Pauses and spacing in learning to program

Abstract

Conventional wisdom holds that time is an integral part of the learning process. Spacing out learning over multiple study sessions seems to be better for learning than having a single longer study session. Learners should also take pauses from the learning process to absorb, assimilate, and analyze what they have just learned. At the same time, pausing too often can be harmful for learning. Participants of two subsequent introductory programming courses completed programming tasks in an integrated development environment that saved detailed logs of their actions, including time stamps of all the participants' keypresses in said environment. Using this data with background variables and a self-regulation metric questionnaire, we study how the students space out their work, identify trends in between the kinds of pauses the participants took and the course outcomes, and their connection to background variables. Based on our research, students tend to space out their work, working on multiple days each week. In addition, a high relative amount of pauses of only a few seconds correlated positively with exam scores, while a high relative amount of pauses of a few minutes correlated negatively with exam scores. Student pausing behaviors are poorly explained by traditional self-regulation measures such as the Motivated Strategies for Learning Questionnaire and other background variables.

Comparison of Time Metrics in Programming

“…Our belief is that this is partially due to focusing on too coarse grained data -it is likely that different students spend different proportions of time in different learning environments and material locations. Thus, effort should not be estimated solely based on the total study time, as even small pauses can account for large variations in learning outcomes [21]. Moreover, it is also important to know what the time is spent on -for example, knowledge on which material paragraphs the students spend their time on could provide additional insights on the students' struggles [22].…”

“…Factors affecting students' performance in programming courses have traditionally been studied with the purpose of being able to predict the students' performance. In this line of research, a myriad of predictors ranging from students' affective states [31], students' programming behavior [8,10,12,13,21,27,38] to complex programming process based metrics [5,20,30] have been constructed.…”

Incentivized Spacing and Gender in Computer Science Education

“…They also found that students who spaced more provided more constructive feedback to others and were more likely to answer correctly on their first attempt [50]. In contrast, Leppänen et al [28] found that students who spaced their practice of programming tasks achieved lower grades than those who did not. Interestingly, they show that short pauses (one to two seconds) while performing programming tasks correlated with higher exam grades, while longer pauses (ten seconds to five minutes) correlated with lower exam grades [28].…”

“…Interestingly, they show that short pauses (one to two seconds) while performing programming tasks correlated with higher exam grades, while longer pauses (ten seconds to five minutes) correlated with lower exam grades [28]. They also collected data from a survey to analyze students' perceptions of their studying and found that their reported self-regulation did not correlate with their spacing habits [28].…”

Predicting Academic Success Based on Learning Material Usage

“…Other studies have identified a multitude of other predictors. Leppänen et al [26] found correlations between the types of pauses students take from writing program code and exam scores in a CS1 course. Leinonen et al [25] were able to predict student learning outcomes based on the key press latencies in their typing.…”