#LAK15 Becoming strategic writers

Ge Vue and Tracey Hall

Purpose to teach middle school students how to become effective writers through web based learning tools. Needing to integrate writing elements, such as making statements, cognitive strategies and affective strategies. Designed a visualisation dashboard to give feedback to students. Qus asked: could students and teachers make sense of the visualisations and which did they find meaningful? How do student and teachers use student information to improve teaching and learning? Enhancing feedback as an instructional strategy to improve knowledge of writing elements, strategies for writing and the quality of the writing. Can we visualise student writing in a way that supports feedback which leads to an improvement? 

Built in 3 different spaces for students to practice: (1) a playground to allow students to try out writing different types of activities and writing aspects to focus on, eg writing a claim with supporting reasons. Students could also set the difficulty of the exercise and had access to knowledge cards to support their understanding; (2) ? and (3) progress monitoring  – a phased prompt to do a specific short activity targeted at an individual student. The activity gets scored by a teacher and the student gets feedback.

Visualisation needs to have some specific detail in order to be useful. Teachers are getting live view of students writing their assignments and these can be shared and critiqued with other students in the group. Students encouraged to share their work and discuss with other peers. Student log data also collected on student id, the event type that they are engaging with, what actions they took, resources used, actual text written, etc. Use Tableau software to offer back a visualisation of what students were doing. Simple visuals seemed to be preferred. Prompts used to trigger remedial actions and to encourage students to try a number of different strategies until they find one that works for them. A greater focus on the process rather than just the content and how this might impact on students’ actual writing.

Posted in Uncategorized | Leave a comment

#LAK15 you’ve got style: detecting writing flexibility across time

Erica Snow, Laura Allen, Matthew Jacovina, Cecile Perret and Danielle McNamara

Writing is not easy! Aim is to better understand the writing process and build that understanding into interventions and strategies to improve writing. Focus on linguistic features that make up student essays and narrativity in particular, elements in text that are story like, actions, familiar places etc and this is commonly assumed to indicate good writing.

Strong writers have a flexible writing style that is adapted to the context. Traditional measures may not be able to capture the changes over time that are encouraged in students who are taking a flexible approach to their writing. Team use dynamic methodologies to see how students adapt content across multiple types of essays.. Use Natural language processing approach to get info on cohesion, emotion and readability. Also dynamic systems theory to focus on complex and fluid interactions, offers a unique means of characterising patterns that emerge across time (a peak behind the window). Wanted to use a technique that would allow tracking over time in quality of students’ writing without the need to get them to write lots of separate essays.

At what point does link between writing flexibility and skill become evident? Study took students and assessed baseline levels, conducted several writing assignments over 2 days, analysed different dimensions of the text to measure narrativity of the content. Students work was categorised on a narrative score range to see whether there were any differences for individual students across the different writing tasks. 

Approach allowed faster understanding of links between narrativity and flexibility in writing.

Posted in Uncategorized | Leave a comment

#lak15 how should we quantify student engagement? 

Perry Samson

Qus: how can we know whether students are engaging with our teaching? What makes a smart student smart? Do smart students engage in a different way?

Outcomes likely to be impacted by motivation, academic background and their participation, life circumstances, etc. Can we build a model which predicts how students will achieve? What are the measures of success? 

Suggests as measures of participation: attendance, questions posed, marks, capturing/downloading lecture, notes taken etc. all this captured in a LA system. Audience suggest attendance is perhaps key :)  Samson shares some data from previous groups. The more students followed the lecture slides, the better they did; the more notes taken (word count and frequency of note taking)  the better they did; reviewing lecture videos also linked. Ok. So far, so intuitive. So, what can be looked for at an early stage in the term/semester to be used as a reliable predictor of likely outcome? 

(Un)surprisingly, no link to class attendance by itself. Attendance is required with some other action.  Previous academic achievement is strongly linked to success. So what is the added benefit of the instructor? Already smart students do more in class. Less smart students engage less in class related activities. So what to make of this? Can we say that students with lower previous academic success have lower success because they are less smart or because they are consistently doing less (motivation) or have lower study skills? Not sure anyone really knows. Is a smart student smart because they are doing the extra activities (the activities add to their smartness) or because they understand that they need to do more? 

The data was shared with students and students asked if they were happy with their test performances. Having seen the apparent link between greater note taking, video views etc, students stated an intention to do more. But would students actually change their behaviour following this? In practice, students did not change their ways. Very short term impact. So what next? Suggests perhaps the use of ongoing feedback (dashboard) which presents back to the student their activity and their performance. Potential that ongoing feedback might lead to a longer term change in behaviours. 

Posted in Uncategorized | Leave a comment

#EP4LA #LAK15 applying learning analytics to a primary school classroom: benefits and barriers

Rodriguez-Triana, Martinez-Mones and Villagra-Sobrino

Based on previous studies into LA in higher education. Little awareness previously paid to data privacy and ethical issues (“because the data was already there and available”). A recognition that automatic data (VLE) is not enough. Always need a personal layer added by the teacher to contextualise and make data set complete.

In following this up, developed a pilot study in a primary school context using blogging tool. Teacher able to identify who is accessing and when (when the pupils are at home). Pupils not aware and cannot see this overview of their access. Parents were offered an explanation of the project and supplied ‘informed consent’.

Main issues were related to data ownership stake and control. Interesting distinction that younger students automatically need greater protection than older students. Permissions granted from parents/headteacher (“more hassle”). In fact, parents created dummy accounts on behalf of their children as pupils were minors and denied permissions to use the software tools.  Data shared with teacher but not with families as it was felt that parents would not understand. Question then around whether parents were able to provide truly informed consent. Interesting.

Posted in Uncategorized | Leave a comment

#EP4LA #LAK15 de-identification in learning analytics 

Khalil and Ebner

Key issues are: Transparency and 3rd party issues, identification, ownership, accuracy, security. Research into how to de identify students. In context of education, tensions between learning analytics and personal information. Students don’t want research into their personal data. 

Aim is to achieve privacy, research reasons (key to be able to analyse data for research purposes rather than just the application of learning analytics to students), available to public (data that can be freely released). In the US and EU, there is relevant legislation HIPAA, FERPA, DPD 95/46/EC which allows anonymised data to be used.

Propose approach which analyses de-identified student datasets (blurring, masking, hashing, suppression – remove key personal information, swapping – deliberately assigning characteristics of one student with another’s,  noising – amending/adjusting results by x%, say). Despite such techniques de-identification is fraught with problems. 

Posted in Uncategorized | Leave a comment

#EP4LA #LAK15 privacy driven design of learning analytics applications – exploring the design space of solutions for data sharing and interoperability 

Tore Hoel and Chen

Privacy control of data and trust essential to learning analytics. What are the implications for LA design of setbacks in learning analytics/privacy issues? Who owns the data? LACE suggests we give parents and students control of their own data…? Will this solve the privacy problem? Privacy is recognised but only superficially so, mostly seen as a barrier that should be overcome. Issues defined by Nissenbaum are contextual integrity, informational norm, actors, information types, transmission principles, contexts (tech, business practice, social domain). Where are the boundaries between formal agreed study and what goes on in the private space? Hoel suggest that socio cultural barriers to sharing data more important than legal barriers.

Should we be asking whether any application of learning analytics would pass a public deliberation on the appropriate use of the data? Good question. Societal approval is key…. Eg the InBloom collapse failed due to public perceptions around the trust issues rather than the application per se. Hoel suggests any LA application should have privacy aspects designed in at the start as an integral part of the solution/approach. Bit.ly/lashare


Posted in Uncategorized | Leave a comment

#EP4LA #LAK15 ethics workshop – student vulnerability, agency and learning analytics: an exploration (Prinsloo and Slade)

We presented a paper around student vulnerability as an interpretative lens for consideration of student control and choices around uses of their data within higher education. Background given on general lack of clarity and policy with HEIs. Range of consent options available (we use all your data and tell you nothing -> we tell you everything and cannot use your data unless you tell us we can).  But consent is tricky to get right. Range of factors which impact on how students understand/engage with the issues: how consent is presented (opt in as opposed to opt out); transparency of issues/options; presentation of options (length, language, even font), etc. 

Examination of TOCs of 3 major MOOC providers suggested a framework for discussion:

Questions posed 1) do workshop participants agree? 2) how can these be put into practice (at scale)?

1. Reciprocal care: about the power balance between students and their institution. The institution should be responsible and transparent about its purpose for using student data: TOCs must be visible/understandable. Students take responsibility for their own data being current/complete/correct.

2. Contextual integrity of privacy and data: must maintain a record of the original context and make this available for scrutiny. As data becomes aggregated and revised is important to retain the original context and purpose.

3. Student agency and privacy self management: there is an asymmetric power relationship between student and their institution. Student must be aware of the circumstances under which their personal information is used to tailor their curriculum and the services/support offered to them. Institutions must consider offering more than a basic process where registration = consent.

4. Rethinking consent/employing nudges: making consent more meaningful/valuable to students. Make clearer what benefits there are in sharing data. Are students willing to share with the right incentives?

5. Partial privacy self management: Define which services a student can/is opting into/out of – clarity of consent/purpose. Eg I can choose not to engage with ‘a’ but am happy to have my info used for ‘b’

6. Privacy timing/focus: data can be reused years after its original purpose. Institutions must be clear what data can be saved and reused later (and why). Setting clear restrictions with a specific focus on timeframe/purpose.

7. Substances v neutrality: rules/legislation can be constraining/incomplete/insufficient/too late but substance/clarity/boundaries always needed. Need to balance with flexibility to negotiate some issues (soft rules).

8. Moving toward the qualified self: students are more than the sum of their data (quantified self). How can we retain a layer of personal context to ensure that data remains meaningful/relevant/representative at scale? Challenge of sense-making 

Posted in education, Ethics, Learning analytics | Tagged | Leave a comment