Introduction

The year is 2026, and Regional State University (RSU) has bounced back from the enrollment woes that resulted from the COVID-19 pandemic of 2020 and 2021. After intensive faculty development efforts during the pandemic, the majority of instructors on campus became skilled online instructors. RSU set itself apart during the pandemic by rejecting remote proctoring systems by issuing several strong recommendations against such tools to their faculty and offering faculty development opportunities around authentic assessments. RSU was featured in several national news articles for their stance on remote proctoring and celebrated for their attention to student privacy. New students were attracted to RSU because of its reputation for having strong online programs and also because of RSU’s commitment to student privacy and authentic assessments.

Aisha is a new student who specifically came to RSU because of its reputation for providing an excellent education with a personal touch. She had a friend, a year older than her, who had ended up at one of the for-profit online education schools and was routinely subjected to various forms of invasive surveillance. So, when Aisha’s math professor told the class that they would be using a remote proctoring system for some of their homework assignments she was a little put off. She knew the school did not have an official ban on such technologies but she had read the news articles and the Provost’s and the President’s statements advocating for student privacy. She tried to talk to her professor and asked specific questions about what data would be collected and how long it would be held but the professor only directed her to a long “terms of service” and privacy policy that was hard to understand. She decides to bite the bullet and just use the technology for this one class.

On her first attempt to use the technology she has to pass an identity check using facial recognition. She keeps getting an error that the software cannot detect her face and decides to call RSU’s technology help desk to see what the problem might be but no one there seems to know about this software. They ask her how she got access to this system and in tracing her steps she mentions that she started by clicking a link in the Learning Management System (LMS). So they transfer her to Summer, an administrator of the LMS. Summer tells Aisha that though she got to this proctoring system through the LMS that the school does not have any administrative access to this remote proctoring system and cannot help her in any way—that she will need to contact the remote proctoring company’s help line.

Little does Aisha know that Summer is a student privacy advocate and is shocked to hear that some students are being subjected to a remote proctoring system. Summer begins the process of trying to figure out what is going on but while waiting to hear back from the customer support rep at the math homework system she sees that RSU is in the national news—headline: “RSU Student Humiliated.” The article features Aisha who is recounting her experience with the remote proctoring company’s help line that Summer referred her to, and it just sounds horrible. They asked Aisha to jump through various hoops including shining a light on her face and even asked her to remove her hijab. Matters are made worse when a week later the proctoring company suffers a major data breach potentially exposing a massive amount of student data on the dark web. Summer is at a complete loss for what to do. Protecting student privacy is part of her job, but the company violating students’ privacy has no real relationship or accountability to RSU. She finds a clause buried in the university’s contract with the math homework system saying that they consider the contract to be binding for themselves but also for any partnerships that they enter into. Summer wonders if this means that the math homework company is legally responsible for the remote proctoring company—even so, it sure feels like the damage to student trust will fall back on the university.

The story above is a semi-fictional speculative account of potential harms that can come from fourth-party vendor relationships, in which outside companies partner with one another with little oversight by the educational institution. Ross (2017) explains how “speculative methods are particularly important for the study and analysis of digital education because of its rapidly changing nature, and the need to anticipate potential ‘unintended consequences’ of such rapid changes.” We based this story on real events that we experienced in our own professional context and the documented experiences of other students, faculty, and staff, which are reviewed in our remote proctoring Harm Index (Table 2). In this paper we will define fourth-party vendor relationships and investigate how the harms of remote proctoring technology are particularly pernicious in the context of a fourth-party deal.

We describe the nature of different types of fourth-party relationships with a focus on the remote proctoring company Proctorio’s partnerships. Our decision to focus on Proctorio is due to our practical experience working with the product through an integration with McGraw Hill Connect, a program that provides electronic textbooks and other learning materials. This paper adds to the larger body of literature that takes a critical view of educational technologies, in that these technologies often act as “mechanisms of economic capture, surveillance, and control” (Paris et al. 2021) and often work against the core pedagogical goals of educational institutions. In that institutions of higher education have a responsibility to treat students’ data responsibly and monitor the data privacy practices of vendors with access to student data, we suggest that fourth-party partnerships obstruct educational institutions’ regular oversight of student information.

Methodologies

Literature review

As remote proctoring technologies are relatively new and knowledge about them and their harms is rapidly evolving, the literature we have reviewed includes not just scholarly works but also news stories, especially accounts from students who have been exposed to remote proctoring technology in educational settings. Because we aim to define and detail the harms of these technologies, both generally and in the context of fourth-party deals, we found it necessary to focus on literature by and about students, faculty, and staff who had been negatively impacted by these technologies.

Content analysis

Our methodology includes content analysis of Proctorio’s website, help documentation, and publicly available information about their partnerships with other education technology companies (Hsieh and Shannon 2005). We analyze these materials to understand the nature of the relationships between Proctorio and its partner companies. The appendix outlines the documents analyzed in our content analysis.

Authoethnography

We use collaborative autoethnography to analyze our experience identifying the existence of a fourth-party surveillance company operating on our university campus, and requesting that the third party restrict the fourth-party from working with our campus users (faculty and students). Autoethnography is a method used to systematically analyze and interpret personal experiences (Ellis, Adams, and Bochner 2011). Collaborative autoethnography enables researchers to work together to combine their shared and personal narratives to make meaning out of experiences (Chang, Ngunjiri, and Hernandez 2016). Our collaborative ethnography was constructed using alternating personal narratives, in which we engage in critical reflection on the experience of identifying and ultimately removing a fourth-party proctoring service from the campus where they are both employed as instructional designers. Our perspective is one of current instructional design and educational technology professionals in higher education, employed at University of Michigan–Dearborn (a small, undergraduate-focused campus of the U of M). Both of us were already opposed to remote proctoring for the reasons detailed in Table 2 before the events in our autoethnography unfolded and had participated in public conversation about the harms of remote proctoring on social media. We additionally co-authored a peer-reviewed paper with colleagues about authentic assessment as an alternative to remote proctoring and issued a press release with Fight for the Future about the paper and expressing our difficulties with the McGraw Hill partnership with Proctorio. These experiences shape our narrative in that we likely had more prior knowledge of the harms of remote proctoring and the potential roadblocks to removing it from our campus. As our narrative will show, the process was complex, frustrating, and time-consuming despite our prior knowledge and preparation.

Defining the Fourth-Party Relationship

Third party services are common in the educational technology industry. For example, instead of building a platform on which to administer online courses, a university will often contract with a company to provide a Learning Management System (LMS). The company hosts and maintains the software, and the university pays a fee to use it. A fourth-party relationship comes about when vendor A, with whom the institution has an established relationship, partners with vendor B with whom the institution does not have an established relationship (Aldoriso 2020). Language can be confusing here because the vendor may refer to the partners as their third-party partners. This is because they see themselves as the originating party. However, this is an erasure of the institution as the originating party.

This kind of vendor relationship in higher education is a problematic one in that it is once removed from a direct contractual relationship with the higher education institution. Contracts and or user agreements with institutions may simply note that external services may be offered through Vendor A’s product, but the use of Vendor B’s product is governed by Vendor B’s terms of use. Note that these clauses are general, referring to a hypothetical Vendor B, rather than a specific one. For example, McGraw Hill Education’s terms of use states,

The Products may contain links or connections to third-party websites, services, or other technology provided by third parties that are not owned or controlled by us (“Third-Party Services”). When you access Third-Party Services, you accept that there are risks in doing so, and that we are not responsible for such risks. A Third-Party Service may include or be accompanied by a separate service, license, privacy policy, or other agreement (“Third-Party Agreement”), in which case that Third-Party Service is provided solely under the terms of that separate Third-Party Agreement. (McGraw Hill Education 2020)

Here, we see McGraw Hill (a third party to the university) referring to other services as a third party. This is confusing because what they refer to as a third party in these terms of use is the same as what we refer to in this paper as a fourth party because we recognize the university as the originating party. Even though McGraw Hill’s “third parties” (fourth parties to the university) are yet unknown to the user, McGraw Hill has already absolved itself of responsibility for these products. While it is entirely possible that in the process of using an add on (such as a Proctorio to McGraw Hill) the user will have to accept the fourth parties terms of use, it is only the end user (likely a faculty member or student) that comes in contact with those terms, rather than a representative of the institution with expertise in learning technology and student data privacy.

Fourth-party partnerships almost always introduce a new feature or functionality to the third-party product. Depending on the product and where any particular individual exists in the university hierarchy, new features from this relationship may be perceived as desirable or neutral (Gogia 2021). However, these partnerships merit increased suspicion when the fourth-party product does not have an obvious connection to the third-party product, or does not have an educational value. While such spurious partnerships could be viewed as simple money-making opportunities, these relationships can have dangerous implications when they involve problematic technologies that have a history of harming students, such as remote proctoring, because these technologies make their way into classrooms with little institutional input or support.

Jones et al. (2020) argue that institutions of higher education are examples of “information fiduciaries” (Balkin 2016), meaning that they have a particular obligation to treat the data of their primary stakeholders (students) responsibly. Fourth-party agreements constitute a threat to educational institutions’ responsibility to their students in terms of data privacy because, as we describe above, they expose students and their information to technologies that may harm their students or operate with bias.

Doorways Between Companies and Institutions

There are many examples of university maintained systems that allow for the possibility of integration with third-party or fourth-party vendors. For example, Google Apps for Education allow for various integrations with their suite of tools and Zoom has a marketplace where vendors can be allowed to offer integrations. One common option for a university-maintained Learning Management System (LMS) has long been the possibility to use plug-ins, and more recently the Learn Tools Interoperability (LTI) protocol (Severance, Hanss, and Hardin 2010), for integrating tools. There is an important distinction between a “true” fourth-party partnership and an integration option with a university maintained system. In instances where there is a possibility for integration, a university has some direct influence to work with the vendor to negotiate and navigate university policies around such integrations; for instance data privacy or procurement protocols. In some way (either by a dedicated employee or by a university committee) the university works directly with the vendor. This is different from fourth-party relationships which are inherently relationships between vendors which bypass the university entirely. It is also important to note that these kinds of integrations can actually act as the mechanism through which fourth-party vendors can come in; considering that once a third party is integrated that third party can then easily partner with a fourth party.

Relationships between a third-party and a fourth-party vendor can vary widely depending on the business partnership but ultimately these relationships act as doors into the institution. Just as there are many different types of doors: french doors, sliding doors, split doors; which can be in multiple states (open, closed, ajar), there are many different kinds of relationships between third- and fourth-party vendors. It is important to understand these relationships as they can deeply alter the power dynamics that are at play. Below are several relevant examples of relationships that can apply to partnerships between educational technology companies:

• • Integration Possibility – an option for two technologies to work together when an institution has separate agreements with both of them. Often a precursor to a fourth -party integration but not necessarily a fourth-party integration in and of itself.
  • Free – the third party offers the fourth-party service or product for free within their system.
  • Freemium – the third party offers some basic functions of the fourth party’s service or product for free with an option to pay for more advanced features. The free product in the Freemium model may not be totally free for the user—they may “pay” using an alternative currency known as “mind share,” or the “development of awareness for the provider’s brand and the consideration for purchase of future commercial products and services” (Pujol 2010). The free version, while a product or service in itself priced at $0, also serves as a type of marketing device for the paid product.
  • Resell – the third party offers the fourth party’s service or product at a cost. The cost can be charged to the institution, but is also frequently placed on the student.

Identifying the Doors: A Content Analysis of Proctorio’s Fourth-Party Partnerships

The remote proctoring company Proctorio lists a number of partnerships with other companies on its website, which are all presented under the category of “assessment platforms.” Our content analysis of these partnerships is summarized in Table 1 and details of the sites reviewed can be found in the Appendix.

The Proctorio website groups all of these partnerships together as “assessment platforms,” a decision which obscures the significant differences between the services that the partner provides (e.g. textbooks, assessments, active learning) but also what the nature of the partnership is between the company and Proctorio. For example, at the time of this writing, Proctorio’s partnership with McGraw Hill Connect allows any instructor who has adopted a McGraw Hill text to enable Proctorio on student activities that are assigned through the Connect platform (and additionally to enable a paid premium product for students), while Proctorio can only be used on the Cirrus Assessment platform when an institution has a separate (paid) agreement with Proctorio—an agreement that likely requires some attention from someone at the university.

Loose Hinges: Inconsistent Messaging in Fourth Party Relationships

Fourth-party relationships can introduce confusion about the purpose and functionality of the products involved. Proctorio defines itself as a tool for promoting academic integrity, which gives instructors information that they can use to make their own decisions about whether cheating has occurred during a given assessment. However, because many instructors encounter and use Proctorio’s product through these other fourth-party products, they may receive all of their information about Proctorio through the help documentation of this other vendor. These Proctorio partners and resellers can describe the purpose of Proctorio’s product in their documentation in ways that are not entirely congruent with Procotrio’s stated purpose. For example, at the time of this writing, Cirrus Assessment’s help documentation page titled Integration with Remote Proctoring from Proctorio states that “Cheating behaviour is automatically detected by the AI” (see Appendix).

Although we strongly object to the idea that any behavior detected by an AI can represent “cheating behavior,” this statement from Cirrus is notably at odds even with Proctorio’s public statements about their product. On their FAQ page under the question “Does Proctorio utilize algorithmic decision making?” Proctorio states that “No,… Proctorio’s software does not perform any type of algorithmic decision making such as determining if a breach of exam integrity has occurred. All decisions regarding exam integrity are left up to the exam administrator or institution” (see Appendix).

Additionally on the Proctorio FAQ page for the question “How do you decide what behaviour counts as ‘cheating’?”, Proctorio themselves state, “Only the exam administrator or the institution can dictate what type of behaviour they want to monitor over the course of an exam. Exam administrators will then review exam attempts to determine whether any flagged behavior was truly infringing on the integrity of the exam” (see Appendix). Moreover, Proctorio’s own “acceptable use policy” prohibits punishment of students based solely on Proctorio reports. The policy states that “Institutions and their representatives are prohibited from making any negative decisions regarding exam integrity or from imposing any other negative consequence or detriment on an End User based partly or entirely on Proctorio’s analysis” (see Appendix).

Miscommunications like these about what the product is for and how it can be used, show how misleading fourth-party relationships can be. But while fourth-party partnerships are complex they are not inherently bad. Returning to our speculative narrative from the introduction, the math homework system we mention could integrate a digital graphing calculator from an outside vendor to assist students with their work with little negative consequence. However, when fourth-party deals are made with companies that offer harmful products these relationships become particularly problematic.

What’s the Harm?

Given that fourth-party integrations can expand the functionality of a third-party tool, often for free, what is the harm? Why would some schools not want free access to more functions? To explore this we review literature around the various harms that have been caused by remote proctoring systems. Reports of the many harms of remote proctoring systems have been widely documented especially since the beginning of the COVID-19 pandemic. Walker et al. (2020) outlined harms of remote proctoring technologies specific to nursing students, the nursing profession, and to the public. To address the important question of “What is the harm?” we have indexed harms from a broad perspective to create our Harm Index of Remote Proctoring Systems, which we present in Table 2. Here we are exploring three different levels of potential harm: harm to the student, harm to the institution, and harm to larger society.

Closing the Backdoor

The following is an autoethnographic reflection between us (the two authors) about our experience discovering and investigating a fourth-party proctoring option at our university, which had a stated anti-proctoring technology stance (Silverman et al. 2021). These events took place between February 5th and May 15th of 2021 and it is of importance to note this is the time it took to have a technology removed that was never vetted or approved by the university in any way. McGraw Hill’s partnership with Proctorio was established as early as February of 2020 (@mheducation 2020) potentially in response to the mass transition to online teaching during the COVID-19 pandemic. However, as this autoethnography will show, no one at the institution seemed to have any knowledge of it for a year. The partnership began to draw wider attention when a parent’s group published an open letter asking McGraw Hill to end their partnership with Proctorio (Ongweso 2020).

Autumm: In February of 2021 I was browsing twitter when I came across conversations about a McGraw Hill Connect (MHConnect) partnership with Proctorio. I did some googling and came across their webpage outlining this partnership and stating that the proctoring options would be part of any text with a 2019 copyright or newer. Days prior, I had been in a Canvas administrators meeting for our institution where the MHConnect integration had been discussed. It was noted that the new integration was mostly a “pass through” over to the publisher’s private platform, which they controlled. I had thought this was a good thing at first, that perhaps it meant that less of our student’s data would be passing between systems but then realized when I saw the Proctorio partnership that this would mean that our students could be subjected to this harmful technology without much oversight from our institution. This upset me for personal reasons but also because our campus had specifically rejected remote proctoring since the beginning of the pandemic. I reached out to Sarah on Twitter DM to vent.

Sarah: When Autumm contacted me to tell me about the McGraw Hill Connect partnership with Proctorio and how it might affect our students I was first and foremost angry. But then I immediately thought about our campus decision to reject remote proctoring (as Autumm mentions) and how that should carry some weight in our working relationship with McGraw Hill. After all, we are a customer and partner of theirs, and it is very common for customers of technology companies to ask to have the product configured to their liking (just take a look at several different universities’ instances of the Canvas LMS to see an example of edtech customization). I suggested to Autumm that we attempt to have Proctorio removed from MH Connect platform for UM-D users, assuming that this would be an easy request to fulfill.

Autumm: When I contacted Sarah I was feeling that we were somewhat at a dead end because of the nature of the integration that I had heard about in the admin meeting days prior. It did not sound like other integrations that I’d had experience with (having LMS admin experience from previous roles) that would bring outside functionality to the LMS—and over which an LMS admin would have more control. Rather, this integration just seemed to pass credentials over to another McGraw Hill controlled platform. Talking with Sarah she questioned my assumptions and asked many questions about configurations and customizations that might be possible. She made good points and I felt like it was a long shot but I reached out to our institution’s lead LMS administrator to point out this new partnership and question if customizations or a shut-off might be possible. He was not aware of this partnership but agreed that it was concerning and said that he would reach out to McGraw Hill. After waiting about two weeks and hearing nothing I reached back out to him and he said he would reach out again. Two days later he wrote to me to say he had met with representatives of McGraw Hill who said that they would turn the integration off for our school but that there were conditions: (1) that they wanted to look and see if there was anyone using the integration and (2) that they needed two weeks to turn it off. I was unhappy about the two week waiting period, especially since we had already waited two-weeks to initially get a response, and I never fully understood why it was required. Later that day one faculty member was identified as already using the integration and McGraw Hill additionally requested an email from the associate provost to perform the shut off. Arrangements were made to work with the faculty member who was using the system to find alternatives and the official email from the associate provost requesting the shut off went out. However, a month later Sarah was working with a faculty member who was using MHConnect and found that the options for proctoring were still there.

Sarah: That particular faculty member hadn’t wanted to use the proctoring feature but noticed that it became available. I think it is important to note that it was a complete coincidence that I was working with an instructor that was using MHConnect, and thus was able to verify that the proctoring feature had not been turned off. Absent my working relationship with this faculty member, none of the instructional designers on our campus would have access to Connect or any fourth-party tools that are connected to it. Together, Autumm relayed that the proctoring feature was not turned off to the LMS administrator and our associate provost, who contacted McGraw Hill again. At this point, we were assuming there was some sort of technical misunderstanding. We eventually heard back that while McGraw Hill could turn off the integration for users that accessed Connect through the LMS, it could not turn it off for those that logged in directly through the Connect site. Late in April I tweeted, “Frustrating day for resisting surveillance and e-proctoring. Found out that McGraw Hill cannot disable the Proctorio integration in “Connect” for all our users. This integration is built on the presumption that eproctoring is an uncomplicated value-add to any course. It is not.” Evidently, someone from McGraw Hill saw this tweet, and reached out to our associate provost by email offering to have a Zoom call in which we could clarify the details of McGraw Hill’s position.

Sarah and Autumm: We accepted the offer for a Zoom call, and decided to use it as an opportunity to better understand how Proctorio partners with McGraw Hill in addition to discussing how Proctorio could be removed for our campus users. We discussed a plan for McGraw Hill to deactivate Proctorio for all our campus users, both those that log in through our LMS and those that log in through the McGraw HIll Connect site. We then inquired as to whether the Data Privacy Agreement (DPA) we had signed with McGraw Hill covered other partnerships that they chose to make, such as the one with Proctorio. They responded that they viewed the DPA as applicable to any other technology companies that they partner with, meaning that our original DPA covered Proctorio being used through McGraw Hill Connect by our users. They also reiterated that they did not want to force Proctorio on anyone, and that they were happy to pursue various avenues to restrict its use on our campus if that is what we desired. In response to our dissatisfaction that they had integrated Proctorio into our users accounts without informing us or asking permission, they maintained that if we had a campus policy against remote proctoring, it was primarily our responsibility to inform faculty of the policy and enforce it.

Discussion and Recommendations

We have described the problematic nature of current fourth-party partnerships, but there is potential for more problematic future partnerships. Fourth-party partnerships may exist explicitly to circumvent campus decisions or policies (such as administrative policies, or faculty governance) or to respond to budget and purchasing constraints (Gogia 2021). The loopholes which are created as part of these deals can do real harm. Considering how we speak about and rationalize such technologies is an important part of analyzing how they end up existing on our campuses.

Caines (2021) described a weaponization of care around surveillance technologies where they are sold and rationalized under a rhetoric of care and gives specific examples from remote proctoring companies. Adjacent to this frame we also see remote proctoring companies using what Herzog (2010) called the “banality of surveillance.” With this construct we see companies making the case that the technology is essential and, though it may not be perfect, we need to suffer with the drawbacks because the good outweighs the bad (McFarland 2021). Third parties that form relationships with remote proctoring companies also implicitly make the case that remote proctoring is essential (and harmless) by integrating it into their product without user or institutional consent. But the banality of surveillance around remote proctoring is nothing but smoke and mirrors, as our Harm Index (Table 2) shows. The harms suffered by students, institutions, and larger constructs such as academic freedom from these technologies is very real. Additionally, multiple educators have pointed out that other, more authentic kinds of assessments do not even require exams (Crosslin 2021; Silverman et al. 2021).

Relationships between educational technology companies and educational institutions can be fraught. Often, there is a disconnect between how institutionally approved technologies are chosen and whether they support the kind of education the institution wants to provide (Cohn 2021). Our first recommendation for institutions to have more control over fourth-party integrations is for universities to consider not only which technologies to adopt, but which technologies, and specifically what functionalities, they do not want to adopt. Methods and tools need to be created to assist with alignment of the university’s mission, strategic plan, or other guiding principles for evaluating not just the benefits but also the harms of technologies. The addition of new functionalities to an existing educational technology tool through a fourth-party relationship can be particularly subtle and they often come with a techno-utopian sales rhetoric that fails to imagine that this addition could be anything but a good thing. Without consideration of where boundaries exist for the institution in regards to what technology is acceptable and what is not it is impossible to express issues with fourth-party relationships. Passing resolutions and offering recommendations against such technologies can go a long way against limiting the use of harmful technology.

Our second recommendation is for institutions to leverage their influence as direct paying customers, or as the provider of a sales environment of these tools, to demand that the third party remove the surveillance functions provided by the fourth-party companies. Enterprise systems are regularly configured to customers specifications. Shutting off these integrations is technically possible. Success in getting a fourth-party integration removed may vary depending on the specifics of the fourth-party partnership (as discussed above). We speculate that McGraw Hill was willing to grant our request for several reasons. For one, on balance it is better for them to retain us as a satisfied customer than to insist on proctoring functionality in our school. In addition, there is the name recognition of our institution and our relationship with our flagship university. Finally, it is important to note that we did take steps to bring public light on our situation. We wrote a peer-reviewed paper about our experiences and issued a press release (Fight for the Future 2021), which could have impacted our ability to successfully get this integration removed. While resisting harmful fourth-party integrations is a difficult, time-consuming, and unpredictable endeavor, we hope that other educators feel empowered to do so based on our experiences.