Student-Mediated Video Lecture Capture: The SUNYIT Experience With ClassX Ronald Sarner and Rosemary Mullick SUNY Institute of Technology United States ron@sunyit.edu rosemary@sunyit.edu Abstract: Video capture of lectures and other classroom sessions can now be done easily and inexpensively. Recent software development enables the viewer to dynamically zoom into portions of the captured image such as the board or the screen. This paper describes the experience at SUNY Institute of Technology in setting up, deploying, and modifying the ClassX system for capturing undergraduate and graduate classes in the Department of Computer and Information Sciences. Findings include widespread use by students, most frequently to view class sessions that they had missed attending. Background: For several academic years a few of us in SUNYIT's Department of Computer and Information Sciences had been video capturing one or more of our undergraduate classes using very inexpensive consumer-grade equipment. A camcorder was brought into the classroom, and mounted on a tripod. The camcorder was operated by a graduate assistant (where and when available), or by a class member who responded to the offer of a few extra credit points in the course. After each class the video files were downloaded to the instructor's PC, transcoded into a low-resolution (to reduce file size) flash file, and subsequently uploaded to the department's UNIX server. While there was substantial evidence that the resulting videos were being used by students there were several issues with this process: (1) the transcoding process used to shrink the files had a significant effect upon the quality of the image, (2) the transcoding and multiple file movements was a labor-intensive manual process, and (3) the image that was recorded was that portion of the classroom that the camera operator, not the viewer, thought was most important. In Spring 2012 the writers became aware of ClassX, an open-source video lecture capture system developed at Stanford University. The description of the ClassX system strongly suggested that the developers had largely solved the problems that were encountered in the local system. One set of features of the ClassX system, the fact that the entire front of the classroom was captured and that the viewer could zoom in on any area of interest, and could change the area of interest at will, was particularly appealing. The authors wrote an application for a 2012-13 SUNY (Innovative Instructional Technology Grant) IITG grant and were awarded $10,000. This paper describes our experience with the ClassX system. Why Video Capture?: To some video capture of classroom sessions, while clearly technically possible, does not represent the forefront of innovation in education. Lectures are inherently a passive learning activity. Critics could argue that the resources devoted to video capture might best be directed in ways that encourage more active learning. However there are several reasonable counter arguments. SUNYIT, like almost all higher education institutions, has large numbers of faculty who are disinterested in developing on-line courses or in employing alternative instructional methods. Video caputre may be a means to enhance the effectiveness of this group of faculty. From a faculty standpoint video capture does not require the any alteration to how the course is structured or presented. The equipment is unobtrusive, and the faculty member can carry on as though it was not present at all. Use by students was expected for a variety of reasons. We expected that some would use the videos to review material that they simply did not understand the first time it was presented, or where their concentration drifted off. Others might use it to review for tests. Where illness or time conflicts resulted in missing a class the videos would likely be much more reliable than the notes taken by a classmate. We could also envision rare circumstances where a student would use the videos as a complete substitution for class attendance. A legitimate example of the latter would include a student who needed a course that was not being offered in the current semester, but videos of the course existed for a prior semester - in effect a directed 1 study. One goal of this project was to determine how students used videos, and to what effect? Did viewers, for example, achieve higher course grades than non-viewers? Is there any relationship between video viewing and course performance? The SUNYIT Experience: Funding from the IITG grant became available in late summer 2012. The first challenge was to procure the necessary equipment within a very tight time frame. Commitments were received from four faculty members in the Department of Computer and Information Sciences to video capture classes and a total of ten sections (nine undergraduate and one graduate) were captured during the Fall 2012 semester. To accomplish this three camcorders meeting the recommendations of the Stanford ClassX developers (about $500 each) and tripods were purchased, along with a number of 32 gb SDHC flash cards. These were stored in the department office between uses and were accessible to all participating faculty. A surplus PC was reconfigured as a temporary ClassX server, but it quickly became apparent that it was inadequate even on an interim basis. On the basis of tests conducted prior to the start of classes it was determined that the internal camera microphone was satisfactory and consequently a wireless lavalier microphone could be dispensed with, simplifying the setup at the beginning of each class and further reducing the per unit equipment cost. Specifications were drawn up and parts ordered to build a server at a cost of about $3000, but this was a slow process and the server was not configured and available until about the last month of the semester. Student assistants reworked some of the Stanford-developed code to put an authentication system in place. While rudimentary, it ensured that only users with SUNYIT credentials (login and password) had access to the videos. There were two substantial obstacles encountered in the Fall semester - process and backlog. Once a class session was finished the instructor removed the flash card from the camcorder and gave it to the graduate assistant assigned to the project. The graduate assistant uploaded the files to the server and initiated the transcoding process, but delays were substantial (sometimes exceeding one week) and unpredictable. Delays were frustrating to both students and participating faculty. Once a card was uploaded it was reformatted and put in a place where faculty could grab a card for their next class. However, there were times when faculty did not have access to this room (a real issue for 8 AM classes), and other times when the backlog was so great that no blank cards were available. As a result a small, but significant, number of class sessions were not captured. Frustration with the backlog and lack of availability of flash cards was the motivation for the development of a self-serve kiosk, with the programming done by undergraduate student assistants. The kiosk was put into service at the beginning of the spring semester and was improved as the semester progressed. In the last weeks of the spring semester each class section being captured was assigned an individual flash card. A signature file on the card encoded the course name and number as well as the lecture number. After the class the faculty member removed the flash card and inserted it into the kiosk that was conveniently placed next to the departmental copy machine in a room for which all departmental faculty had keys. The kiosk has three flash card readers, and so can upload three classes at a time. In general, the uploading process takes about fifteen to twenty minutes for a two hour class. Upon completion of the uploading, the kiosk script increments the lecture number in the signature file of the flash card and also deletes the uploaded video files from the flash card. The instructor is sent an automated e-mail that the uploading has completed and can retrieve the flash card - ready for the next class session. The e-mail also reminds the instructor to log on to the ClassX server to initiate the transcoding process. At the completion of that process the video is automatically accessible to the SUNYIT community. These refinements largely eliminated the problems that had been encountered earlier. During the Fall semester a presentation on the ClassX project was made to SUNYIT faculty at a Provost's lecture. During the Spring 2013 semester one additional faculty member in the Department of Computer and Information Sciences began capturing classes; thus five of the department's twelve full-time faculty were participating. In the Spring semester a total of thirteen class sections were captured. Costs and Benefits: To suggest that the ClassX system is cost-effective is an understatement. Initial expenses were approximately $600 for each 2 setup consisting of a camcorder, tripod, and a number of 32 gb SDHC flash cards. The cost of the server was approximately $3000, and while it will likely require additional storage space in the future, storage is now in the range of $50 per terabyte. Ongoing labor costs are minimal. Set up and take down of equipment is handled by the faculty member teaching the course. Even if student usage is not extensive the cost is so low and the set up so unobtrusive that it is clearly worthwhile. While usage within the Department of Computer and Information Sciences is widespread, concerns have been raised by faculty elsewhere on campus. One concern is intellectual property rights, and the concern is multifaceted. On the SUNYIT campus the video lecture capture process only minimally engages the use of staff resources, and as such the faculty member retains the IP rights. However, it is also clear that there are limits to the restrictions that the faculty member can place on access. As a result of prior campus discussions, the case of Russo v. Nassau County Community College (see http://www.law.cornell.edu/nyctap/I93_0189.htm) is well known by content, if not by name, by many SUNYIT faculty. From the Russo case it is absolutely clear that whatever transpires in a public college classroom is subject to the State's Freedom of Information Law. In the Russo case the petitioner sought access to a film shown in a Human Sexuality course. Respondent, Nassau Community College, raised several arguments to deny petitioner access, including that the film was subject to copyright and the respondent did not own the intellectual property rights. The New York State Court of Appeals rejected respondent's arguments holding that the materials in question fall within the scope of the Freedom of Information Law, and that the fact that the classroom is a deliberative environment "...does not alter the status of items used in the classroom." One parallel here is that the faculty member, not the college, retains the IP rights to the captured lecture, so as was the situation in the Russo case, the college can not deflect a FOIL request on the basis that it does not own the intellectual property. Several SUNYIT faculty have expressed reservations about engaging in video capture specifically because of possibility of a FOIL disclosure. Another concern that is raised is that the privacy of student class participation is breached. Students who do not wish to appear on a video can easily avoid it by sitting in the classroom behind the camcorder. However, faculty have voiced concerns about the chilling effect of an audio recording of any comments students may make during a class. The only way that students can avoid an audio recording is to refrain from speaking and that is a chilling effect. At SUNYIT we have sought to balance these interests by restricting access to videos to persons holding valid SUNYIT credentials. Some have suggested that access to the videos be restricted to a link in the campus LMS (for SUNYIT it is Angel). While it is certainly possible to have Angel point to the ClassX server, we have not yet developed the tools for ClassX to use Angel authentication. So far we have two options - disable authentication on the ClassX server, in which case any person with knowledge of a session URL can bypass any authentication, or enable authentication on the ClassX server, in which case a user authenticated by Angel must be re-authenticated. Once authenticated into the ClassX server the user has access to all ClassX videos, not only those for the class in which the viewer is enrolled. Research Design: A pre-post non-experimental design using intact groups was employed. A total of 19 classes were included in the study, nine in Fall 2012 and ten in Spring 2013. At the beginning of each semester students were given a brief pre-test survey to tap their attitudes about the possibility of accessing video-captured lectures, a projection of how likely they would be to use them, and an assessment of the reasons they would use video-captured lectures. Participation in the study was totally voluntary so not all students enrolled in the classes elected to participate in the survey. At the end of the respective semesters a post-test questionnaire was administered to all students in 18 of the 19 classes. The post-test contained the same questions with one additional item asking how likely they would be to recommend the use of video-captured lectures to other students. Findings: A total of 301 students participated in the study, 127 in Fall 2012 and 174 in Spring 2013. The major questions on both surveys were in the form of a rating scale from 1 through 9. Question 1 determined how helpful the video-caputred lectures would be, Question 2 asked how often they thought they would access the videos, and Question 3 asked how likely they thought students would use the lectures. On the post-test Question 4 determined how likely it was that the student would recommend use of the videos to peers. The value of 1 represented not at all helpful, never accessed, highly unlikely, and the value of 9 represented very helpful, access after every class, highly likely. The mean, median, and mode was calculated for each of the questions for both the pre and post tests. 3 Q1 Q2 Q3 PostQ1 PostQ2 PostQ3 PostQ4 N 110 111 110 53 54 52 52 Mean 8.04 6.86 6.99 7.00 4.19 6.67 6.96 Median 7.00 7.00 6.00 7.00 5.00 7.00 7.00 7 5 7 7 Mode 7 7 7 Table 1: Descriptives for Pre and Post Questions Fall 2012 Q1 Q2 Q3 PostQ1 PostQ2 PostQ3 PostQ4 N 140 140 140 98 98 98 98 Mean 7.23 6.44 6.66 7.08 5.14 6.30 6.77 Median 7.00 6.00 7.00 7.00 5.00 7.00 7.00 7 5 7 7 Mode 7 7 7 Table 2: Descriptives for Pre and Post Questions Spring 2013 Tables 1 and 2 clearly indicate that students perceived that the helpfulness to themselves and the usefulness to others was high. The mean values for Q1 and Q3 on both the pre and post test over the two semesters ranged between 6.66 and 8.04 on a 9 point scale. In addition students indicated that they would be likely to recommend the use of the videos to other students, mean = 6.96 and 6.97 for Fall 2012 and Spring 2013 respectively. The paired t-test statistic was computed on the pre-post test items for each semester. The results of these analyses are depicted in Tables 3 and 4 for the Fall 2012 and Spring 2013 semesters respectively and clearly indicate that there were no significant changes from pre to post evaluations of Questions 1 and 3. The t value of 2.06, p= .045 on Q1 approaches significance for Fall 2012 and the t of .957, p=.342, the t-value of .651, p=.519, and the t=.434, p=.665 for Spring 2013 indicate that students did not change their perception of helpfulness of the videos for themselves, or the usefulness to other students. Table 3: Paired Samples t-test Fall 2012 4 Table 4: Paired Samples t-test Spring 2013 When examining the pre-post values for the question, How often are you likely to view the videos, significant results were found in both semesters. A t=9.702, p=.000 on Table 3 and a t=6.468, p=.000 on Table 4 show that students tended to estimate their likelihood of viewing lectures higher on the pretest, mean=7.00, than they actually reported viewing on the post-test, mean=4.08. One item on both the pre and post test asked students to indicate reasons for using video-captured lectures. They included: 1- to make up a class I missed 2- to clarify difficult concepts 3- to review for quizzes and/or exams 4- to review examples of topics. In addition there was an open-ended question that asked them to list other reasons. The most frequent use indicated by students was to make up a class I missed, followed by the choice to clarify difficult concepts. A fairly large number of students also reported use for review for quizzes and/or exams. The same uses were reported for the post-test although the number who chose these options was smaller than on the pre-test. The same pattern of results was found in both semesters. The ClassX server automatically collects information on use of the system, but because of the mixed use of a temporary server and a permanent server in Fall 2012, no actual use data is available for that semester. Technical difficulties precluded data collection early in the Spring 2013 semester; data is available only for about ten or eleven weeks of that semester. Over that period a total of 143 of the 174 unique enrolled students in sections (a student enrolled in multiple ClassX courses is only counted as a single student) captured using ClassX accessed one or more videos (an 82% participation rate). There were a total of 1220 logins, so the mean usage was 8.5 viewings per actual user. Total viewing time was 349.61 hours, an average of just under 2 1/2 hours per user. Viewings ranged from 26 for the least frequently accessed course to 270 for the most frequently accessed course. This paper was written less than one week after the end of the Spring 2013 semester. There has not been adequate time to do more than rudimentary descriptive statistical analysis. In the weeks ahead additional analyses will be performed. Student performance using class grades and overall GPA will be compared for students who partiipated in the study and those who did not. A more thorough examination of the system data should shed some light on the relationship between student use of the videos and course performance. The analysis to date suggests that students are highly positive about the availability of video-captured lectures. They tended to report having video-captured material as helpful to them and potentially useful for other students. They also indicated they would be highly likely to recommend the use of the video-captured material to others. The results also indicated that students tended to overestimate their likelihood of viewing the materials compared to their report of how frequently they actually accessed them. 5