Data Sharing (DaSh) for Collaborative Learning in Laboratories

Untitled Document

Data sharing (DaSh) for collaborative learning in laboratories

Abstract

In a set of independent student focus groups conducted at Leeds University by the Royal Academy of Engineering, a common concern regarding laboratory sessions was that an emphasis on assessment often meant that, due to time pressures, students concentrated on gathering data rather than understanding the underlying engineering principles.

In 2008, with support from the Higher Education Academy Engineering Subject Centre, the DaSh system was developed to allow students to upload and share data, in order to promote collaborative and deeper learning. Data could be uploaded in real-time, via the internet, allowing results to be made available immediately for comparison with their peers’, whilst still in the laboratory in front of the equipment.

This facilitated learning in two distinct ways. At a basic level, students were able to rapidly spot and correct measurement or calculation errors. At a more advanced level, conclusions could be drawn from the observation of hundreds of data points (rather than just five or ten), something that would otherwise have been impossible, had they worked independently.
Following its positive impact on the student learning experience, data sharing was incorporated into a ReLOAD (Real Labs Operated At Distance) session, allowing the experiment to be repeated anywhere with internet access.

DaSh was evaluated by comparing performance on similar sessions run with and without data sharing, observing behaviour within the laboratory, completion of questionnaires after the sessions and from focus groups. Through these methods, results have shown that the system has worked well, has been popular with the students and has been beneficial to their learning.

Introduction

Laboratory based learning is an important aspect of many engineering programmes. Laboratory sessions give hands on experience teaching practical skills and allow theoretical principles to be put into practice. They are, however, resource-intensive in terms of laboratory space, equipment required, and support from technicians and demonstrators. Moreover, the time spent by academic staff assessing laboratory sessions, often through marking laboratory reports, can result in feedback being returned weeks after the laboratory sessions were undertaken. In addition there is limited time (due to the pressures on space and equipment) in which students can complete experiments which often results in them feeling they need to concentrate on gathering data rather than thinking about the underlying engineering principles (Hanson et al., 2008). More meaningful learning would take place if students had the opportunity to interact and reflect on what they are doing (Gunstone and Champagne, 1990).

Building on successful work already carried out as part of the ReLOAD-SAFE system (Gallagher et al., 2008), a new data sharing learning technology system (DaSh) was developed in 2008. DaSh aimed to maximise potential learning opportunities within laboratory based sessions by rapidly sharing data in a real-time environment, promoting face-to-face dialogue among peers and hence encouraging the interaction necessary for collaboration to occur (Dillenbourg, 1999). As Salmon (2002) describes, ‘the best experience of collaboration by participants for learning purposes enables them to experience both personal, individualistic, useful learning whilst contributing to a community of learners and the support and development of others.’

DaSh was developed to allow students, at a basic level, to spot any simple errors, such as insufficiently accurate measurement or submitting results using incorrect SI units and, at a more advanced level, to allow them to learn from their peers mistakes or from more capable peers. Vygotsky (1978) describes the difference between what you are able to learn individually to that of collaborating with more capable peers as the Zone of Proximal Development (ZPD). Throughout the session the students work towards a common goal (Lewis, 1997) to compile a graphical description of how the transient dynamic characteristics of systems change as different parameters are varied. Students are able to work in pairs submitting their results via a PC and viewing them on a large collaborative graph displayed elsewhere in the laboratory.

The DaSh system has the additional benefit of allowing the tutor to quickly and easily identify any issues or concerns that arise with both individuals and laboratory class groups during the session. These can then be addressed by the tutor and any necessary feedback or support given immediately during the laboratory session (Cortez, 2009).

Methodology

As part of a core 20 credit level 2 module “Vibration and Control”, 121 students participate in two laboratory sessions. The first laboratory session concentrates on a cantilever beam experiment, in which students measure the frequency of vibration of the beam and then examine how its dynamic characteristics change as mass is added to its free end. The second session focuses on an electro-mechanical position servo mechanism. Again students vary parameters, this time the forward path gain and velocity feedback gain, and observe how these affect its transient dynamic performance. During both experiments the students work in pairs to measure and collect data for a range of parameters. In both sessions, the data collected allows them to calculate the roots of the characteristic equation which can be plotted on a root locus or S-Plane diagram. These results can be compared to mathematical model predictions that the students are asked to make.

During the first laboratory session students were able to submit their data using a personal response system (PRS), based on the work of Boyle and Nicol (2003) who used the system to give ‘almost instantaneous feedback’. The submitted data was available to the tutor only, allowing for quick collation of results and marking of the students’ work from several identical sessions. This allowed feedback relating to common errors to be returned in a more timely manner during the following week’s lecture. This would not have been possible with a traditional laboratory report format, which takes considerably more time to assess. Attendance at the lecture in which feedback was given was also recorded.

Students were then asked to perform a very similar measurement exercise using a large cantilever beam that could be accessed via the ReLOAD system, a system for completing laboratory experiments remotely via the internet (Weightman et al., 2007). Damping levels applied to the beam were different for each student and not known to the student beforehand, ensuring each student’s data was different to that of their peers. The ReLOAD experiment was used to assess how effectively students had learned from the feedback given. Students were allowed to repeat the experiment as often as they wished but were only allowed a single submission. They were prevented from seeing the group’s data until they had submitted their own data and the scale on the axis was deliberately omitted in an attempt to prevent students from trying to guess responses.

In the second laboratory session students were able to submit their data in real-time, via a web form to a database. From this data two web pages were generated, each containing a visual representation of their data:

An S-plane diagram showing data submitted by a particular pair of students, allowing only them to see only their data, on their computer screen, following a request by them.
An S-plane diagram, displayed using a data projector, showing all data submitted by students in their group conducting the experiment, allowing staff and all students to view an image of the whole data set. This page was refreshed automatically allowing students to see the group’s data as it emerged.

As in the first laboratory session, following the face-to-face laboratory work, students were asked to complete an individual, remote laboratory experiment using similar equipment attached to the ReLOAD system. The students were able to repeat the experiment as many times as they wished and, unlike the first laboratory session, they were able to compare their data with that of their peers and were allowed to resubmit data if they subsequently felt they had made any errors.

During both laboratory sessions videos were recorded. This was to enable the level of interaction and behaviour of individual students, pairs of students and the laboratory group to be identified. In addition, during the second laboratory session the video would show how many students viewed the cumulative graph and any interactions that occurred while viewing the graph. The videos could then be analysed to identify any differences between laboratory sessions one and two.

Immediately following the completion of both the first and second laboratory sessions students were asked to provide feedback and comments. They were asked to complete the same questionnaire after both laboratory sessions, on their experiences of working:

individually,
in pairs and
as part of a larger group during the session.

Responses were collected using the PRS allowing for anonymity and quick collation of overall percentage responses from all of the laboratory groups. The results were later analysed to see if there were any differences in response between the first and second laboratory sessions.

In addition to the questionnaires, students were asked on a voluntary basis to participate in independently-run focus groups. The focus groups allowed students to give more general feedback and comments about the two laboratory sessions and to discuss in more detail the responses from the questionnaires.

Experiment design

In the first laboratory experiment, students (working in pairs) deflect the free end of a steel cantilever beam and then, using strain gauges and a computer data acquisition system, record the free vibration over a period of time. Students come into the laboratory in groups of between 20 and 28 and each pair of students work on their own piece of equipment. They measure the damping ratio using the log decrement method (since the beams are very lightly damped) and measure the damped natural frequency. From this data they can calculate the undamped natural frequency and hence calculate the roots of the characteristic equation for the beam. They are then encouraged to add mass to the end of the beam using magnets and observe how this affects damping, natural frequency and most importantly, the roots of the characteristic equation. They plot the position of the roots of the characteristic equation on a root locus diagram using a paper proforma given out during the laboratory. The amount of mass added and the number of experiments completed is left up to the students. By weighing the added mass, students are able to calculate the vibrating mass and stiffness of the beam from its change in natural frequency only and are asked to compare this to theoretical values. They are also asked to consider if the roots of the characteristic equation have moved as expected when mass is added.

Following the face-to-face laboratory, students are then required to perform two short experiments using the two ReLOAD enabled large cantilever beams shown in Figure 1. The beams are identical, except that one has a known additional mass attached at its free end. Students are again asked to calculate the roots of the characteristic equation and submit these for assessment. Once submitted they are able to see an electronically generated root locus diagram showing their data compared to their peers’.

Figure 1.Cantilever beams, one with added mass at its free end

Figure 1.Cantilever beams, one with added mass at its free end

In the second laboratory session, students investigate the transient dynamic performance of the electromechanical position servomechanism (shown in Figure 2), and how this changes as the forward path gain and velocity feedback gain are varied. A step input is applied to the system and the response is analysed using the same data acquisition systems as used previously. Students calculate frequency of oscillation and damping by measuring the position and magnitude of the peak overshoot. Again they can calculate the roots of the characteristic equation.

Figure 2. Photograph and system diagram of the position servomechanism

Figure 2. Photograph and system diagram of the position servomechanism

Prior to the session students were given a personalised theoretic investigation to complete, based on the system shown in Figure 2. They were asked to calculate what would happen to the roots of their particular system as the forward path gain and velocity feedback gain were varied.

Following the laboratory sessions, students were again required to conduct several short personalised experiments using the ReLOAD enabled position servo mechanism shown in Figure 2 for assessment purposes.

Key learning objectives for both laboratory sessions are:

to give students experience of conducting experiments and collecting and analysing real experimental data
to enable students to compare mathematical predictions to experimental data
to show students how changing key parameters in dynamic systems affects the position of the roots of the characteristic equation and hence their transient dynamic performance.

When calculating the roots of the characteristic equation two commonly made errors have been identified. These are:

students don’t work in the correct SI units (they work in Hertz rather than radians per second) and
students get the sign of the real part of the roots incorrect (they assume the real part is positive rather than negative).

The DaSh system enabled both of these to be immediately highlighted to the student without the need for a lecturer’s input. In addition, the system also allows for data obtained from one piece of equipment to be compared with another, allowing an appreciation of variation in data that can arise from nominally identical equipment. It was also envisaged that the DaSh system would encourage students to look for trends in data across a greater number of experiments than could be conducted by themselves alone and therefore shifting the focus from simply conducting many experiments to more fully understanding the data.

Submission of results using DaSh

Laboratory session

In the laboratory, pairs of students were able to upload (and subsequently remove) their results to a web accessible database using a graphical user interface, developed on a local network server using PHP and MySQL. The pair would be confronted with a web form (as shown in Figure 3) through which they could submit their data. Once the data had been submitted, their results would be displayed on a web page as a root locus diagram (as shown in Figure 4) allowing them to view their results and discuss them in their pairs.

Figure 3. Screenshot: submission page, showing previous submitted answers

Figure 3. Screenshot: submission page, showing previous submitted answers

Figure 4. Screenshot: root locus page, showing two sets of data, one shows variation in velocity feedback the other forward path gain

Figure 4. Screenshot: root locus page, showing two sets of data, one shows variation in velocity feedback the other forward path gain

In addition to the personal data displayed on dedicated PCs, an S-plane diagram displaying data from all students conducting the experiment was made available (as shown in Figure 5). The cumulative data was displayed using a data projector in a separate part of the laboratory. Locating the screen to one side of the laboratory allowed staff and demonstrators to easily observe whether students made use of this facility. The collaborative S-plane diagram was automatically refreshed on a regular basis to ensure that all submitted data was included. From the cumulative diagram staff and students were able to easily view an image of the whole data set, promoting discussion not only between the pairs of students but also amongst the whole class. Although individual data points were not identifiable, a scale was made available allowing students to easily recognise their own data points. The students were then given the option of removing and resubmitting their data following any discussions if they found they had made a mistake.

Figure 5. Results for all students from the laboratory session showing two sets of data, one shows variation in velocity feedback; the other forward path gain

Figure 5. Results for all students from the laboratory session showing two sets of data, one shows variation in velocity feedback; the other forward path gain

Data sharing using ReLOAD

During the follow-on ReLOAD session students were again able to submit individual results via a web form (part of the ReLOAD interface) to a database. Once they had submitted their results they were then able to view a root locus diagram containing all of the submitted data up until that point. In order to view a more complete diagram they were able to log back into the system at a later date to inspect the emerging data.

By viewing a cumulative plot, the students were able to compare their results with those of their peers. If, having viewed the group’s data, they decided they had made a calculation error or spotted another mistake they could repeat their experiment and resubmit their data.

Results

Laboratory session 1

Figure 6. Submitted experimental data from the first laboratory session

Figure 6. Submitted experimental data from the first laboratory session

Figure 6 shows the results submitted using the PRS during the first laboratory session. Despite support from two postgraduate assistants and the module leader in each laboratory, the common errors mentioned earlier are clearly apparent. Working in incorrect SI units produces a cluster of erroneous data points close to the origin, while getting the sign of the real part of the roots incorrect produces data points on the right hand side of the plot. Experiment 1A represents data taken from the beam without mass, while 1B represents the data from the beam with added mass. The spread of data is caused partly by the variation in the 14 separate (but nominally similar) pieces of experimental equipment and partly by measurement errors.

This data was presented in the following week’s lecture and, after this lecture, students were asked to complete the associated ReLOAD experiment.

Figure 7. Submitted experimental data from the follow up ReLOAD session

Figure 7. Submitted experimental data from the follow up ReLOAD session

Figure 7 shows the data collected from the ReLOAD experiment. Again, Experiment 1A represents data taken from the beam without mass, while 1B represents the data from the beam with added mass. The spread of data is caused partly by the variation in damping applied to each student’s experiment and partly by measurement errors. Despite having no support from staff during the experiment, the number of errors made is clearly much reduced. In addition it is also possible to compare the student data with a model solution made using ReLOAD, although this model solution data is not something that the students see.

Figure 8.Submitted experimental data from the second laboratory session

Figure 8.Submitted experimental data from the second laboratory session

Figure 9. First submission of experimental data from the follow up ReLOAD session

Figure 9. First submission of experimental data from the follow up ReLOAD session

Figure 10. Submission of experimental data from the follow up ReLOAD session

Figure 10. Submission of experimental data from the follow up ReLOAD session

Laboratory session 2

In the second laboratory session the real-time DaSh system was introduced. Figure 8 shows the data submitted by the entire cohort.

Experiment 2A represents data taken from the position servo mechanism as the forward path gain was varied, while 2B represents the data captured as the velocity feedback gain was changed. The spread of data is caused partly by the variation in the 14 separate (but nominally similar) pieces of experimental equipment and partly by measurement errors. In Figure 8 the only two clearly erroneous data points submitted are visible; five anomalous data points (three from experiment 2a and two from 2b) are also visible in the lower part of the graph. Though anomalous, these five are not necessarily erroneous as the complex roots appear in conjugate pairs and, in this case, it would appear that the students have elected to submit the negative root of the pair. Interestingly both the two erroneous points and the five anomalous points were submitted by a single pair of students.

Following the face-to-face session students were given a short break of one to two weeks then asked to complete the ReLOAD version of the position servo laboratory with data sharing enabled. Figures 9 and 10 show the data submitted via ReLOAD. Figure 9 shows how the data would have looked if only their first submission was allowed, while Figure 10 shows each student’s final submission after they had been given the opportunity to view and amend their data. In addition to the student data, model solution data is also given in the figures. Again, erroneous data of both common types was seen in the first submission. This was dramatically reduced as students were allowed to amend their data following comparison with their peers’.

Student behaviour

During both laboratory sessions video recordings were taken so that the level of interaction between pairs of students could be observed and to determine whether students changed their behaviour from session one to session two as a result of being able to view a collaborative image of the group’s data using DaSh. Detailed evaluation was, however, not required as the outcome was apparent within the session and was backed up via the focus groups and questionnaire data given in the following section.

It was obvious that in laboratory session one, students worked within their pairs occasionally asking for help from the support staff and, even less frequently, consulting with the pair of students sitting immediately adjacent to them (Figure 11). In session two, once students had started to submit their data, they were keen to gather around the collaborative display to compare and discuss their results (Figure 12).

Student feedback

Individual questionnaires

After both laboratory sessions the students were asked to complete a questionnaire about their individual and collaborative experiences during the sessions. Figure 13 shows that the students felt that the collaborative work was more important to them in the second laboratory session than the first. There was a rise from 23.7% to 34.0% in those thinking it was important to them to complete the task with a noticeable decrease in the number that felt it was not important to them.

Group questions

In addition to the data collected immediately following the laboratory sessions, students were also asked three questions in a lecture during the week after the second (DaSh) laboratory session. The questions and responses are shown in Table 1.

Table 1. Questions and results asked after laboratory session two
1. I changed my submitted data as a result of seeing other students’ results projected up in the lab

True 50.7%	False 49.3%
2. It was useful to be able to see other students’ results projected up in the lab
True 97.3%	False 2.7%
3. I felt I learnt more from the lab as a result of seeing other students’ results projected up in the lab
True 58.8%	False 41.2%

Student focus groups

In addition to the feedback collected through questionnaires, two focus groups were run with an independent evaluator from the Royal Academy of Engineering. The students anonymity was maintained at all times and the interviews were recorded and transcribed to ensure accuracy.

As mentioned previously, many students felt that the collaborative experience had been of importance to them. Through the focus groups it became apparent this was because they were able to quickly spot any errors they had made and correct them or, if necessary, start again.

‘You could go up to the board and see everyone’s results, which was really good, ‘cos if you were like miles away, it was really good ‘cos you could like see that and start again.’

‘It means you learn there and then that you have done it wrong and can start to put it right from that point onwards as oppose to waiting for someone to mark it and give you results.’

The students were also in agreement that working collaboratively improved their confidence and assisted them in making sure that the next stages of the laboratory were correct.

‘Having the instantaneous feedback gave you the opportunity to rework it and learn from your mistakes, often if we write lab reports it is a long time before we get feedback.’

Increased motivation to participate in the collaborative laboratory session was also commented on. Particularly appreciated was the ability to view trends as they were forming in real time as opposed to a simulation of what the data should look like. This enabled the learning experience to be enhanced.

‘Does make you learn as well though […] ‘cos like with labs and graphs with trends you know what they are supposed to look like […] but when everyone is putting up their results you can see the trend forming […] it’s not just this is what should happen you can see it actually happening.’

In addition to the learning experience being enhanced, the students also felt that comparing their own data with their peers’ increased their motivation to perform well.

‘Yes – it also put pressure on you to do better as you could see how you were placed compared to others. Like I should be as good as this guy.’

However, it was also noted that although students could compare data points with others’ this was not necessarily a trustworthy resource and therefore it may not be advantageous to make changes based on others’ data.

‘If they are all wrong which is quite feasible, then it’s not a good guideline to go by, they are all students they are all learning, so that doesn’t mean that they have done it right.’

Feedback from academic staff

Feedback from academic staff using the systems was very positive. Staff noted that it was easy to spot both group and individual errors and that it would have been easy to intervene if required. There was significantly more engagement between the support staff and the students and, importantly, plenty of peer pressure when it became obvious that certain members of the group were submitting erroneous data.