38 1995 12 12.77 Total number of events 94 (a) N is 71, 60, 93, 224. Table 3 Event Coverage Rates Articles Number of Number of % of per event events % of events articles coverage 1 99 42.61 99 5.49 2 42 18.26 84 4.66 3 15 6.52 45 2.50 4 7 3.04 28 1.55 5 9 3.91 45 2.50 6 7 3.48 42 2.33 7 2 0.87 14 0.78 8 8 3.48 64 3.55 9 4 1.74 36 2.00 10 7 3.04 70 3.89 11 6 2.61 66 3.66 12 3 1.30 36 2.00 13 3 1.30 39 2.17 14 1 0.43 14 0.78 15 2 0.87 30 1.67 19 1 0.43 19 1.05 20 1 0.43 20 1.11 21 1 0.43 21 1.17 24 2 0.87 48 2.67 34 2 0.87 68 3.78 35 1 0.43 35 1.94 36 1 0.43 36 2.00 38 1 0.43 38 2.11 62 1 0.43 62 3.44 65 1 0.43 65 3.61 77 1 0.43 77 4.28 162 1 0.43 162 9.00 438 1 0.43 438 24.32 230 100 1,801 100 Table 4 Distribution of Event Packaging Number of articles N % Mean News routines Monday 263 14.6 Newswire 519 28.8 Beat reporter 368 20.4 Globe and Mail 840 46.6 Event characteristics Tactics (standoffs excluded) March 77 4.3 1.4 Road blockade 624 34.6 7.1 Rail blockade 89 4.9 5.2 Land occupation 76 4.2 3.8 Other tactic 224 12.4 5.0 Total all tactics 1,090 60.5 4.8 Standoffs Kanesatake--The Pines--Oka 438 24.3 Khanawake--Mercier Bridge--Oka 77 4.3 Kettle and Stoney Point/Ipperwash 34 1.9 Gustafsen Lake 162 9.0 Standoffs (total) 711 39.5 178 All events (total) 1,801 7.90 Front page Length of article N % Mean News routines Monday 54 17.0 538 Newswire 48 15.1 350 Beat reporter 104 32.8 697 Globe and Mail 156 49.2 609 Event characteristics Tactics (standoffs excluded) March 3 0.9 401 Road blockade 92 29.0 504 Rail blockade 23 7.3 606 Land occupation 4 1.3 331 Other tactic 24 7.6 426 Total all tactics 146 46.1 477 Standoffs Kanesatake--The Pines--Oka 96 30.3 646 Khanawake--Mercier Bridge--Oka 32 10.1 760 Kettle and Stoney Point/Ipperwash 4 1.3 590 Gustafsen Lake 39 12.3 721 Standoffs (total) 171 53.9 673 All events (total) 317 17.6 554 Picture N % News routines Monday 53 10.1 Newswire 114 21.8 Beat reporter 163 31.2 Globe and Mail 205 39.2 Event characteristics Tactics (standoffs excluded) March 22 4.2 Road blockade 155 29.6 Rail blockade 21 4.0 Land occupation 6 1.1 Other tactic 58 11.1 Total all tactics 262 50.1 Standoffs Kanesatake--The Pines--Oka 151 28.9 Khanawake--Mercier Bridge--Oka 27 5.2 Kettle and Stoney Point/Ipperwash 7 1.3 Gustafsen Lake 76 14.5 Standoffs (total) 261 49.9 All events (total) 523 29.0 Table 5 GEE Regression of Factors Associated with Differential Media Packaging across Events Number of articles 1a 1b Intercept -0.96 ** -0.97 ** (0.33) (0.32) Monday Canadian Press Beat reporter Globe and Mail Timing of article (days since event start) Event duration 0.01 0.02 (0.06) (0.05) Number of participants 0.01 *** 0.005 (a) (0.003) (0.002) Tactic (versus march) Road blockade 2.43 *** 2.55 *** (0.38) (0.36) Rail blockade 2.32 *** 2.36 *** (0.51) (0.48) Land occupation 3.18 *** 1.94 *** (0.54) (0.48) Other tactic 1.99 *** 2.17 *** (0.41) (0.39) Fit statistics BIC 1,049 * 973 * BIC (intercept only) 1,128 1,015 QIC QIC (intercept only) N 224 220 Front page 2a 2b Intercept -3.23 *** -3.19 *** (0.58) (0.59) Monday 0.32 * 0.29 (0.14) (0.21) Canadian Press Beat reporter Globe and Mail 0.10 -0.14 (0.28) (0.29) Timing of article (days -0.05 -0.01 since event start) (0.07) (0.08) Event duration Number of participants 0.01 0.01 (0.06) (0.05) Tactic (versus march) Road blockade 1.44 * 1.38 * (0.59) (0.60) Rail blockade 2.07 ** 2.11 ** (0.64) (0.65) Land occupation 1.08 0.35 (0.69) (0.81) Other tactic 0.98 1.04 (0.63) (0.64) Fit statistics BIC BIC (intercept only) QIC 1,773 856 (a) QIC (intercept only) 1,705 863 N 1,801 1,090 Picture 3a 3b Intercept -0.12 -0.23 (0.35) (0.34) Monday -0.52 -0.58 (0.22) (0.25) Canadian Press -0.49 -0.41 (0.30) (0.23) Beat reporter 1.02 *** 0.90 *** (0.16) (0.21) Globe and Mail -0.81 *** -0.92 *** (0.17) (0.16) Timing of article (days - 0.26 -0.13 since event start) (0.11) (0.08) Event duration Number of participants 0.01 0.02 (0.05) (0.04) Tactic (versus march) Road blockade -0.76 * -0.63 (0.36) (0.35) Rail blockade -0.99 * -0.75 (0.50) (0.48) Land occupation -0.47 -1.56 ** (0.64) (0.47) Other tactic -0.54 -0.39 (0.40) (0.37) Fit statistics BIC BIC (intercept only) QIC 2,405 1,150 * QIC (intercept only) 2,200 1,206 N 1,801 1,090 * p < .05. ** p < .01. *** p < .001. Models 1a and 1b are event based and the remaining models are article based. Models a standoffs, models b with four standoffs omitted (removal of one road blockade and three land occupations). (a) Model improves overall fit if nonsignificant predictors are removed. The QIC statistic is used to compare model fit for GEE models. GEE. generalized estimating equations. QIC, Quasi Likelihood Independence Criterion. BIC, Bayesian Information Criterion.