Student Perspectives on Assignment Deadline Policies in Computer Science Courses

While the COVID-19 pandemic encouraged higher education to be more flexible, there is now growing uncertainty regarding the usefulness of flexibility as students return to in-person classes, particularly in fast-paced, rigorous fields such as Computer Science (CS). A big part of providing flexibility in education is through assignment deadlines. On one hand, flexible deadline policies may reduce student stress, help students face unexpected circumstances, and improve student learning; on the other, too much flexibility may lead to procrastination and poor time management. In order to help CS instructors make informed decisions when choosing an assignment deadline policy, we compared student perspectives on different deadline policies and assessed their benefits and drawbacks. Results from a survey of 56 undergraduate students showed that deadline policies that provide both flexibility and structure for students have the advantage of being perceived as more welcoming, causing less stress, and reflecting instructor concern for mental health without having negative effects on time management and leading to procrastination. Participants also overwhelmingly preferred flexible deadline policies to a strict deadline policy, citing reduced stress and the ability to deal with unexpected circumstances as their primary reasons.


INTRODUCTION
The COVID-19 pandemic had a tremendous impact on various sectors of the world, including higher education.Colleges and universities had to transition from traditional face-to-face classes to remote ones, and students found themselves continuing their learning amidst adjusting to the abrupt changes and facing various personal circumstances.Much of the student experience during the pandemic was therefore negative, with fear of COVID-19 being correlated with uncertainty, depression, anxiety, and stress for college students [11].Their enjoyment, interest, learning, attention, and effort in classes also decreased after shifting to online learning due to the pandemic [9].Overall, students were facing difficulties both in their quality of life and learning experiences.
One positive change that emerged from the pandemic, though, was course flexibility [9].Instructors understood that both themselves and their students were going through an unprecedented time, and became more flexible with the structure and the deadlines of their courses.However, as higher education now returns to in-person classes, there is uncertainty as to whether institutions should continue providing flexibility to the students or go back to more strict policies with little room for grace.
Many arguments support the notion of providing flexibility to students.It can help students who are faced with unexpected circumstances outside their control and those who have to juggle their coursework with other responsibilities.Knowing that an extension is available to use may also reduce the stress of the students.More indirectly, providing flexibility may send a message that the instructors are understanding of the struggles that students face, which can make CS education more welcoming for the students.
On the other hand, some forms of flexible deadlines could hinder students from succeeding in their classes.Knowing that they can postpone their work may encourage students to procrastinate until they are in a rush to submit poor quality work, and the absence of due dates that give structure to class can lead to students feeling lost and that they are not receiving sufficient guidance from their instructors.Logistically, for some courses, it may also be difficult to implement a flexible deadline policy that is manageable for the instructor and fair for the students while maintaining sufficient academic standards.
The question of flexibility is especially significant for Computer Science (CS) classes, where frequent assignments are given to students to apply what they have learned and receive feedback on concepts that they may not have understood fully.Assignments in CS classes are also unique in that even a small error or typo in one's code can cause the entire program to malfunction, and debugging these errors can take an indeterminate amount of time to locate and fix, a situation where flexible deadlines could greatly help students successfully complete their work and be more likely to achieve course learning outcomes.With all these in mind, it is crucial to assess the benefits and drawbacks of various deadline policies, especially from the perspective of students as they are affected most.In order to help instructors make an informed decision on the course structure and the level of flexibility they wish to provide, this paper seeks to answer the following Research Questions: (1) What are student perspectives on the benefits and drawbacks of various assignment deadline policies in CS education?
(2) Why do students prefer one type of deadline policy over others?

RELATED WORK
In this section, we present prior work on the benefits of flexible deadline policies, the potential drawbacks of flexibility, and how our work differs from what has been done previously.

Benefits of Flexibility
A 2023 Student Voice survey found that most students want some kind of flexibility in their classes, with 72% agreeing that deadlines should be flexible for extenuating circumstances, 45% agreeing that deadlines should generally be flexible, and only 9% agreeing that deadlines should remain firm [5].Withington and Schroeder reported that implementing rolling deadlines that allowed students to choose their own due date from a range of dates lead to a pass rate that is significantly higher than the institutional average [27], while Nickels and Uddin found that providing two late days for submissions led to students learning more from assignments [20].Academic performance is not the only benefit that flexible deadlines can bring to classes.In a 2022 study, students reported that along with producing better quality work, they were able to better manage other academic responsibilities and had benefits related to reduced stress and mental health with flexible deadlines [12].In another study, providing students with just a five-day extension in one of the assignments led to reduced stress and a more positive attitude towards the instructor [23].This decrease in stress was even reported by the majority of students who did not use the extension.Perceived flexibility was also found to be significantly related to the perceived supportiveness of the instructor, and both flexibility and supportiveness of the instructor led to students reaching out more to instructors to discuss their academic struggles [18].Providing flexibility to students therefore can help students feel supported by the instructor and the department, which is an important factor to foster a sense of belonging [22] and can ultimately help in retaining more students [10].
In CS education specifically, a 2005 paper introduced a time bank of two days and received overwhelmingly positive feedback from the students [1].More recently, Vahid et al. allowed resubmission of assignments within 14 days with a 1% deduction per day in an introductory CS class [26].Most students used the policy moderately, and student feedback was overall positive.Another paper that allowed CS students to redo their work and submit late without penalty for formative assignments reported 65% of the students increasing their letter grades by redoing their assignments.[16].

Drawbacks of Flexibility
Despite all the reported benefits, one difficulty involved with implementing flexible deadline policies is setting the extent to which students are given flexibility.Studies that report benefits of flexible deadlines vary widely in their implementations, and a myriad of situational factors including class size, assignment type, course subject, and goals of the course may influence the appropriate level of flexibility needed for the course [12,27].
Furthermore, another factor to consider is that many students rely on deadlines to keep pace with their work, with 51% of students in a 2023 survey agreeing that they depend on deadlines to stay on track and be motivated [5].Procrastination is a challenge for most college students, and around 70% of students identify themselves as procrastinators [28].This is a problem as procrastination has been negatively related to academic performance in a meta-analysis of 33 studies [13].Two of the environmental factors that can promote procrastination are large degree of freedom and long deadlines [25], both of which may arise with flexibility in classrooms.In a study involving two sections of a course with a flexible deadline in one and a weekly deadline in another, only students in the weekly deadline section were able to keep pace with the course materials [19].
In multiple undergraduate CS courses, a binary grading system that assessed assignments as either satisfactory/unsatisfactory also provided opportunities for students to resubmit their assignments [3].Many students appreciated the flexibility but also stated that they needed due dates to not procrastinate and keep pace with the work.In another paper, McCane et al. implemented a mastery learning model where flexibility was provided by allowing students to redo their assignments and tests until they obtained mastery of the concept [17].Some students commented that being able to complete work at their own pace was the best part of the course, but students also complained about the lack of structure.
In a 2022 panel that discussed CS educational ideas that did not turn out as planned, the removal of late assignment penalties led to students falling behind and the instructor experiencing their highest attrition rate [7].In a 2023 panel with the same goals, Stephens-Martinez introduced a policy of soft, hard, and late deadlines [8,24].While some students valued the available flexibility, many students struggled with time management and were confused, while the policy also put extra workload on the instructor and the TAs.Bricker, who incorporated a resubmission policy for students to improve their assignments, also observed 20-35% of students not submitting their assignments on time [8].A similar outcome was observed in another paper that implemented a self-paced introductory CS course with no deadlines, with students falling behind and wanting more structure from the instructors [4].

Current Work
Despite all this important previous research, we are not aware of any work that compares various deadline policies in CS courses through quantitative results and statistical evaluations of student perspectives on flexible deadline policies as we do here.
Most work in this area has focused on a specific implementation of flexibility, and few have evaluated various implementations of deadline policies.In 2006, Becker reported about her experiences with various policies used over seven years in introductory CS classes, including a strict deadline policy, policies with late penalties or early incentives, and a policy with no deadlines, concluding that some degree of flexibility increased performance and that students reported greater satisfaction with the early incentive policy [2].More recently, a 2022 study comparing multiple deadline policies showed that students rated the early incentive policy highest, while early incentive combined with late penalties led to higher grades and earlier submissions [14].
In our paper, we expand this comparison of deadline policies to more specific outcomes including a sense of welcoming and concern for mental health, and explore both the benefits and the drawbacks that are related to flexibility.

METHODOLOGY
In order to answer our Research Questions, we conducted a survey on the perspectives of students on various deadline policies that they experienced in CS courses.This research was granted an exemption by the Institutional Review Board at the second author's institution as protocol #IRB-FY23-24-2.

Recruitment
Data was collected using an anonymous online survey that was distributed nationwide in the United States via institutional partners of AccessComputing, an organization that aims to increase the participation of people with disabilities in computing fields.Partners were asked to forward the recruitment email to all undergraduate students in their department and/or CS classes so that students could opt into the study if they so chose.
Survey data was collected in June and July 2023.Participants were not compensated for completing the survey.

Survey
The survey first asked the participants to select all the deadline policies that they have experienced in their CS courses from the following list: • Strict deadlines with no extensions • Deadlines with limited number or length of extensions, e.g. late days, tokens, etc. • Deadlines that could be extended if the instructor were notified early enough • Soft deadlines that could be extended without notifying the instructor • No deadlines, i.e. everything due at the end of term • Deadlines with an incentive for early submission, e.g.bonus points Except for the strict deadline policy with no extensions, all other policies are implementations of flexible deadlines with varying levels of flexibility.An "other" option was also given to describe any other policies that were not included in the survey.
For each deadline policy that the participants experienced, they were asked how much they agree with six statements on a 5-point Likert scale (1 = Strongly disagree, 5 = Strongly agree).A list of the statements is shown in Table 1.Each statement represents an element of the students' experiences in the class, with many being associated with flexibility in past research (e.g.[4,8,12,20,23]).

Statements
The deadline policy made the class feel more welcoming.The deadline policy showed that my instructor cared about my mental health.The deadline policy was stressful for me.The deadline policy helped my learning in the class.The deadline policy helped me to not procrastinate.The deadline policy helped me manage my time more effectively.
Students were also asked how often they used an extension for each flexible deadline policy on a 5-point Likert scale (1 = Never, 5 = Always).
Lastly, participants had the option to indicate their most preferred deadline policy in CS classes and the reasons for their choice.

Analysis
Each participant only provided responses for the deadline policies that they have experienced.Therefore, a linear mixed model was used rather than repeated measures ANOVA to better accommodate missing data [15].This model involves both a fixed effect (variables that you are interested in) and a random effect (variability that you want to account for).For our study, the fixed effect was the type of different deadline policies and the random effect was each individual to account for variability caused by individual differences.
For each statement, a linear mixed model with post hoc analysis was used to determine whether there were significant differences between the various deadline policies.Then, for each deadline policy, we also tested whether there is a correlation between the usage of extensions and the statements.

RESULTS
This section summarizes the comparison of different deadline policies for various elements and the participants' most preferred policy in CS classes.A total of 56 students participated in the study, with the following demographics.

Comparison of Deadline Policies
Table 2 shows the number of participants who reported having experience with each type of deadline policy in CS classes.Out of the 56 respondents, 84% indicated that they had experience with more than one type of policy.While 46% of the participants stated that they experienced a strict deadline policy, no participant had the strict deadline policy as the only deadline policy that they experienced in their CS classes.Due to the small number of participants experiencing a policy with no deadlines (n = 3) or a deadline policy with an early incentive (n = 9), we concluded that the numbers from these two policies were not representative and these policies were not included in our analysis.Table 3 shows the mean rating and the standard deviation for our elements of interest for each type of deadline policy.
4.1.1Welcoming environment.All three deadline policies that provided some sort of flexibility (limited extension, extension with notification, and soft deadline that allowed extension without notification) were perceived to be significantly more welcoming than the strict deadline policy (p < 0.001).Both extension with notification (p = 0.003) and soft deadline that allowed extension without notification (p = 0.004) were also perceived to be significantly more welcoming than the limited extension policy.

Mental health.
A similar result was observed for the instructors' perceived concern for mental health.All three flexible deadline policies were significantly higher than the strict deadline policy (p < 0.001), and both extension with notification (p < 0.001) and soft deadline allowing extension without notification (p = 0.004) had significantly higher perceived concern for students' mental health than the limited extension policy.

Stress.
All three flexible deadline policies were perceived to cause significantly lower stress than the strict deadline policy (p < 0.001).The only other significant comparison was the soft deadline policy that allowed extension without notification being perceived to cause significantly lower stress than limited extension policy (p = 0.003).

Learning.
For perceived learning, the policy that granted an extension with notification was significantly higher than the strict deadline policy (p = 0.007), but no other comparisons were significant.

Time management & procrastination.
We found no significant differences among the four policies in terms of helping with effective time management or helping students not procrastinate.

Usage of policies.
Lastly, for all four deadline policies, the frequency of usage of extensions had no significant correlation with any of the elements discussed above.

Most Preferred Policy
A total of 38 participants answered the most preferred policy question, and Table 4 shows the number of participants who preferred each type of policy.The two "other" responses included a combination of deadline policies, with one combining limited extensions and soft deadlines and the other combining limited extensions with an early incentive.
No students selected strict deadlines or no deadlines as their favorite policy.Instead, most participants preferred a policy that provided both structure and flexibility, including limited extension, extension with notification, and soft deadlines allowing extension without notification.For example, students who preferred these policies stated that: "Having a deadline helps me manage my time and not procrastinate, but knowing that I am able to turn a certain amount of projects in a few days late was helpful so I was less stressed about getting things done and working correctly, which was very helpful in computer science especially.""I do want deadlines but I also like the option of asking for an extension if it becomes too stressful for me, if there were no deadlines, I think half of the students would not turn in the work.""Deadlines that could all be extended if notified early enough is strict enough to stop me from procrastinating but soft enough that I can ask for an extension if anything comes up." "Having deadlines but having the ability to extend them if something surprising comes up provides flexibility for unplanned conflicts as well as for assignments that turn out to be more difficult than expected."Out of these policies that provided both flexibility and structure, some participants specifically preferred a deadline policy with limited extensions as they believed more lenient policies would lead to procrastination.
"Soft deadlines almost always lead to me just doing everything later... Having a small number of no questions asked extensions preserves my privacy, adds flexibility if something pressing comes up, and maintains the structure of the class.""Life happens throughout the semester, sometimes last-minute, so I appreciate having the flexibility to use an extension as needed... Classes with too many extensions / flexible deadline policies give me too much stress because I end up procrastinating."Conversely, some did not prefer a policy with limited extensions but instead opted for more lenient policies as they believed limited extensions do not account for every situation.
"Limited late days can't account for certain situations.Extendable deadlines can account for most situations and are stricter than soft deadlines, which are a bit too easy in my opinion.""It's not always reasonable or fair to put a set number of extensions."Some students also preferred having extensions without being required to notify the instructor, stating that asking for extensions can itself be a stressful task.
"I also like not having to notify the instructor because that can feel really stressful and I get scared to ask so I end up turning in half completed work instead of asking for an extension.""Soft deadline policy gives me the most peace of mind...It frees me from the stress of having to email my professor and explain why another situation is more important than finishing a problem set.I appreciate the flexibility of this policy and the classes I've taken with this policy have been historically less stressful." Lastly, some students preferred having an early incentive policy rather than extensions as it motivates them to finish work on time.
"For me, deadlines that allow extension actually may encourage procrastination, which messes up with my overall schedule.""The incentive helps motivate starting early."

DISCUSSION
In our analysis, we compared students' perpsectives on the first four policies in Table 2. Except the strict deadline policy, the other three policies represent a type of flexible deadline policy.

Stress, Mental Health & Welcoming
There were clear patterns for these three elements.All three flexible deadline policies were perceived to be more welcoming, more reflective of instructor concern for students' mental health, and less likely to cause stress compared to the strict deadline policy.This was further supported by 86% of participants choosing one of these three deadline policies as their most preferred policy, citing both the ability to deal with unexpected circumstances and reduced stress as reasons for their choices.
There also seems to be a possible relation between the level of flexibility and these elements, with the two more lenient implementations of flexibility (extending deadlines with notification or soft deadlines that allowed extension without notification) being perceived as better in terms of concern for mental health and being welcoming than the limited extensions policy.Having a limited number of extensions would mean that students have to choose the assignments on which they would use the extensions, which could feel stressful and more restricted than the other two that allow extensions to be provided for all assignments.The responses from the most preferred policy also showed that some students find the process of asking for the extension itself stressful, which could partially be the cause for our results.

Learning
The policy with deadlines that could be extended with notification was the only policy that was perceived to be better for learning than the strict deadline policy.We believe that this form of flexible deadline may have provided sufficient structure to keep pace with the material while also giving enough flexibility to take more time to learn the material thoroughly if needed.
The two other flexible deadline policies also had higher (though not significantly so) scores for learning, which showed that at the very least flexible deadlines do not diminish students' perceived learning.Perhaps a form of flexibility that would better target learning would be resubmission, which allows the students to receive feedback and improve from their mistakes and is used in models such as Mastery Learning [4,17] or the more recent paradigm of "A's for All" [6].

Procrastination & Time Management
In contrast to previous papers that observed procrastination and students falling behind due to flexibility [3,4,19], the three flexible deadline policies in our analysis did not differ in terms of perceived help with procrastination or improved time management compared to the strict deadline policy and each other.This was perhaps because the three implementations of flexible deadlines were not extreme in terms of leniency and still provided standard due dates by which students were expected to submit their assignments.The qualitative responses of the participants also showed that deadlines present in these flexible policies help with procrastination.However, our paper measured perceived procrastination and time management from students' perspective while the previous papers reported observed procrastination, which may also have been the reason for the differences in results.

Summary
Overall, the results showed that it is important for deadline policies to have both flexibility and structure.However, the more specific implementation will depend on the focus of the instructor.Having more flexibility seem to have benefits of being more welcoming and allowing students to have more leeway for situations outside their control.On the other hand, some students seem to not want too much leniency as it can cause procrastination.

RECOMMENDATIONS
The purpose of this work is to help instructors to make an informed decision on their course assignment deadline policies, and our results show that CS classes in general should adopt a policy that provides both flexibility and structure.Students perceived no benefits from the strict deadline policy, while the three flexible policies were perceived to be more welcoming, cause less stress, show concern for students' mental health, and perhaps even improve learning without compromising students in procrastination and time management.The perceived benefits were even present for participants who did not take advantage of the flexibility, similar to a study where the majority of participants reported reduced stress even when they did not use the available extension [23].
Shown in this work and previous studies, there is a myriad of flexible deadline policies available for instructors that perceive them to be worthwhile.An instructor who wishes to start providing flexibility could start with a time bank that provides a fixed number of grace days to students.This kind of limited extension policy does not require much effort to implement [1] and also requires less adjustment from the students who are used to having set deadlines with no extensions.
On the other hand, we did observe more pronounced benefits of flexibility as the policies became more lenient, so an instructor whose goal is to maximize the positive experiences of students may utilize deadlines that could all be extended with or without notifying the instructor.If an instructor is worried that too much flexibility may promote procrastination for some students, perhaps incorporating an incentive for earlier submissions could motivate the students to start their assignments earlier.
Regardless of the implementation, we believe it is also important that instructors clearly communicate the level of flexibility they wish to provide and the reasons for doing so.It would also be beneficial to remind the students of the available flexibility throughout the term.A 2020 study has shown a difference in the perception of students and instructors, with instructors willing to be more lenient than students believe them to be [21].This clear communication between the instructor and the students will emphasize to the students that they are welcome to utilize the offered flexibility and make the process of asking for extensions less stressful.

LIMITATIONS AND FUTURE WORK
As with all research, there are limitations to the study presented in this paper.The list of deadline policies that we had in our survey was by no means comprehensive of all deadline policies that are present in CS education, including a resubmission policy that is widely used in CS education We provided an "other" option for the participants as a way to tackle this issue, but it was sparsely used.We also were not able to include policies with no deadlines or with early incentives in our analysis due to a small fraction of participants who experienced them.However, we believe that these policies are less frequently used in CS classes so it would be hard to make generalizations about them, though future work could explore these policies specifically.
Similarly, there could have been other elements that are influenced by the flexibility that we did not consider, shedding light on even more benefits or potential drawbacks of these flexible deadline policies.Therefore, future research could investigate both a wider range of deadline policies and other metrics that may be impacted by these policies.
While students' perspectives are important in assessing these deadline policies, another crucial factor to consider is the perspective of the instructors.In the end, implementing flexible deadlines puts more responsibilities on the instructors, graders, and TAs, and the efforts required to provide flexibility along with the goals of the instructor for the course could be assessed in the future for a more complete evaluation of flexible deadline policies in CS education.
The survey itself also has limitations, as students may have changed their perception over time and the same policy may have had subtle differences depending on the instructor and the course that we might not have captured.Although we put focus on the deadline policies by framing our statements in the form of The deadline policy made/showed/helped..., future research could measure the outcome variables before and after implementing a policy to empirically evaluate the impact of the policy.
Future work could consider the impact of different policies on different levels of courses, as well as variations in perspectives based on students' demographic information, e.g. to see whether certain policies are considered more welcoming to certain demographics.This could help instructors target their deadline policies based on the general needs and makeup of their classes.Our survey did ask the participants for the demographic information, but due to the limited size, the numbers were not representative and could not be used for analysis.

CONCLUSION
After assessing the perspectives of CS students on various assignment deadline policies and their most preferred policy, we have observed that policies that provide both flexibility and structure to students are more beneficial compared to a strict deadline policy in terms of being more welcoming, causing less stress, and showing a concern for mental health without having negative effects on time management and procrastination.We hope that the results of this paper can encourage more instructors to make informed choices on their course flexibility and structure.

Table 2 :
Number of participants that experienced each deadline policy (n=56)

Table 3 :
Extent to which participants agreed with elements of interest for various deadline policies (n = 56)No deadline and early incentive policies were not included in the analysis.Statistical analysis of post hoc Holm-Bonferroni test vs Strict Deadline (** = p < 0.001, * = p < 0.05).

Table 4 :
Most preferred policy of the participants (n = 38)