Electronic petitioning as online collective action: Examining the e-petitioning behavior of an extremist group in we the people

Abstract

This study aims to reveal patterns of e-petition co-signing behavior that are indicative of the mobilization of online “communities” engaging in collective action to express policy preferences on We the People (WtP), the first web-enabled US government petitioning system initiated by Obama. This Internet-based tool allowed users to petition the Obama Administration and solicit support for policy suggestions. Using petition data from WtP, this case study examines a set of 125 petitions that were created by individuals that are associated with a white supremacist group called The White Genocide Project (The White Genocide Project has recently changed their name to Fight White Genocide). Using data mining techniques, namely market basket analysis and social network analysis, I found evidence of the mobilization of “communities” of an extremist group of white supremacists who systematically and strategically used the WtP platform to broadcast their message by creating and co-signing petitions every month for almost four years.

Keywords

Electronic petitioning online collective action market basket analysis social network analysis policy suggestions we the people social media

1. Introduction

In September 2011, President Obama’s administration initiated the first web-enabled petition system for the US federal government called We the People (WtP) (https://petitions.whitehouse.gov/) as part of the Open Government Partnership (OGP) launched in July 2011. Obama’s first National Action Plan for the OGP was released in September 2011 featuring WtP as an innovative tool for citizen engagement (Sinai & Smith, 2013). WtP gives individuals the opportunity to petition the US federal government to undertake actions suggested by the petitioner and to register signatures from supporters. New users must create an account with an email address, first name, last name, and zip code ( WtP stopped asking for a zip code from the user in April 2017).

When a petition garnered at least 100,000 signatures within 30 days it became eligible to receive a response from the Obama Administration. Under Obama, usage of WtP had become increasingly popular attracting a growing amount of petitioning activity. As of November 2016, WtP had generated over 28 million users, more than 475,000 petitions, and 38.5 million signatures (Pew, 2016).

However, when President Trump took office on January 20, 2017, WtP became part of the new presidential administration. All of the petitions that were created under the Obama administration from September 22, 2011 to January 20, 2017 have been archived. WtP continued to function the same way it did before Trump took it over, with a few caveats. Petitions that received 100,000 signatures in 30 days (which is the threshold for a response) were not being responded to. Also, petitions remained on the site for the public to see and sign even after the 30-day window regardless of whether they reached the 100,000 signature threshold. Under the Obama administration, petitions would remain on the site for 30 days and then they would be retired.

In December 2017, Trump said that they would create a new platform that would save the taxpayers more than $1m a year (Zurcher, 2017). As of January 2021, WtP is operating as it was when Trump took it over in January 2017. People are still creating and signing petitions and none of these petitions are getting responses by the Trump Administration. Clearly, WtP was never a priority for the Trump Administration throughout his term as the President of the United States (at the time of writing of this article, Trump lost the race for a second term as President to Senator Joe Biden).

In the case study that follows, I explore how electronic petitioning functioned as online collective action in mobilizing support for policy suggestions to the US government by an organized group of white supremacists.

2. Background

2.1 E-petitioning systems

Electronic petitioning systems have surfaced as a contemporary and potentially effective way for citizens to communicate with their governments about policy issues and have facilitated making public participation in policy discussions more easily accessible.

However, citizens’ rights to petition their governments are not new. The act of petitioning dates back to the 13th century and has long since been a way for people to communicate with local, national or parliamentary governments. With the advent of the Internet and advances in Information Communication Technologies (ICTs), e-petitioning has emerged as a mechanism for citizens to participate in the policy-making process. Germany’s Bundestag’s e-petition system has demonstrated that these systems are at the “forefront of official, fully operational e-participation opportunities provided to citizens by governments and parliaments” (Lindner & Riehm, 2011, p. 1). The purpose of petitioning is to change public policy, demand officials to make statements, or induce public institutions to take action (Lindner & Riehm, 2011). E-petitioning provides a safe “playing field” for citizens to take part in the policy-making process and is well-suited for a representative democratic society (Lindner & Riehm, 2009; Bochel, 2012).

E-petitioning systems are typically designed with the purpose of enabling citizens to have some influence over decision-making in the policy process. Several studies of e-petitioning systems have examined whether such influence actually takes place, with mixed results. Macintosh et al. (2008) found the Public Petitions Committee in the Scottish Parliament system “useful in influencing politicians about issues considered important” (p. 10). The Scottish Parliament was also one of the two successful e-petitioning systems in Bochel’s (2012) study (the other being the Welsh Assembly). In other systems (the House of Commons and the Coalition in Scotland), there was evidence that the petitions had no effect at all on actual outcomes involving policy-making decisions (Bochel, 2012). In the case of the Royal Borough of Kingston, one of the first local authorities to implement e-petitioning in 2004, findings showed that citizen engagement had an impact on policy making decisions (Panagiotopoulos et al., 2010).

In the US system, a few successful e-petitions have been followed by actions that are consistent with those requested by the petitions, some with attribution to the petition. For example, the petition “Make Unlocking Cell Phones Legal.” was created on January 24, 2013 in response to the decision of the Library of Congress to remove cell phones from the exceptions to the Digital Millennium Copyright Act (DMCA). The petition garnered 114,000 signatures within the 30-day threshold needed for a response from the Obama administration. The response included a plan of action that involved working with the FCC, wireless companies, and Congress to make unlocking cell phones legal. According to Ezra Mechaber (2015), Deputy Director of Email and Petitions at the time in the White House Office of Digital Strategy, the popularity of the e-petition (through signature accumulation), the concerted efforts of the White House policy team, FCC, and the wireless companies and the national attention by multiple media outlets encouraged both the House of Representatives and the Senate to pass legislation to legalize cell phone unlocking in March 2014. It was the first time an e-petition on the US platform apparently led to legislative change.

2.2 Online collective action

Collective action found a new vehicle online with the advent of the Internet. One of the earliest examples of online collective action and online protest using the Internet as a tool for organizing mobilization is the case of the Lotus Marketplace in 1990 (Gurak, 2003). Email and Usenet newsgroups (two of the earliest forms of online communication tools) were used to protest a compact disc (CD) called Marketplace: Households that Lotus Development Corporation revealed had the names, addresses, and expenditure behaviors of 120 million people from different households in the US (Gurak, 2003). The CD was intended to be sold to people who wanted to do direct marketing from home and who did not want to pay the high costs that direct market mailing companies were charging (Gurak, 2003). One of the more influential activists/protester on the email and Usenet newsgroups, Larry Seiler, sent a message to the email and Usenet newsgroups explaining that they should be worried about the fact that Lotus was selling their personal information to marketers (Gurak, 2003).

Seiler sent another message and over 30,000 people, who were on the email and Usenet newsgroups, contacted Lotus asking to have their names removed from their database (Gurak, 2003). Shortly after, Lotus responded by cancelling the release of the CD Marketplace: Households because of “public concerns and misunderstandings of the product, and the substantial, unexpected costs required to fully address consumer privacy issues” (Gurak, 1997, p. 66).

In 2011, several protests around the world surfaced beginning with rebellion in Tunisia and uprisings in the Middle East and North Africa (MENA), and the revolutions in dictatorship countries Egypt and Libya which became known as the Arab Spring (González-Bailón et al., 2013). Even democratic countries such as Spain, Greece, Chile, and the US experienced demonstrations against social injustice which collectively became the global Occupy movement in which thousands of people took to the streets in cities all over the world chanting the mantra “We are the 99%” (González-Bailón et al., 2013). In these protests, different strategies were used to mobilize people to participate in the movements, such as living in tents in public places and establishing local media facilities to enable the spread of the protest information via online networks, such as Twitter and Facebook (González-Bailón et al., 2013).

González-Bailón et al. (2013) conducted an empirical study of the information diffusion in the protest in Spain in May 2011, the Spanish indignados movement, using Twitter data with a focus on how new information and communication technologies decrease the costs of organizing and managing collective action. This movement provided an illustration of how the spread of information online frequently goes together with/complements or may incite the spread of offline behaviors (González-Bailón et al., 2013). Tens of thousands of people were mobilized online and organized to set up tents in public spaces in massive offline protest (González-Bailón et al., 2013). González-Bailón et al. argue that analyzing protest activity via the “lens of diffusion” ties collective action to “studies of social influence and interdependence in networks which are … better equipped to make sense of digital protests than classic approaches to the logic of collective action” (2013, p. 945).

Online networks help spread participation in collective action by allowing people to see who else has participated (González-Bailón et al., 2013) which may contribute to their decision to participate. There are different thresholds for individuals in regards to how many of the people in their networks need to participate before they decide to do so as well (Granovetter, 1978; Valente, 1996). Research studying the effects of social influence in online networks posits collective action as “more as a process of contagion than of incentive design” (González-Bailón et al., 2013, p. 947) which “makes collective action similar to other diffusion processes” (González-Bailón et al., 2013, p. 947). Their analyses shows that the development of “digitally born protests” is contingent on the tactical utilization of groups of people who are already connected and being able to take advantage of the prominent actors in the networks who are highly-connected (González-Bailón et al., 2013, p. 945).

2.3 E-petitioning and collective action

Bimber et al. state that there are three basic functions that are essential to all collective action: “(a) a means of identifying people with relevant, potential interests in the public good; (b) a means of communication messages commonly perceivable among them; and (c) a means of coordination, integrating, or synchronizing their contributions” (2005, p. 374). The costs of organizing individuals for collective action have been greatly reduced with the online information and communication technologies. The need for formal organization in collective action is not as necessary as it once was before the Internet especially when there is a network already in place using established online communication channels to pass along information instantly. As we saw in the case of the Lotus Marketplace, a large number of people, namely, 30,000, were mobilized to take action after reading a message that was sent to them online (by already established email or Usenet newsgroup networks). Gurak acknowledged that the people using these information and communication technologies to protest Lotus Marketplace were like-minded in that they were all computer savvy users and/or techies which, at this time, were a minority (2003). There was a sense of trust in the Internet protest network because the people writing and forwarding the messages were all part of a “common core of like-minded computer privacy advocates” who shared “common values about what “cyberspace” should be” (Gurak, 2003, p. 8).

Previous research on the US e-petitioning platform, We the People, found evidence that people signed more than one petition related to gun control laws after the Sandy Hook school shooting on December 14, 2012. During the week (December 14–21) after the fatal shooting of 20 children and 6 school personnel by a lone gunman, 33 petitions were created. Of these, 12 were in favor of gun control laws and the other 21 were against gun control laws or offered alternative policy suggestions such as improvements in mental health, and putting armed guards in schools. Using market basket analysis and social network analysis, Dumas et al. (2015a: 2015b) analyzed the extent to which individuals who signed one petition in favor of a particular policy option, for example mental health care reform, also signed other petitions favoring the same policy suggestion. Dumas et al. (2015a: 2015b) found substantial patterns of co-signing of petitions proposing similar policy options, indicating that signers recognized similarities in the policy positions expressed, and endorsed multiple petitions with thematic similarities. These patterns of co-signing were also reflected in our network analyses of communities, which supported specific policy proposals that conformed to a particular theme. This study concluded that the signing of petitions related to Sandy Hook was used strategically by individuals to express opinions about and influence gun control policy in the future (Dumas et al., 2015a: 2015b). The results of these studies support claims that e-petition data can provide information about how collective political action can take place on e-petitioning platforms.

2.4 Social media promotion

WtP was created by the Obama Administration to be an innovative tool for online citizen engagement. The system was designed to make it easy for people to bring issues to the forefront of the policy making agenda and potentially have an effect on the policy making process. Users are encouraged to garner support for their petitions through social media via the “Promote this Petition” buttons for Twitter and Facebook on the petition pages. The initial petition creators spread the petition information to their personal online social networks which may or may not be shared with other online social networks of the individuals who also sign the petition; thus, resulting in a snowball effect.

Promoting e-petitions through online social networks is one way to gather support in the form of signatures. Research has shown that most people will not actively seek out e-petitions to sign; they are more apt to sign if they see e-petition information on their social networking sites when it initially appears (Gleason et al., 2013; Lin, 2013; Yasseri et al., 2014). Thus, an important design feature for platforms is to consider how to make it easy for e-petition creators and signers to disseminate links to the e-petitions via online social network sites, blogs, discussion forums, and email. By providing users with the ability to diffuse e-petition information online, the platform and its e-petitions will receive more exposure to the public and may result in more participation. Additionally, issues will get more attention from a larger population more quickly.

To better understand the social dynamics of e-petitioning behavior, researchers can examine the diffusion of e-petition information in online social networks. E-petition initiators have the opportunity to immediately spread information about their e-petition to their related social networks when they are given the option to use platform-provided access to Twitter and Facebook. Groups of people actively seeking support for an issue or cause and coming together for a common goal frequently engage in online collective action. E-petition signing patterns involving social networks can reveal one important form of online collective action. This information may be used by policy makers to understand which actors are influential in mobilizing online support for certain policy issues that are salient in the public sphere. Dumas et al.’s (2015a; 2015b) study of e-petitioning on We the People following the Sandy Hook tragedy used centrality measures and community detection algorithms to find core groups of people who were mobilizing for and against gun control laws. These studies provide evidence that individuals used the US platform to register their support for certain specific petitions that offered a particular policy option.

The Internet’s role in collective action within social media is apparent in enabling massive amounts of people, strangers in some cases, to come together to mobilize or participate online through petitioning or calling attention to issues via retweeting or hitting the “Like” button on Facebook (Margetts et al., 2013). Do people search for opportunities to mobilize or do they respond to messages virally spread through their online social networks? Social influence plays a role in signature accumulation and collective action (Margetts et al., 2016). People have varying thresholds for joining a mobilization, with some willing to join when numbers of other participants are very low (the leaders), while others will only join when there are large numbers already participating (the followers) (Gonzalez-Baillon et al., 2011). The diffusion of petition information from the e-petitioning platforms to individual’s personal online social networks needs to be explored further.

3. Case study: The white genocide petitions

The case study that follows focuses on 125 petitions which upon further investigation of the petition texts (title and body), appear to be created on WtP by individuals who are associated with a national white supremacist group called the White Genocide Project. Figure 1 is a screenshot of the White Genocide Project’s website where they announce that their messages will be displayed every month at the White House, stating that “On the FIRST of every month, we put STOP WHITE GENOCIDE messages like the ones below on public display at the White House”. There is also a link to this page on the lower right sidebar called “Petitions” – Monthly Messages displayed at the White House (see bottom right in Fig. 1). On the lower left side, there is a link that takes you to a page (See Fig. 1) with links and texts to all of the petitions that they have created which have been archived.

Figure 1.

Screenshot of the white genocide project website http://whitegenocideproject.com/petitions/.

Interestingly, on another page on the White Genocide Project website, there is a post announcing the end of their project to mobilize support for their cause on WtP in anticipation of the end of Obama’s e-petitioning platform that reads:

The October, 2016 White House meme-message below marks exactly 4 years that we’ve displayed Stop White Genocide memes every month on the website of the President of the United States! With President Obama’s term now coming to an end, this project now also comes to an end, and we thank heartily all of you who have helped!

Some of the text from the petition titles and bodies include phrases from “The Mantra”, which is a strategy to fight White Genocide, created by Bob Whitaker, a recent presidential candidate. The Mantra is displayed on Bob Whitaker’s website.

In January 2015, the White Genocide Project also claimed responsibility for posting similar anti-diversity messages on a billboard they paid for along Alabama highway I-59 (Terry, 2015). The sign read “Diversity Means Chasing Down the Last White Person” with “#whitegenocide on the bottom (Terry, 2015). The billboard can be viewed in a NBC News video.

The billboard was taken down five days after it was put up (Terry, 2015). In 2017, two years later, also in January, signs reading “Diversity is a code word for #WHITEGENOCIDE” were found outside two schools, a middle school and an elementary school, in Manchester, New Hampshire (Terry, 2015). Investigators traced the messages on the signs back to the White Genocide Project’s website where there was evidence that members of the White Genocide group were thanking the people who put the signs up (Terry, 2015).

So, what is the message that these individuals are broadcasting? According to the website belonging to the creators of these petitions, “White Genocide” is a result of mixing other races with the white race. It appears that this is a group of white supremacists that have decided to use WtP to try to mobilize support in a strategic and systematic fashion by creating a number of petitions every month that contain similar, if not the exact, text.

In this study, techniques from market basket analysis are used to explore questions about whether individuals who signed one white genocide petition also signed other white genocide petitions. Methods from social network analysis are used to determine if there are groups of individuals who signed the white genocide petitions, thus suggesting the creation of “communities” if individuals whose actions were similarly aligned in support of stopping white genocide.

The analysis that follows considers the set of 125 petitions that were created and signed on a monthly basis for four years that called for stopping white genocide as a policy suggestion to the US government. I was explicitly interested in exploring answers to the following research questions:

RQ1. Do the same people sign petitions related to a particular policy suggestion across a longer time period, across different months? Do people sign petitions and return to the site to sign other petitions related to a particular policy suggestion? RQ2. Is there evidence of e-petitioners forming core groups or “communities” that are characterized by a particular policy suggestion?

3.1 Data description

The petition data that is being used for this case study were obtained from a publicly available White House database containing information about all petitions and signatures (coded to ensure anonymity) appearing on the WtP website between Sept 22, 2011 and April 20, 2016. Petition titles and petition signatures are used in the analyses that follow. Within each dataset, a distinct signature ID consists of unique first and last initials followed by a five-digit zip code. Any ID that does not include a valid five-digit zip code has been eliminated.

The dataset for the white genocide petitions case study consists of petition data and signature data for 132 petitions created between on WtP between November 18, 2012 and March 30, 2016 that contain the phrase “white genocide” and that successfully achieved 150 signatures within 30 days, making them available for the public to see and sign (see Fig. 2). There are 47,059 total signatures with 31,520 that have valid zip codes. ( I acknowledge the possibility that a distinct ID of two initials and a zip code may reference more than one individual). There are 12,881 unique signature identification codes.

Figure 2.

One hundred and thirty-two white genocide petitions with signatures over time.

After reading the titles and the bodies of the 132 petitions, I found a set of 125 petitions which expressed opinions against what the petition creators are calling “White Genocide”. This set of 125 petitions will be referred to as Stop White Genocide (SWG) petitions. The other seven petitions appear to be created in response to one or more of the SWG petitions and will be referred to as Anti-SWG.

There are a total of 11,979 unique signature identifications in the 125 SWG petitions and 1212 unique signature identifications in the seven Anti-SWG petitions. There is an overlap of 255 unique signature identifications that signed both SWG and Anti-SWG petitions.

Table 1 includes six SWG petitions that were created in the month of February, 2013. These are examples of the petitions that were created monthly for almost four years. Petitions need to obtain 150 signatures in 30 days to be put on the website for the public to sign. Over the course of four years there were only four months where an SWG petition did not make it to the website. Table 1 is an example of a set of six petitions that were created during the month of February 2013. As Table 1 indicates, these petitions did not accumulate a large number of signatures. Additionally, these six SWG petitions were created on the same day within minutes of each other and by the same user, AD94705.

Table 1

Example of SWG petitions for the month of February 2013

ID	Creator	Title	Date created	# signatures
7	AD94705	(Feb. #1 of 6) STOP WHITE GENOCIDE! Halt (Feb. #1 of 6) STOP WHITE GENOCIDE! Halt MASSIVE third world immigration and FORCED assimilation in White countries! MASSIVE third world immigration and FORCED assimilation in White countries!	2/8/2013 2:00:00	405
8	AD94705	(Feb. #2 of 6) Provide Federal Funds for the Million Name March on Washington, D.C. to STOP WHITE GENOCIDE!	2/8/2013 2:07:00	360
9	AD94705	(Feb. #3 of 6) Teach public school children the truth: ANTI-RACIST IS A CODE WORD FOR ANTI-WHITE!	2/8/2013 2:19:00	450
10	AD94705	(Feb. #4 out of 6) ’äÎ $\raise 2.15pt\hbox{$\scriptstyle 3$}\kern-1.0pt/\kern-1.5pt\lower 1.075pt\hbox% {$\scriptstyle 3$}$ Let White GeNOcide Project give the “national conversation on race” TWO sides: STOP WHITE GENOCIDE!	2/8/2013 2:24:00	335
11	AD94705	(Feb. #5 of 6) Award the Presidential Medal of Freedom to each Woman in the video “STOP WHITE GENOCIDE”!	2/8/2013 2:33:00	320
12	AD94705	(Feb. #6 of 6) STOP GENOCIDE of WHITES! Invite Public to White House Screenings of “How Whites Took Over America”!	2/8/2013 3:01:00	504

Table 2

Seven anti stop white genocide petitions

ID	Creator	Title	Date created	# signatures	# unique sigs w/valid zips
61	JT (no zip code)	Denounce the “White Man March” as a Hate March	3/3/14 23:19	1088	756
65	JT (no zip code)	Explain to the American public that the term “Anti-White” is a lie.	4/15/14 1:02	357	222
68	TT (no zip code)	Defend Justice Sotomayor against the racist charge that she is “anti-white”!	5/13/14 0:48	624	215
77	TT (no zip code)	Explain to the American public that “ANTI-RACIST IS A CODE WORD FOR ANTI-WHITE” is a white supremacist hate slogan.	7/17/14 20:34	206	103
81	AM (no zip code)	Denounce “AntiRacist Hitler”!	8/17/14 11:13	205	96
86	TT (no zip code)	SUPPORT “our nation’s ‘No Child Left White’ policy”!	10/13/14 1:47	177	35
126	TT (no zip code)	Reject presidential candidate Bob Whitaker’s demand for tax audits of institutions that promote DIVERSITY!	11/3/15 21:55	167	56

There were nine different users who created the 125 SWG petitions. The majority of the 125 SWG petitions were created by two users, AD94705 (80 petitions, 64%) and CM99201 (33 petitions, 26%). AD94705 initiated the first SWG petition on December 13, 2012 and created petitions on a monthly basis until July, 30, 2015 and then stopped. Five months later on January 1, 2015 CM99201 appears to take over and created the monthly petitions for a little over a year ending on February 29, 2016. Of the 12 remaining petitions beyond those created by AD94705 and CM9920, seven other different users created some of the other 125 SWG petitions.

The length of time between when a petition was created and when it became public on the website to be seen and signed varies, but for the most part is short. There were eleven petitions that became public within hours of the time they were created. For the rest of the petitions, the length of time between creation and becoming public ranges from one day to 18 days. The average length is 4.2 days for the 132 White genocide petitions to achieve the 150 signatures needed to become public for people to sign.

As mentioned previously, there were seven petitions in this dataset that appear to have been created in response to one or more of the SWG petitions and will be referred to as Anti-SWG. Table 2 shows the seven Anti-SWG petitions.

Table 3

Seven anti stop white genocide petitions with body

ID/title	Body
61 – Denounce the “White Man March” as a Hate March	A small international group of white supremacists are planning hate actions around the world on the 15 of this month, and are calling their collective action “the White Man March.” They reportedly plan to display banners that say, “Diversity” $=$ White Genocide. This hateful slogan is an insult to the peoples and cultures that our government represents. We believe this small group may be associated with some petitions that have appeared on We the People, since those petitions have favored other slogans of the group: “Asia for the Asians, Africa for the Africans, White countries for EVERYBODY?” “Anti-racist is a code word for anti-white.” We support free speech for everyone, but we also believe our government has the obligation to denounce hate wherever and whenever it appears. Thank you.
65 – Explain to the American public that the term “Anti-White” is a lie.	Some petitions here have called our government “anti-white” for allowing “massive immigration” and for forcing “assimilation” that will “blend whites out of existence” by intermarriage. The petitions call that “white genocide.” This is an attempt by people who enjoy white privilege to provoke a grievance in other people who enjoy white privilege, in order to maintain white privilege! The term “anti-racist” is NOT a “code word” for “anti-white,” as those petitions claim! Nor is diversity a code word for “white genocide,” as they claim. And “Asia for the Asians, Africa for the Africans, White countries for EVERYBODY?” is the natural result of the imperial white countries being in decline. We support free speech, but we also ask our government to always speak up against white privilege!
68 – Defend Justice Sotomayor against the racist charge that she is “anti-white”!	“Asia for the Asians, Africa for the Africans, White countries for EVERYBODY?” “Diversity is a code word for White Genocide.” “Anti-racist is a code word for anti-white.” Those racist hate slogans appear in a petition that asks President Obama to “apologize” for appointing Justice Sotomayor because she is “anti-white.” WHITE SUPREMACISTS USE THE TERM “ANTI-WHITE” AS A CODE WORD FOR “ANTI-RACIST”! Affirmative action rightfully demands that minorities who have suffered centuries of oppression at the hands of the privileged white patriarchy be given a hand up and be moved to the front of the line for jobs and other benefits. Justice Sotomayor supports affirmative action. That does not make her “anti-white.” It makes her anti-racist! “Anti-white” is a code word for anti-racist!
77 – Explain to the American public that “ANTI-RACIST IS A CODE WORD FOR ANTI-WHITE” is a white supremacist hate slogan.	We commend the White House for upholding free speech, but the White House should also speak out against these hate slogans that are now appearing regularly on its petition site and in other places around the country: “Anti-racist is a code word for anti-white” “Asia for the Asians, Africa for the Africans, white countries for EVERYBODY?” “Diversity is a code word for white genocide” These are a blot upon the progress that our country has made toward racial harmony, and they are a rude slap in the face to our many immigrants of color who arrive daily and may see them and feel that we do not welcome them to our shores. They are a display of hate toward our human brothers and sisters that has no place in America in the year 2014. Please state publicly that you do not condone these statements!
81 – Denounce “AntiRacist Hitler”!	A disgusting neo-Nazi hate petition on We the People urges readers to watch the hate video “AntiRacist Hitler.”

Table 3, continued
ID/title	Body
81 – Denounce “AntiRacist Hitler”!	That video presents Hitler as an anti-racist who promotes diversity as an instrument of genocide, an absurd idea that not only expresses hate against immigrants who give strength and diversity to our country, but also expresses hate toward those who suffered a real genocide by a racist hater who was worse than all these racist haters who want to continue making racist hate the foundation of this country.Like other white supremacist hate petitions on this site, that one screams about “White Genocide” and concludes with the absurd hate statements, “anti-racist is a code word for anti-white,” and “diversity is a code word for white genocide.” Mr. President, denounce this hate!
86 – SUPPORT “our nation’s ‘No Child Left White’ policy”!	That may seem like a horrible thing to ask, until you examine the hateful racist petition on this site that asks the opposite. That petition asks to turn immigrants away from our shores, and to end “forced” integration that causes whites to intermarry and “be blended out of existence.” But we know that intermarriage will end the scourge of racism! When we are all non-white, racism will disappear! I’m White and cannot escape my racism, but my grandchildren will not be racist because my grandchildren will be non-white. Diversity is NOT “a code word for white genocide,” as we read every month in these hateful racist petitions that end absurdly in “anti-racist is a code word for anti-white.” Diversity is a plain word for the end of whiteness and other human differences that cause racism!
126 – Reject presidential candidate Bob Whitaker’s demand for tax audits of institutions that promote DIVERSITY!	Diversity is NOT a code word for “white genocide”! Diversity does NOT mean “chasing down white people”! Anti-racist is NOT a “code word” for “anti-white”! But it IS true that “White self-hatred is SICK”! What is more self-hating than fearing new people coming into our country and marrying our children? How little confidence can we have in ourselves and our familes [sic] to fear something so natural? If Mr. Bob Whitaker had a generous heart and a mind for the future, he would appreciate the public funding of groups that promote multiculturalism and diversity, not try to end that funding. It’s from diversity that our future American identity and unity will emerge. We march into the future under the banner “Love conquers all,” NOT under the slogan “Diversity is a code word for White Genocide”!

Table 3 shows the seven Anti-SWG petitions with titles and the bodies of the petitions.

The Anti-SWG petition 61 “Denounce the “White Man March” as a Hate March” created on March 3, 2014 appears to be a reaction to a petition to SWG petition 60 “Please get the administration on message that “Diversity” $=$ White Genocide” created two days prior. The body of petition 60 contained the following text: That message will be seen throughout the United States and the world during the White Man’s March on March 15. It appears that the creator of Anti-SWG petition 61 is trying to call attention to petitions that have been created on WtP by an extremist group and possibly warning that the petition creators were trying to mobilize people to join their cause by participating in a “White Man March”. The first sentence in the body of petition 61 reads: A small international group of white supremacists are planning hate actions around the world on the 15 of this month, and are calling their collective action “the White Man March.” (see Table 3, in red text). Additionally, Anti-SWG petition 77 is calling on the administration to act and “speak out against these hate slogans that are now appearing regularly on its petition site and in other places around the country”.

There is a post about the third Anti-SWG petition 68 “Defend Justice Sotomayor against the racist charge that she is anti-white” on the White Genocide Project’s website. The Anti-SWG petition 68 was created on May 13, 2014 in response to the SWG petition 66 “Apologize for appointing anti-white Justice Sotomayer to the Supreme Court.” which was created on April 30, 2014. The author of the blog post, Henry Davenport, calls it “an attack” on their own petition, urges people in his group to sign it, and provides a link to the petition. Davenport states that he saw it being promoted on Twitter which is where he must have seen it because this post is dated May 14, 2014 and the Anti-SWG petition 68 was created on May 13, 2014 but was not public until May 19, 2014.

There is also another post on the web page that is calling for people to continue to sign any Anti-SWG petition that appears on the site. Recall that I found that 255 people signed both SWG and Anti-SWG petitions. This page also has an archive of all seven Anti-SWG petitions.

These web pages present an appeal for members of the group to sign the Anti-SWG petitions; however, it also demonstrates that this group is committed to getting their ideas out in the public sphere, so much so that they do not even mind if there are petitions against their perspective because they see that as helping to get their issues thrust into the public sphere. This particular campaign is very small; it is however being run by a committed group of activists, and they are engaged in a systematic pattern of behavior. The petition creation and signing behavior in this case is an example of a very systematic and purposeful campaign. This is a small group of committed activists who are attempting to get their concerns on the agenda.

4. Methodology: Data mining techniques

Data mining (sometimes called data or knowledge discovery) is the process of extracting information analyzing data from a data set and transforming it into a structure that can be analyzed to see if there is any useful information. Many different methods have been developed to analyze data looking for patterns or trends that cannot be observed through traditional statistical methods. For this study, I use techniques from market basket analysis and social network analysis on the 125 white genocide petitions that are the focus of this study. These techniques will be explained briefly below; for additional information see Easley and Kleinberg (2010); Newman (2010); and Tan et al. (2006).

4.1 Market basket analysis

Market basket analysis is used to identify patterns of co-occurrences of objects. In the case of e-petitioning, e-petition transactions (or market baskets) contain the set of petitions each user signs. This data collected over time can be analyzed to see which petitions users frequently signed together.

Some definitions that will be useful to help understand the concept of frequent co-occurrence of objects in the context of petition data:

•
Itemset: any set of items; each transaction is an itemset (subset of petitions signed by user).
•
Support of an itemset $S$ : This is the fraction of transactions which include all the items in $S$ ; that is, the support for $S$ is the ratio of the number of transactions that include all the items in $S$ to the total number of transactions.
•
Frequent itemset: Any itemset $S$ whose support exceeds a chosen support level. Thus, frequent itemsets represent subsets of petitions that are signed together often by users of the petitioning system.

In addition to frequent itemsets, analysis of market basket data can also reveal other patterns related to co-occurrences. For example, for some items x, y and z, a large fraction of users who sign petitions x and y may also sign z. These patterns are captured through an association rule which is usually shown as {x, y} $\to$ {z}. The importance of an association rule is specified using a measure called confidence. The confidence of the association rule {x, y} $\to$ {z} is the ratio of the number of transactions that contain all the items x, y and z to the number of transactions that contain the two items x and y. (Formally, confidence gives the conditional probability that the customer’s basket contains item z given that it contains both x and y.) Thus, association rules with large confidence values also provide insights regarding co-occurrences.
4.2 Social network analysis

Online social media networks (such as Facebook, Twitter, Instagram), and professional networks (such as LinkedIn) are just a few examples. Social network analysis (SNA) helps us to explore the roles of actors or entities and relationships between these actors or entities in these networks. SNA methods are used to try to understand these relationships. SNA is used to study a variety of networks: communication networks, biological networks, economic networks, and terrorist networks (Newmann, 2010).

Centrality is commonly used to measure the level of importance or influence of an actor or entity in a social network (Freeman, 1979). Along with Freeman’s seminal paper, numerous other papers have acknowledged a variety of centrality measures for social networks (Newmann, 2010). Some of these include degree centrality, closeness centrality, betweenness centrality and eigenvector centrality. (Definitions for these can be found in Easley & Kleinberg 2010; Freeman 1979; Newmann, 2010). In the context of this study, the concept of actors or entities having large centrality measure values signifies a more central or important role in determining certain behavior in a given network.

A community (or cluster) is used to identify a group of actors or entities with similar behavior in a social network. Similarity in behavior can be defined in many ways and algorithms are available for partitioning the nodes of a social network into communities according to those definitions (Newmann, 2010).

4.3 Results

The goal of this research is to determine if there is evidence of collective action using the WtP platform. In order to accomplish this goal, I first determine how many signers have signed more than one of the 125 SWG petitions. Additionally, given that a majority of the 125 SWG petitions are created in clusters of two to seven petitions within minutes on a monthly basis, I wanted to see if there are signers who have signed an SWG petition in one month and then came back in a different month to sign another one. The following are my research questions:

For RQ1, I began the analysis by focusing on the first 150 people who signed the petitions before they became available for the public to see and sign. In the analyses that follow, I refer to the first 150 people who signed petitions as “Early Signers”. I was interested to see if any of the Early Signers were signing petitions in one month and coming back to sign similar petitions in another month or multiple months. Individuals who signed a similar petition in more than one month are classified as “Repeat Signers”. Individuals who are part of the first 150 signers and have signed in more than one month are classified as “Early Repeat Signers”.

4.3.1 Early signers and early repeat signers for 125 SWG petitions

For this analysis, I examined the first 150 signers for each of the 125 SWG petitions to see what percentage of the early signers signed a petition in more than one month before the petition became public on WtP to see and sign. As shown in Table 5, note that in the first 150 signatures there were 1500 unique signature identifications. Of the 1500, there were 573 unique Early Signers who were also Repeat Signers which means that 38% of these users are found in the first 150 signatures of the 125 SWG petitions and signed an SWG petition in more than month. Additionally, 1637 or 14% of the 11.979 total unique signature identifications signed petitions in more than one month.

Table 4
Classification breakdown of the total unique signature identifications of the 125 SWG petitions

	Repeat signers	Single month signers	Total unique signature identifications
Early signers	573	927	1500
Not early	1064	9415	10,479
Total	1637	10,342	11,979

Early Signers signify users who signed in the first 150 signatures for the 125 SWG petitions. Repeat Signers signify users who signed a petition in more than one month. All signatures are zip code validated.

Table 5

Months signed with signer counts for 125 SWG petitions by first 150 unique signature identifications

# months signed	# unique signature identifications
1	927
2	116
3	114
4	52
5	30
6	37
7	41
8	72
9	52
10	19
11	13
12	5
13	1
14	6
16	1
17	1
18	2
19	1
20	2
24	1
25	1
26	1
27	2
32	1
33	1
Total # unique signature identifications	1500

Table 5 shows the breakdown of the number of months and the respective numbers of the unique signature identifications who signed an SWG petition in $n$ number of months.

The 573 Early Repeat Signers include all the signers in Table 5 except for the 927 Early Signers who only signed in one month. The 537 Early Repeat Signers can be viewed as a group of hardcore activists. As the number of months increase, the number of signers decreases, however we still see some individuals in this core group of hardcore activists who are coming back in several different months to sign SWG petitions. As shown in Table 5, 116 out of the 537 (total Early Repeat Signers) or 22% sign petitions in two different months, 114 out of the 537 or 21% sign petitions in three different months, 52 out of the 537 or 10% sign petitions in four different months, and 30 out of the 537 or 6% sign petitions in five different months. These hardcore activists are signing petitions in more than one month before the petitions are made public. The Early Repeat Signers must be part of the network (ie. social media, email, etc.) of the creators of the SWG petitions because they know about the petitions. The petition creators have mobilized a small group of hardcore activists who are engaging in online collective action by immediately responding to a direct call to action from the petition creators who are part of their social networks.

4.3.2 Applying market basket analysis to stop white genocide petitions (RQ1)

I used techniques from market basket analysis on the data collected for 125 SWG petitions. Market basket analysis is used to identify patterns of co-occurrences of objects. In the case of e-petitioning, e-petition transactions (or market baskets) contain the set of petitions each user signs. This data collected over time can be analyzed to see which petitions users frequently signed together.

For this analysis, I wanted to see if I could find petitions that different users may have signed in common. In this analysis of the 125 SWG petitions, each person who signed at least one of the 125 petitions represents a market basket and the subset of the 125 petitions signed by the person represents the items in that basket. Since a total of 11,979 people (total number of unique signature identifications with valid zip codes for the 125 SWG petitions) signed one or more of these petitions, the data set for market basket analysis consisted of 11,979 baskets, with each basket containing at most 125 items (or petitions). I used the arules package in R to identify frequent itemsets, association rules and their confidence values.

I computed the confidence values of various association rules of the form {x} $\to$ {y}, where both x and y represent the IDs of one of the 125 petitions. I considered a confidence value of 75% and 50% using a support of 1%. I constructed the following graphs: each node of the graphs represents a petition ID and each edge (x, y) implies that the association rule {x} $\to$ {y} has a confidence value of at least c (which in this case is 75%, for the graph on top and 50% for the graph on the bottom). The graphs in Fig. 8 depicts the association between the petitions. Each node contains a petition ID that represents one of the 125 SWG petitions.

Figure 3.

Association rules at 1% support at 75% (top) and 50% (bottom) confidence for the 125 SWG petitions.

In the structure on the top in Fig. 3, we see 10 clusters of petitions. Two clusters emerge that contain petitions that were created on different days in different months. For visualization purposes, the nodes with the same color are petitions that were created on the same day. For example, in the cluster containing petitions 41, 42, 44, 45, 46, 47, and 48, petitions 41 and 42 were created on 9/1/2013 (colored red), petitions 44, 45, and 46 were created one month later on 10/1/2013 (colored dark blue), and petitions 47 and 48 were created on 11/1/2013 (colored dark green) another month later. This indicates that people are signing SWG petitions on one day and coming back to sign others. In the cluster containing petitions 49, 50, 51, 55, 56, 57, 58, and 59, petitions 49, 50, and 51 were created on 12/1/2013 (colored light pink), petitions 55, 56, and 57 were created six weeks later on 1/13/2014 (colored light blue) and petitions 58 and 59 were created on 3/1/2014 (colored light green). The remaining eight clusters contain petitions (colored grey) that have been created on the same day in the same month. In other words, each of the grey clusters represent a group of petitions that were created and signed on the same day. Most of the clusters show that all of the petitions are highly connected on the basis of common signers and constitute frequent itemsets.

In the graph on the top in Fig. 3 we see that there are clusters that are composed of petitions that are all created on the same day, so that when a person signs one of them, that person can sign the others as well. However as shown in the graph on the top in Fig. 8, two groups of petitions are being signed by people who are signing one petition on one day and then coming back on a different day after at least one month to sign other SWG petitions.

In order to see if I could find additional patterns of repeat monthly signing behavior within the 125 SWG, I lowered the confidence level to 50%, keeping the support of 1%. The graph on the bottom in Fig. 3 depicts the association between the petitions. Each node contains a petition ID that represents one of the 125 SWG petitions.

In the structure on the bottom in Fig. 3, we see 12 clusters of petitions. One cluster emerges that contains petitions that were created on different days in different months. Again, for visualization purposes, the nodes with the same color are petitions that were created on the same day. The large cluster contains all of the petitions that had different creation dates (created in different months, see graph on the top in Fig. 3) as well as an additional cluster which contains petitions 62, 63, and 64 (see the graph on the bottom in Fig. 3). By lowering the confidence to 50% the two clusters of petitions that we saw in Fig. 3 (in the structure on the top), becomes connected to form one large cluster of petitions being signed by people who are signing one petition on one day and then coming back on a different day after at least one month to sign other SWG petitions.

As a result of this analysis, I found many stable patterns of co-signing behavior in the 125 SWG petitions within individual months as well a few patterns of co-signing behavior across months. This indicates that people are signing petitions related to a particular policy suggestion within a short period of time. In other words, there is evidence of people signing multiple petitions that were created on the same day related to the policy suggestion in this case, stopping white genocide. There are also a few instances where there is evidence of people signing petitions that were created on the same day and coming back to the site months later to sign more petitions that support the policy suggestion of stopping white genocide.

In the following section, analyses and results from the social network analysis conducted on the 125 SWG petitions will be presented. The goal of this analysis is to investigate the possibility of communities of e-petitioners forming around particular policy issues, in this case stopping white genocide, which would indicate that individuals are engaging in collective action on WtP. In order to accomplish this goal, the following research question was developed:

RQ2. Is there evidence of e-petitioners forming core groups or “communities” that are characterized by a particular policy suggestion?

4.3.3 Applying social network analysis to stop white genocide petitions (RQ2)

From the petition data, I constructed social network graphs that will allow us to identify highly central petition signers and groups of similar petition signers. To ensure that my conclusions were not affected by users who exhibited low levels of petitioning behavior and that the graph has a large enough sample of people signing n common petitions, I have restricted the network to users who signed at least five of the 125 SWG petitions. In the constructed network, each node represents a person who signed at least five petitions. An edge is added between two nodes if the corresponding pair of users co-signed at least five petitions. The resulting graph has 955 nodes and 89,285 edges (see Table 6).

Table 6
Number of nodes for different number (1–20) of common petitions signed

Graph	# nodes	Edges
G ${}_{1}$	11,979	5,121,904
G ${}_{2}$	3627	714,953
G ${}_{3}$	1835	232,409
G ${}_{4}$	1207	120,441
G ${}_{5}$	955	89,285
G ${}_{6}$	814	69,587
G ${}_{7}$	656	50,089
G ${}_{8}$	589	41,776
G ${}_{9}$	541	36,338
G ${}_{10}$	474	30,897
G ${}_{11}$	450	28,229
G ${}_{12}$	421	25,615
G ${}_{13}$	397	22,895
G ${}_{14}$	381	20,905
G ${}_{15}$	356	18,758
G ${}_{16}$	342	17,228
G ${}_{17}$	327	15,575
G ${}_{18}$	315	14,338
G ${}_{19}$	296	12,992
G ${}_{20}$	282	12,080

The graph consists of one large component containing all 955 nodes. Thus, the component (called the giant component) of the network consisted of 100% of all the nodes. The large number of edges indicates that the nodes in the giant component form a cohesive group (Easley & Kleinberg, 2010).

In the above discussion, I considered a social network in which each node represents a person who signed at least five petitions. Table 6 shows how the number of nodes in the graph drops rapidly as we increase the level of petition signing activity from 1 to 20. (In the table, I use G $i$ to denote the graph where each node represents a person who signed at least $i$ petitions.) Since this analysis is focusing on a network where each node represents a person who signed at least five petitions, the network will be referred to as G ${}_{5}$ (see Table 6).

Table 6 also shows that out of the 125 SWG petitions garnering a total of 11,979 signatures, 3627 people (30%) signed at least two petitions in the set. Additionally, 1835 people (15%) signed at least three SWG petitions. These results indicate that there are people who are signing multiple SWG petitions.

4.3.4 Centrality measures for G

{}_{5}

of the 125 stop white genocide petitions

The network for G ${}_{5}$ , consists of nodes that represent people who signed at least five common petitions, in other words there is an edge between two nodes if they have signed five common petitions. Within this network there will also be an edge between two nodes (people) if they have signed 6 to 125 common petitions. The relationship or edge between two nodes is the number of common petitions that they signed together, and in this case they must have signed at least five to be a part of the network.

After constructing the network for G ${}_{5}$ , I computed centrality measures (closeness, betweenness, and eigenvector) for each node to find the nodes that are the most central according to the structure of the graph. These computations were done with the igraph package in R. For each centrality measure, I computed the set of 100 nodes with the highest values. I found 83 out of the 100 nodes (i.e., 83%) appeared in all three sets, indicating that the group of nodes playing an important role in determining the behavior of the network are roughly the same, no matter which of the three centrality measures is used to find such nodes. The structure of the chosen graph suggests these people are core or integral to connecting others to the graph. They are connected to more people in the graph compared to the other nodes in the network. These people may have created some of these petitions or influenced others to sign petitions by promoting the petitions via their social networks.

4.3.5 Community detection

I used an R package, igraph, and a function called cluster_louvain to identify the communities in the graphs. This tool partitions the nodes of the graph into subsets, with each subset representing one community. Communities represent the subgraphs where nodes in a subgraph have more edges to other nodes in the same subgraph as compared to the nodes outside of the subgraph. The relationships between the nodes in these communities are stronger than they were in the original network graph that I constructed because there are more edges between these nodes than the other nodes in the network. We know that in any of these communities, each pair of people signed at least five common petitions. Again, as with the original network, there can be an edge between two nodes/people who signed six to 125 common petitions. It is not necessary that all the people in the community signed the same set of n petitions. The communities are separated by the petitions that they do not have in common, thus there are fewer edges between the communities. I refer to these communities as “Communities of Signers” see Table 7).

Table 7
The sizes of the four communities of signers that signed at least five SWG petitions

Community of signers	Community size by unique signature identifications
$C_{1}$	269
$C_{2}$	439
$C_{3}$	208
$C_{4}$	39

For G ${}_{5}$ , the community detection algorithm found four communities of signers, denoted by $C_{1}$ , $C_{2}$ , $C_{3}$ , and $C_{4}$ with signature (unique signature identifications) sizes presented in Table 7.

This analysis shows that there are core groups or communities of people signing petitions characterized by a particular policy suggestions, namely stopping white genocide.

4.3.6 Social network analysis 125 SWG petitions: Multiple month signers

In order to see if I could find evidence of users signing SWG petitions in more than one month, I constructed social network graphs that will allow us to identify highly central petition signers and groups of similar petition signers who signed in more than one month. In the constructed network, each node represents a person who signed an SWG petition in $n$ number of different months. An edge is added between two nodes if the corresponding pair of users co-signed petitions in at least $n$ months.

I also ran community detection on all of the graphs. As discussed earlier, the “Communities of Signers” represent the subgraphs where nodes in a subgraph have more edges to other nodes in the same subgraph as compared to the nodes outside of the subgraph. We know that in any of these communities, each pair of people signed a petition in the same n (number) of different months. It is not necessary that all the people in the community signed the same set of months in which a petition was signed.

Table 8
G ${}_{n}$ $n=$ number of different months a signer signed a SWG petition and the communities of signers found with the community detection algorithm

Graph	# nodes/% of users that signed petition	Edges	Communities of signers
G ${}_{1}$	11979	6415423	1703, 1095, 1963, 232, 2157, 1062, 307, 2078, 1382
G ${}_{2}$	1637 (14%)	158675	564, 445, 257, 371
G ${}_{3}$	808 (7%)	55547	335, 253, 220
G ${}_{4}$	529 (4%)	31334	159, 147, 223

Table 8 shows that 1673 out of 11,979 (14%) of the users signed a petition in at least two different months. We also see that 808 (7%) of the users signed a petition in at least three different months. Finally, 529 (4%) of the users signed a petition in at least four different months.

As a result of the social network analysis with the 125 SWG petitions, there is evidence of groups of individuals or “communities” that are signing more than one petition within six month intervals, within one month intervals, and across different months.

5. Discussion and conclusion

The case study of the white genocide petitions is an extreme case of a few individuals systematically creating petitions on WtP on a monthly-basis for four years as a strategy to spread their ideas and mobilize support for their policy suggestion. There is also a smaller group of individuals who saw the SWG petitions that were being created and counter mobilized by creating petitions that were clearly not supportive of the SWG message. Additionally, the SWG advocates were so intent on getting their ideas out in the public sphere, that they encouraged their members to sign the Anti-SWG petitions as well. Recall that 255 people signed both SWG and Anti-SWG petitions. Here we see evidence of the mobilization of e-petitioners engaging in collective action in support of white supremacy and against white supremacy. The SWG campaign is being led by a committed group of activists engaged in a methodical and purposeful pattern of behavior.

I found evidence that individuals were strategically using WtP to mobilize and engage in collective action to gain support for their policy preferences. Two individuals supporting white supremacy systematically created a majority (113 or 88%) of the 125 SWG petitions which were anywhere between two to seven petitions within minutes of each other on the same day on a monthly basis. The remaining 12 SWG petitions were created by seven other individuals. The seven Anti-SWG petitions were created by three separate individuals in an effort to mobilize support against the SWG petitions.

Using techniques from market basket analysis, I found that people were signing more than one SWG petition but they were petitions that were created on the same day. However, in two clusters of petitions, there was evidence of petitions created months apart which were signed together by individuals. So, there were 1637 people who signed SWG petitions in one month and returned to the site months later to sign additional SWG petitions, however the majority of the co-signing of petitions occurred on the same day or within the month.

Both the market basket and the social network analysis were feasible because a sizable number of individuals signed more than one SWG petition. Recall that all of the SWG petitions are similar in that they essentially say the same thing. As I show in Table 4, out of the 125 SWG petitions garnering a total of 11,923 unique signatures, 3627 people (30%) signed more than one petition in the set. Additionally, 1835 people (15%) signed at least three SWG petitions.

The social network analysis I conducted to see how many people signed an SWG petition in multiple months (see Table 4) found that 1673 out of 11979 (14%) users signed a petition in at least two different months. There were 808 (7%) users who signed a petition in at least three different months. Finally, there were 529 (4%) users who signed a petition in at least four different months.

The SWG petitions do not contain suggestions to policy that represent laws that already exist. Instead, the petitioning behavior in this case is a systematic and strategic campaign of a small organized group of people who would like to see stopping white genocide become a policy and have their ideas enter the public sphere. Here we see the what is known as social construction of a problem in agenda setting theory (Birkland, 2007) by this organization who provide a narrative describing their problem in the text of their petitions and the content on their website as they promote what they believe is truly a problem for which they propose several solutions to. Also in agenda setting theory, agenda-building explores how policy issues emerge in the public sphere and gain the attention of policymakers (Cobb & Elder, 1971). Agenda building reflects “the process by which demands of various groups in the population are translated into items vying for the serious attention of public officials” (Cobb et al., 1976, p. 126). This extremist group was attempting to build the agenda, raise the salience of “white genocide” and racist issues, and solicit presidential response. According to Cobb & Elder, “for an issue to attain agenda status, it must command the support of at least some key decision-makers, for they are the ultimate guardians of the formal agenda” (1983, p. 89). The White Genocide Project group are using the WtP platform to get their racist issues to be seen by the Obama Administration which is comprised of “key decision makers”.

Clearly, the policy suggestions proposed by the White Genocide Project group are not mainstream ideas and are representative of a fringe faction of individuals in society. Whether or not stopping white genocide is really a problem that is recognized by policy makers is not the point. However, the actions of the White Genocide Project group can be seen as one example of agenda building in order to articulate their policy ideas in an attempt to gain the attention of policy makers, the media, and the public. The petition data in this case study has brought to the surface a group of individuals who are engaging in radical, extreme, racist, and at times, violent behavior in our society that used WtP as a platform to broadcast their message. Governments with e-petitioning platforms should be aware of the important information that petition data can provide when it is collected and analyzed.

This case is an extreme example of mobilized, collective action e-petitioning signing behavior. As of now, not enough work has been done to characterize a typical case of e-petitioning behavior; however, it will be useful to use the case of the white genocide petitions to serve as a sort of benchmark for future work with other case studies because of the nature of how calculated and systematic the actions of the individuals creating the petitions are and how the strategic online mobilized collective action of small groups of people can occur on an e-petitioning platform such as WtP. This research provides insights into the dynamics of e-petitioning behavior. This work will contribute to the scholarship of online activism by demonstrating that there are segments of the population that use e-petitioning to actively mobilize support for their policy preferences.

The popularity of e-petitioning on WtP when it was under the Obama administration from Sept 2011 to January 2017 suggests that it has the potential to be an important mechanism for citizens to use to participate in policy decision making. This would underscore the importance of understanding the dynamics of e-petitioning activism, which may provide a useful foundation for theory generation in the future. However, as previously mentioned WtP continued to be part of the Trump administration when he took office on January 20, 2017. Recall that in December 2016, Trump announced that they would create a new platform that would cost taxpayers less money (Zurcher, 2017). At the time of this writing, the site does not look any different. It appears that the current administration is not paying any attention to the few petitions that are still being created. Petitions that reached or exceeded the 100,000 signature threshold are not being responded to and are left up on the site for people to see and sign long after the 30 days they are given to reach the threshold.

It is not clear what the future holds for WtP given that we now have new Presidential leadership with Senator Joe Biden winning the election and will be taking over January 20, 2021. WtP could potentially re-gain the interest of the White House and the public who wish to communicate with their government by expressing their opinions on policy issues and mobilizing support for them. E-petitioning is a novel source of information that can be used by policy makers and governments in the course of decision making. Discussion between government leaders and the public can lead to policy making that more effectively address the needs of the citizens. Governments should have a vested interest in providing e-petitioning platforms as a fundamental participatory mechanism for its citizens to interact with them and should actively seek ways to interpret and understand this new form of participation and policy discourse.

6. Limitations and future work

Despite the amount of data presented in this study, there are inherent limitations in analyzing data provided by petitioning platforms in general, and WTP in particular. Petitioning platforms must satisfy the privacy of users but they need to make sure that the signers are real. When WtP went live on September 22, 2011 users who created or signed a petition had to create an account which required their first and last name, zip code and valid email. The signature data that was obtained from WtP and used for this study contained the first initial of a creator or signer’s first and last name and their zip code. Recall that in my analyses a distinct signature ID consists of unique first and last initials followed by a five-digit zip code. This data enabled me to conduct a majority of the analyses in work that I have done examining e-petitioning behavior on WtP. On April 20, 2016 there was a change in required WtP input which omitted the zip code so I can no longer create a unique user by concatenating the first initial of the first and last name with the postal code with any of the data created after this date. As a result, I am unable to work with more current petitions which were created after April 20, 2016 by identifying unique signers.

As I have been working with e-petition data I have found that e-petitions are sensitive to events that occur in the public sphere and that some petitions raise policy issues or propose solutions to existing ones that gain attention from the media, policy makers, and the public. Additionally, I have found that e-petition activity and digital campaigns such as #BlackLivesMatter have followed similar timeline of events that can be documented. One line of future research will explore e-petitioning activity and digital campaigns on social media that are a reaction to the same events that take place in the public sphere to further explore the relationship between these two platforms of social mobilization. This kind of analysis would provide a more fine-grained and interactional understanding of how activist communities mobilize online and engage in collective action using e-petitioning and digital campaigns on social media.

References

Bimber

(2001). Information and political engagement in america: the search for effects of information technology at the individual level. Political Research Quarterly, 54(1), 53-67.

Bimber

Flanagin

, & Stohl

(2012). Collective action in organizations. New York.

Bochel

(2012). Petitions systems: contributing to representative democracy? Parliamentary Affairs, 1-18.

Birkland

(2007). Agenda Setting in Public Policy. In Fischer

Miller

, & Sidney

M.S.

(Eds.). Handbook of public policy analysis: theory, politics, and methods, pp. 63-78. Boca Raton: CRC/Taylor & Francis.

Cobb

R.W.

, & Elder

C.D.

(1972). Participation in American politics: The dynamics of agenda-building. Baltimore, MD: The Johns Hopkins University Press.

Cobb

R.W.

, & Elder

C.D.

(1981). Communication and public policy. In D.D. Nimmo & K.R. Sanders (Eds.). Handbook of political communication, pp. 391-416. Beverly Hills, CA: Sage Publications.

Cobb

R.W.

, & Elder

C.D.

(1983). Participation in American politics: The dynamics of agenda-building (2nd ed.). Baltimore, MD: The Johns Hopkins University Press.

Dumas

LaManna

Harrison

T.M.

Ravi

S.S.

Hagen

Kotfila

, & Chen

(2015a). E-petitioning as collective political action in We the People. Conference 2015 Proceedings.

Dumas

C.L.

LaManna

Harrison

T.M.

Ravi

S.S.

Kotfila

Gervais

Hagen

, & Chen

(2015b). Examining political mobilization of online communities through e-petitioning behavior in We the People. Big Data & Society, 2(2), 2053951715598170.

10.

Easley

, & Kleinberg

(2010). Networks, crowds, and markets: Reasoning about a highly connected world. Cambridge University Press.

11.

Freeman

(1979). Centrality in social networks: a conceptual clarification. Social Networks, (1), 215-239.

12.

Freeman

L.C.

(1996). Some antecedents of social network analysis. Connections, 19(1), 39-42.

13.

Gleeson

J.P.

Cellai

Onnela

J.P.

Porter

M.A.

, & Reed-Tsochas

(2013). A simple generative model of collective online behavior. Proceedings of the National Academy of Sciences, 111(29), 10411-10415.

14.

González-Bailón

Borge-Holthoefer

Rivero

, & Moreno

(2011). The dynamics of protest recruitment through an online network. Scientific Reports, 1.

15.

González-Bailón

Borge-Holthoefer

, & Moreno

(2013). Broadcasters and hidden influentials in online protest diffusion. American Behavioral Scientist, 57(7), 943-965.

16.

Gurak

L.J.

(1995). Rhetorical dynamics of corporate communication in cyberspace: the protest over Lotus MarketPlace. IEEE Transactions on Professional Communication, 38(1), 2-10. doi: 10.1109/47.372386.

17.

Gurak

L.J.

(2014). Trust and Internet Activism: From Email to Social Networks. In Cyberactivism on the participatory web, pp. 17-29. Routledge.

18.

Lindner

, & Riehm

(2009). Electronic petitions and institutional modernization. International parliamentary e-petition systems in comparative perspective. JeDEM-eJournal of eDemocracy and Open Government, 1(1), 1-11.

19.

Linder

, & Riehm

(2011). Broadening participation through E-Petitions? An empirical study of petitions to the german parliament. Policy & Internet, 3(1), 1-23.

20.

Macintosh

Adams

Whyte

, & Johnston

(2008). ePetitioning in the Scottish Parliament. In Digital Government, pp. 487-501. Springer US.

21.

Margetts

John

Hale

, & Yasseri

(2015). Political Turbulence: How Social Media Shape Collective Action. Princeton University Press.

22.

Mechaber

(2014). Here’s How Cell Phone Unlocking Became Legal. Retrieved on January 6, 2016 from: https://www.whitehouse.gov/blog/2014/08/15/heres-how-cell-phone-unlocking-became-legal.

23.

Newman

(2010). Networks: an introduction. Oxford University Press.

24.

Panagiotopoulos

, & Elliman

(2012). Online engagement from the grassroots: Reflecting on over a decade of ePetitioning experience in Europe and the UK. In Empowering Open and Collaborative Governance, pp. 79-94. Springer Berlin Heidelberg.

25.

Pew Research Center. (2018, February 5). Social Media Fact Sheet. Retrieved February 11, 2018, from http://www.pewinternet.org/fact-sheet/social-media/.

26.

Racially Charged Signs Found Near Schools. (2017). Retrieved May 20, 2018, from http://www.necn.com/news/new-england/Racially-Charged-Signs-Found-Near-Schools-411546655.html.

27.

Sinai

, & Smith

(2013, December 6). The United States Releases its Second Open Government National Action Plan. Whitehouse.Gov. https://obamawhitehouse.archives.gov/blog/2013/12/06/united-states-releases-its-second-open-government-national-action-plan.

28.

Tan

Steinbach

, & Kumar

(2006). Introduction to data mining. Reading, MA: Addison-Wesley.

29.

Terry

(2015, January 20). Group Behind White Genocide Billboards Tries To Explain Its Actions. Retrieved May 20, 2018, from https://www.splcenter.org/hatewatch/2015/01/20/group-behind-white-genocide-billboards-tries-explain-its-actions.

30.

Terry. (2015, January 16). White Genocide Billboard Comes Down in Alabama Town. Retrieved May 20, 2018, from https://www.splcenter.org/hatewatch/2015/01/16/white-genocide-billboard-comes-down-alabama-town.

31.

Yasseri

Hale

S.A.

, & Margetts

(2013). Modeling the rise in internet-based petitions. arXiv preprint arXiv:1308.0239.

32.

Zurcher

(2017, December 19). White House to shut petition website. BBC News. Retrieved September 15, 2018 from https://www.bbc.com/news/world-us-canada-42409634.

Electronic petitioning as online collective action: Examining the e-petitioning behavior of an extremist group in we the people

Abstract

Keywords

1. Introduction

2. Background

2.1 E-petitioning systems

2.2 Online collective action

2.3 E-petitioning and collective action

2.4 Social media promotion

3. Case study: The white genocide petitions

4.1 Market basket analysis

4.3 Results

4.3.1 Early signers and early repeat signers for 125 SWG petitions

Table 4 Classification breakdown of the total unique signature identifications of the 125 SWG petitions

Table 6 Number of nodes for different number (1–20) of common petitions signed

4.3.5 Community detection

Table 7 The sizes of the four communities of signers that signed at least five SWG petitions

Table 8 G n n = number of different months a signer signed a SWG petition and the communities of signers found with the community detection algorithm

6. Limitations and future work

References

Table 4
Classification breakdown of the total unique signature identifications of the 125 SWG petitions

Table 6
Number of nodes for different number (1–20) of common petitions signed

Table 7
The sizes of the four communities of signers that signed at least five SWG petitions

Table 8
G ${}_{n}$ $n=$ number of different months a signer signed a SWG petition and the communities of signers found with the community detection algorithm