The Prevention and Detection of Academic Dishonesty: Watch the Online Exam Proctoring Videos

>Compared to the traditional seated classroom, the online environment invites unique challenges for students (Celuch et al., 2021; Clesham, 2010; James, 2016; Karim et al., 2014) and instructors (Christie, 2003; Cluskey et al., 2011; Trenholm, 2007). Some of these challenges include information technology security issues, a lack of necessary technology skills or hardware, a lack of social interaction and communication, a lack of time to complete the exam, privacy issues (Clesham, 2010; James, 2016; Karim et al., 2014) as well as the obvious risk of cheating (Chiesl, 2007; Christie, 2003; Cluskey et al., 2011). As the demand for online courses continues to grow, the risk of cheating becomes a greater concern (Alessio et al., 2017, 2018; Bedford et al., 2011; Hobbs, 2021; Hylton et al., 2016; Karim et al., 2014; Moten et al., 2013). Cheating wastes a significant amount of time and resources intended to develop new skills and knowledge in students; cheating can also damage a university’s reputation and cause employers to lose confidence in the potential of graduates (Norris, 2019).

The COVID-19 pandemic forced millions of students to move from a partially or fully traditional face-to-face environment to a fully online learning environment (Nigam et al., 2021). Since the start of the Pandemic, many universities in the United States reported seeing a substantial uptick in the number of cheating allegations (Hobbs, 2021). Hobbs (2021) explained how a North Carolina State University course coordinator caught about 200 students cheating on the final exam for a statistics class. Mr. Johnson used a computer program that created a unique group of questions for each student. Those unique questions quickly appeared on a for-profit website that exposed approximately one fourth of the students in his course for cheating.

North Carolina State University’s cases of academic dishonesty increased by more than 100% during the 2019-20 academic year, with the largest increase after the students moved to online learning (Hobbs, 2021). During the same time period, Texas A&M University reported a 50% increase in cheating allegations, and the University of Pennsylvania reported a 71% increase (Hobbs, 2021). After catching dozens of cadets cheating, the U.S. Military Academy at West Point terminated a policy that protected cadets who admitted to honor code violations, indicating the lack of tolerance for academic misconduct within this military institution.

Academic dishonesty and academic misconduct are often used interchangeably to describe cheating in an academic setting (Hughes & McCabe, 2006). Cheating is nothing new in higher learning. McCabe and Trevino (1993) studied 6,096 students from 31 colleges and universities and found that 64% of the respondents had engaged in exam misconduct, and 66% engaged in written misconduct. Exam-related misconduct, which is the focus of this paper, occurs when exam information is provided to or received from others, when forbidden resources are used during an exam, or when the examinee is impersonated (Faucher & Caves, 2009; Hughes & McCabe, 2006; Hyltan et al., 2015).

Several studies examined the effect of remote proctoring on the scores of online exams, and the results were mixed. Some studies found that remotely proctored examinees scored significantly lower than unproctored examinees (Alessio et al., 2017, 2018; Carstairs & Myors, 2009; Prince et al., 2009). Other studies found no statistical significance in exam scores between monitored and unmonitored examinees (Beck, 2014; Hylton et al., 2016; Ladyshewski, 2015;). For example, Hylton et al. (2016) found no statistical significance in online exam scores for webbased proctored and unproctored examinees. However, they found that students took a significantly smaller amount of time to complete the proctored exams.

Neutralizing attitudes are justifications that students use to rationalize their defiant cheating behaviors (Pulvers & Diekhoff, 1999; Rettinger & Kramer, 2009; Sykes & Matza, 1957). Neutralizing attitudes are positively correlated with academic misconduct (Pulvers & Diekhoff, 1999; Rettinger & Kramer, 2009). Students express various justifications for committing academic dishonesty, including the “desire to get ahead” (Simkin & McLeod, 2010, p. 441), the lack of time to study due to their working schedule, the need to help a friend, the perception that many others are cheating (McCabe et al., 2001; Rettinger & Kramer, 2009), the view that cheating is easy to do, the desire to earn income, and the lack of adequate skills or confidence to do the work (Christie, 2003; Cizek, 1999; Moten et al., 2013; Norris, 2019). Amigud and Lancaster (2019) studied why students contracted with cheating services; the reasons were grouped into five categories: academic aptitude, perseverance, personal issues, competing objectives, and self-discipline. In addition, in an online learning environment, students perceive a sense of disconnect with the instructor that may translate to a misconception that the student will not get caught cheating, the instructor does not care, or the instructor will not notice the cheating (Anderman et al., 2010).

Several factors play a role in the decision to cheat. Students are more likely to cheat when they perceive their risk of getting caught is low (Whitley & Keith-Spiegel, 2002). Although both individual and situational determinants for academic dishonesty exist, the psychological literature historically found dishonesty is “mostly a function of opportunity” (Landon & Robson, 1999, p. 441). Researchers have found that students perceive a greater opportunity and willingness to cheat in an online learning environment than in the traditional seated classroom (King et al., 2009; Moten et al., 2013; Watson & Sottile, 2010). This article will attempt to educate instructors on how to increase the examinees’ perceived risk of getting caught and decrease the opportunities for students to cheat.

Several studies offered suggestions to try to prevent cheating. Some researchers recommend that institutions require students to report cheating and strengthen a culture of reporting misconduct by implementing honor codes (McCabe et al., 2016; Rettinger & Kramer, 2009). Rettinger and Kramer (2009) found that students cheat more if they believe a substantial amount of cheating occurs. They suggested making sure students realize they can get caught cheating and be penalized with more severe consequences than the students might imagine.

Some researchers have attempted to offer suggestions on preventing cheating, specifically in online courses, through the communication of the curriculum. Volpe et al. (2008) communicated the need for faculty to use their syllabi to communicate their academic integrity policy. Christie (2003) took a broader approach by suggesting a careful focus on (1) designing the syllabus and assessments, (2) presenting content in the course, and(3) developing an excellent student-instructor relationship. Mills (2010) claimed the most effective means of preventing cheating is moral strategy, noting the most common reasons students reported cheating are a genuine desire to learn, finances, morality, legalism/ rules, and fear of consequences.

Aimed at deterring academic misconduct, Cluskey et al. (2011) recommended online control procedures for exams such as offering the online exam at one designated time, randomizing exam questions, presenting exam questions one at a time, limiting the amount of time students may take the exam (Brothen & Peterson, 2012), implementing a lockdown browser during the exam, and modifying at least one third of the exam questions every semester. Brothen and Peterson (2012) suggested making questions more difficult, forcing the students to know the material well to perform well.

“The opportunities and empowering technologies for cheating are numerous” (Norris, 2019, p. 15). Today’s students operate differently than their instructors did in school (Lieneck & Esparza, 2018). Students share their grades, notes, and assignments from their current and previous courses. Lieneck and Esparza (2018) encourage instructors to know that test banks and answer guides are published for all to see. The researchers recommend adapting to the changing times by modifying the method of evaluation in courses frequently as well as creating different quizzes for each student generated from a large pool of test questions.

Other studies focus on the use of modern technology to attempt to thwart cheating (Nigam et al., 2021). Over the last decade, universities and inventors saw the need for computerized proctoring systems (Bedford et al., 2011,; Frank, 2010). These early systems discussed by Frank (2010) have come a long way as technology advances. As online schooling becomes even more prominent, the demand for virtual test proctoring is rapidly growing. Bedford et al. (2011) also noted that accrediting bodies such as the Southern Association of College and School, the Accreditation Council for Business Schools and Programs, and the Association to Advance Collegiate Schools of Business expect online programs to reflect the “integrity and rigor of on-campus programs” (p. 5).

To boost integrity, some universities are turning to a web-based proctor. A “web-based proctor” is an instructor “who utilizes a webcam for video surveillance to observe the users and their environment during the online exam session”(Hylton et al., 2016,). Nigam et al. (2021) wrote a systematic review of 43 articles on artificial intelligence-based proctoring systems between 2015 and 2021. Hussein et al. (2020) compared the features of several popular proctoring systems, including ProctorU, Respondus, Proctorio, and AIProctor. With so many different risk factors associated with online classes, technological advances are working to solve these problems and allow for online testing integrity.

Current technologies seek to provide more than just proctoring software to prevent cheating. Software provides facial ID and recognition to ensure that the intended test taker is truly the one taking the online exam (Smith, 2015). Current systems also monitor exam behavior cues, including phone detection, user verification, text detection, gaze estimation, voice detection, and active window detection (Atoum et al., 2017). This technologymakes it easier for instructors to verify examinees and detect academic misconduct.

The present study involves the use of remote, online exam proctoring software. “Remote, online exam proctoring” entails using the internet and technology to monitor examinees remote from a standard in-person testing location (Foster, 2013). “Remote proctoring technology” refers to the range of technology used to recreate the proctored environment (Karim, 2014).The three goals of this study were to analyze the exam proctoring videos of students taking an individual, closed-book online exam in order to (1) analyze conduct that should raise “yellow flags,” or suspicion, for cheating, (2) identify what conduct exhibits overwhelming evidence (“red flags”) for cheating, and (3) articulate suggestions that will help instructors decrease the opportunity for cheating and increase the students’ perceived risk of getting caught cheating. The author could find no research specifically addressing these issues in closed-book online remotely proctored exams.

The author had used exam proctoring software for approximately 2 years in online courses. During the first semester of using the software, the author did not see the necessity to watch the exam proctoring videos until a colleague mentioned that she would view the proctoring videos for her class to ensure the students were not cheating. The author decided to do the same. When the author watched the videos, the author was shocked to discover that numerous students had suspicious behaviors or were cheating. Despite the author’s numerous warnings about the consequences of academic dishonesty throughout several semesters, students continued to cheat at a surprising rate on closed-book remote, online proctored exams, even though (1) the students were told they would be videoed by exam proctoring software, (2) the instructor would watch the recording, and (3) the students would be severely penalized for academic misconduct.

The current study was performed at a public, medium-sized university in the Midwestern United States. The university permitted faculty to use Proctorio, an online exam proctoring software that videos and records students taking online exams (e.g., Tromblay, 2020). The software recorded the sounds heard during the exam, the movements of the test taker, and the images on the test taker’s screen. The instructor of the course activated exam proctoring as an automatic function of the exams within the sections of the course in this study. When students opened their exams, Proctorio automatically began recording them and saved the recording, which enabled the instructor to view the recording later. The recording was stored on the university’s server and accessed by the instructor through the university’s learning management system.

Proctorio tracked the examinee’s (1) abnormal changes in audio levels, highlighting examinees with significant audio changes during the exam; (2) abnormal head movements, highlighting examinees who look away from the camera, (3) interaction with the quiz page, highlighting less interaction than other examinees through using the mouse, scrolling, and clicking, and (4) computer performance including internet performance and video quality. The software records the examinee as well as the examinee’s screen. The software positioned both recordings side by side and synchronized them so that the proctor may compare the examinee’s actions to the information available on the examinee’s screen. Proctorio also ranked the examinees for each exam based on suspicion of cheating.

The researcher studied the exam proctoring videos of 164 undergraduate students who took an online introductory business course in a business college at a public university. All subjects took the same online business course taught by the same professor over five consecutive semesters within the 2019-20 and 2020-21 academic years. When taking the five online exams in the course, the students were tested using the same pool of true/false and multiple-choice questions. In addition, the student subjects were told the exams were closed-book, individual exams that would be proctored by exam proctoring software.

The author observed the conduct during proctoring videos to look for commonalities (i.e., eye movements, arm/hand/head movements, and sounds) between those students who had overwhelming evidence of cheating and those who did not. Based on these commonalities, the author offers practical recommendations for preventing and detecting exam misconduct in closed-book remote, online proctored exams. While this is not the paper’s focus, most students admitted to the cheating during a conversation with the professor during the semester the student took the course. These known admissions of cheating behaviors help to validate this study’s recommendations.

The researcher obtained approval from the university’s internal review board to perform this study. Most students who cheated in the exams within the study were issued consequences during the semester the cheating occurred. The investigator determined during this study that a few students committed academic misconduct that was not detected during the semester the cheating occurred. The researcher did not issue consequences for any newly discovered instances of cheating uncovered due to the observation of the proctoring videos within the study.

As the title suggests, the principal recommendation from this study is for instructors to review exam proctoring videos, especially videos that are flagged by the software as having a higher than average suspicion level compared to the other students who took the assessment. The author recommends watching the videos with the highest suspicion for cheating in an accelerated play mode to look for suspicious behavior. Over the course of five semesters included in this study, the author concluded that at least 26 of the 164 students committed academic misconduct due to the overwhelming evidence in the proctoring videos. Interestingly, many students did not try to conceal that they were cheating.

When analyzing the exam proctoring videos of students taking individual, closed-book exams using remote proctoring technology, instructors should look for certain “yellow flag” behaviors that raise the suspicion of cheating. When a proctor sees these yellow flags in a proctoring video, the proctor should spend more time carefully watching the videos at a slower speed to analyze the examinees’ conduct. Based on the commonalities among 5. students who committed academic dishonesty in the exam proctoring videos within this study, the author found the following behaviors raise suspicion for cheating:

1. An examinee’s eyes do not stay focused on the screen but instead consistently look to the left and/or right and/or front of the screen. This action is highly suspicious 6. behavior when the examinee reads something other than the exam question. Eyes that consistently move from reading an exam question to moving off-screen may amount to overwhelming evidence of academic conduct, as explained later in this article. This behavior is something the proctor should closely examine. Some examinees look off-screen just to think during the exam or locate the mouse, which can certainly be normal behavior.

2. The sound of paper rustling can be heard during the exam. An examinee taking a closed-book exam that does not require any calculations or scratch paper should not be handling any type of paper during the exam.

3. A paper document or cell phone is viewed by the webcam. A test taker of a closed-book exam should not use outside resources during the exam. Sometimes a cheating test taker "slips up" and accidentally moves the cell phone or paper in front of the webcam, which is strong evidence of academic dishonesty.

4. An examinee fails to complete a room scan and a desk/table scan, as required by the exam rules prior to or during the exam. In this study, many examinees who committed academic dishonesty omitted the required desk/table scan during the room scan. However, some examinees who cheated did complete a desk/table scan and must have had their outside resources (i.e., cell phones or notes) hidden from the view of the webcam during the desk/table scan and then moved the resource in place for use during the exam.

5. Loud sounds such as music, television, or people talking in the background can be heard. Cheaters often used loud sounds to cover up the sound of paper rustling during the exam. In one instance, the examinee obviously communicated with another person throughout the exam and used music to attempt to cover up the conversation.

6. The examinee’s eyes are not viewable by the webcam. Sometimes the examinee wore a hat, which blocked the webcam’s view of the examinee’s eyes; in this case, the examinee may position the bill of the hat so that the webcam may not view the examinee’s eyes. In other cases, the examinee positioned the lid of the screen, so the eyes are not in view of the webcam; instead, only the examinee’s nose, mouth, and neck may be seen.

7. Other people are in the room and/or communicating with the examinee. In some cases, the examinee obviously communicated exam material with another person throughout the exam and committed academic dishonesty. Occasionally, a person, whom the examinee does not anticipate, walks into the room during the exam; in this case, honest examinees quickly made clear to the person entering the room that he or she is taking an exam. A careful examination of the video is likely to determine whether the additional person was intentional or accidental.

8. The distance between the student’s eyes/ face and the screen is larger than normally expected. In some cases of academic dishonesty, the student kept a significant distance between the computer screen and the student’s eyes. The closer the examinee’s eyes are to the screen, the more the eyes and head have to move back and forth to read the exam questions on the screen, triggering the software to flag for suspicion of misconduct. Several exam takers who cheated placed a longer than average distance between the screen and their eyes, which minimized their eye and head movements.

9. The examinee eats during the exam, and the examinee’s eyes linger near the food, appearing to be reading something. This case requires a careful examination of whether the examinee is reading something near the food. The proctor can quickly eliminate this issue by refusing to permit students to eat during the exam.

10. The examinee’s arms and hands appear to be holding/turning paper or an electronic device. Several examinees who cheated used their hands to turn or hold an outside resource. If the upper arms are in view of the webcam, the proctor can check for off-screen reading near the time of the arms moving and listen for the sounds of paper or a device moving.

Based on the observation of the exam videos, the key to detecting academic dishonesty in an online, closed-book, remotely proctored exam is to focus on the examinees’ eyes. Proctors should determine the position of the examinee’s eyes when he or she is reading an exam question based on the size and position of the exam question on the screen. The exam proctoring software used in this study synchronized the webcam’s view of the test taker and the view of the test taker’s screen, so a proctor knows what the test taker is doing and can see on the computer screen at a particular moment in time. The synchronized recordings of the test taker and the test taker’s screen are pivotal to detecting academic misconduct.

In uncovering academic dishonesty, the question becomes whether the examinee’s eyes move away from reading the exam question to reading something other than the exam question. When examinees read something other than the exam question, their eyes typically, but not always, track to the right and left,typically looking to the side of the screen or in front of the computer; this is the most important evidence that ties the examinee to misconduct. The typical pattern of a student committing academic dishonesty in this study during a closed-book, online exam is as follows: (1) the test taker reads a question, (2) then the test taker reads something to the side or front of the screen, and (3) then the test taker answers the question. The examinee repeats this sequence for multiple questions during the exam.

An occasional look off the screen is not a problem and can be expected. For example, proctors should be aware that some examinees will look off the screen briefly to locate the mouse or think about a question. However, as a proctor becomes more proficient in examining videos, the proctor will notice the difference between a brief off-screen look to locate a mouse or other harmless look off-screen and a lingering off-screen view performed to read or study something other than an exam question.

The examinee’s eyes are the key to detecting academic dishonesty; however, in some cases, the exam proctor may find additional evidence to boost a claim of academic dishonesty, such as (1) the proctor can hear the examinee moving outside resources such as a cell phone or piece of paper, (2) the webcam catches a glimpse of the test taker reading a piece of paper or a cell phone, (3) the examinee hides his or her eyes from the webcam, or (4) the examinee fails to perform a desk scan.

Based on the findings of this study, the author makes several suggestions for instructors who assign closed-book online remotely proctored exams. In order to decrease the examinees’ opportunity for cheating (Landon & Robson, 1999) and increase the students’ perceived risk of getting caught cheating (Whitley & Keith-Spiegel, 2002), the author suggests the following instructor communications with students (e.g., Christie, 2003; Volpe et al., 2008):

1. The instructor should communicate that the exam proctoring videos will be viewed.

2. The instructor should communicate and impose severe consequences for academic misconduct, such as an F for the course.

3. The author recommends that instructors mandate the following rules for individual remotely proctored, online exams:

(a)This is an individual, closed-book exam.

(b)Keep your eyes focused on the exam questions, not elsewhere.

(c)After starting the exam, the proctoring software will ask you to scan the room. During the scan, you must show the room and the desk or table on which you will be taking the exam.

(d)The testing environment must be quiet. No music, television, or talking may be heard in the background during the exam.

(e)No hats, headphones, or earbuds may be worn during the exam.

(f)Your entire face must be shown in the camera at all times.

(g)Your face and testing space must be illuminated and may not be dark during the exam. You may need to adjust your webcam so your face is illuminated.

(h)No other people may be in the room while you are taking the exam. Make appropriate people aware that you will be taking an exam, so they do not talk to you or distract you during the exam.

(i)You must stay seated during the entire exam.

(j)You may not eat during the exam.

Failure to follow any of the above rules may result in a claim of academic misconduct. The instructor watches exam proctoring videos. While it may be tempting for students to cheat during an online exam, it is very easyto detect. Do not waste your time/money and the instructor’s time by cheating. The minimum consequence for academic misconduct is an automatic F in the course.

The exam proctoring software used in this project ranked the examinees’ proctoring videos in the order of the most suspicious recordings to the least suspicious recordings. The suspicion rank order is not foolproof, but it does give the instructor a place to start viewing videos. The author recommends reviewing the videos ranked highest for suspicion. To save time, the author recommends watching the proctored recordings on an accelerated-play speed to look for suspicious activity. If suspicious behaviors are noted, the proctors can slow down the videos to analyze the examinees’ conduct more thoroughly. Viewing the videos in a fast-forward mode permits instructors/proctors to view more examinees in a shorter time. If the instructor has time, the instructor may benefit by watching all of the videos because the author has found that cheating occurs in several videos labeled by the software to have a lower-than-average suspicion level than the other videos.

Some students in this study used outside resources for multiple exams in the course. Most students’ behavior tended to get “sloppier” with each successive exam. During the first exam of misconduct, a cheating student’s behavior tended to be sly, covert, and careful. In later exams, the cheating was much more apparent (e.g., having more pronounced head movements, reading off-screen for longer, hiding eyes from the screen, and showing paper on the webcam). This behavior raises the question of whether instructors should address the academic misconduct as soon as possible or wait for the cheating behaviors to be more obvious.

One of the primary findings from this study is the importance of watching the exam proctoring videos for online, remotely-proctored exams. Misconduct in this context is fairly easy to detect; proctors do not need to be an expert in academic dishonesty to notice the misconduct. When students exhibit suspicious behaviors, the video is worth analyzing more stringently and for a longer time. The key to detecting academic dishonesty is to closely monitor the examinee’s eyes, know the eyes’ position when the examinee is reading a test question, and determine whether the student reads an outside resource during the exam.

The author recommends implementing exam proctoring rules to minimize the examinees’ opportunity to cheat and increase the students’ perceived risk of getting caught (Whitley & Keith-Spiegel, 2002). In addition, the author suggests communicating to examinees that the exam proctoring videos will be viewed and communicating substantial consequences for academic misconduct. This study has broad ramifications regarding business education and other educational disciplines that use online exams through distance learning. The research also applies to the professional licensing exams (Weiner & Hurz, 2017) and employment exams.

This research had no external sponsors.

The author received internal review board approval from the author’s institution to complete this research project.