Learning analytics is defined as ‘the measurement, collection, analysis and reporting of data about learners and their contexts, for purposes of understanding and optimising learning and the environments in which it occurs’(“Guest Editorial – Learning and Knowledge Analytics. Educational Technology & Society, 15 (3), 1–2. Siemens, G., & Gašević, D. (2012)”, n.d.).

Applications of learning analytics include Early Alert and Student Success, Course Recommendation, Adaptive Learning and Curriculum Design(Sclater, 2017).

Can and should this learning analytics be extended to identify normative behaviours of students and recognise changes to those behaviours, aiding pastoral support?

Although much of the data for informing pastoral support is the same as that for Early Alert and Student Success the aims and implications are different. The data needs may also be more demanding. For example, there will be additional considerations around data sharing and protection as mental ill-health is classified as a protected characteristic.

For inclusion of engagement data from virtual learning environments, this would involve understanding the seasonality of student interactions with online course content, cohort interactions, how a student’s interactions are differing from both their cohort and their own normative behaviour with respect to the seasonality.

Prinsloo and Slade (Prinsloo and Slade, 2017) note that ‘Not  only  do various  stakeholders  in  the  institution  work  in  silos,  responding independently  of  each  other and resulting  in  overlap and inconsistencies, institutional  sense-making  of  students  at  risk is  also  fragmented’, which may hinder student well-being support.

Evidence on the effectiveness of learning analytics based interventions in unclear. A systematic review and quality assessment of studies on learning analytics in higher education by the University of Exeter(Sonderlund and Smith, 2017) was only able to include 20 of 560 papers identified due to the methods employed in the studies, only 4 studies evaluated the effectiveness of interventions based on learning analytics. The key recommendation from the review is that more research into the implementation and evaluation of scientifically-driven learning analytics to build a solid evidence base.

The combination of the lack of evidence of the effectiveness of learning analytics based interventions and the potential negative consequences for both our students and institutions, therefore causes us to question whether learning analytics should be used to support student mental well-being.

Conclusion

Student mental wellbeing and in particular student mental ill-health is of major concern, with 48% of UK HEIs having appropriate policies in place(Universities, 2015), and continually needs to be addressed.

An “unhelpful divide” of distinguishing intellectual needs from emotional needs, then students mental health may suffer if the emotional needs are ignored(“What Happened to Pastoral Care? | HuffPost UK”, n.d.). Therefore, pastoral care in addition to academic support is crucial student mental wellbeing.

The Higher Education Academy UK Professional Standards Framework (“UK Professional Standards Framework (UKPSF) | Higher Education Academy”, n.d.) dimension A4, Developing effective learning environments and approaches to student guidance,  indicates that student support is an area of activity in which those teaching and supporting learning in higher education should be involved. Additionally, the Universities UK #stepchange(“#stepchange”, n.d.) guidance states that HEIs should seek to promote a diverse, inclusive and compassionate culture as part of their preventive actions.

Unfortunately, there are a number of inadequacies with the current provision of pastoral care in UK Higher Education Institutions.  I propose that learning analytics can be used to help to address some of these inadequacies by providing timely and meaningful data to personal tutors about their tutees., this is in alignment with the Univerisites UK guidance(“#stepchange”, n.d.) to align learning analytics with student wellbeing. However, this will require action on behalf of tutors, and there are legal questions still to be answered around negligence and failing to act on or engage with information provided via learning analytics.

Despite the potential ethical and legal issues around using learning analytics to support pastoral care and student mental wellbeing, I believe that this application area that should be explored.

References:

Research has shown that students who are distressed and at risk from mental ill-health will often exhibit one or more of the following indicators concurrently: academic struggles and failures, excessive absences from classes and obligations, excessive substance use, loneliness and isolation, social and interpersonal difficulties with others on campus, changes in self-care and lack of self-care, extreme risky behaviours, inability to tolerate frustration and normal stressors in college, inability to regulate emotions, hopelessness and despair(Anderson, 2015).

Gemmill and Peterson(Gemmill and Peterson, 2006) have found that internet communication may have the same buffering effects of stressful life circumstances in the same way as non-internet communication by increasing measures of social support and perceived social support.

This corresponds to findings by Gordon et al. (Gordon et al., 2007) who investigated types of student internet usage (meeting people, information seeking, distraction, coping and email) and four indicators of well-being: depression, social anxiety, loneliness and family cohesion.

Their findings suggest that it is the type of internet usage, more so than the frequency of use that relates to depression, social anxiety and social cohesion. Using the internet for coping purposes was significantly associated with lower levels of family cohesion and higher levels of depression and social anxiety. Whereas, information seeking and email were positively associated with family cohesion.

Research into predicting depression with social media(“Predicting Depression via Social Media – 6351”, n.d.), in this instance Twitter, found that social media contains useful signals for characterising the onset of depression in individuals, as measured through decrease in social activity, raised negative affect, highly clustered ego networks, heightened relational and medicinal concerns, and greater expression of religious involvement. It is noted that in order to identify changes in some behaviours, it was important to know the normal behaviours of the user e.g. an indicator of depression is a tendency to be more active at night. To be able to identify if there has been a change in activity the authors defined the normalised difference in number of postings made between the night window (9pm and 6am) and day window to be the “insomniac index” on a given day.

In summary, these studies show that identifying behavioural changes are key to identifying student mental ill-health, this therefore implies that an understanding is needed of normative behaviours of students.

References:

 

Byrd and Mckinney (Byrd and Mckinney, 2012) found that the combined effects of both individual and institutional level measures were associated with student mental health, accounting for 49% of the variance in mental health after controlling for background and demographic characteristics.  The IPPR state that their findings(“not-by-degrees-summary-sept-2017-1-.pdf”, n.d.) suggest that a majority of HEIs should take measures to ensure that the nature of course content and delivery does not result in academic rigour being sought at the expense of students’ mental health and wellbeing.

Social problem solving, coping, was identified as a more useful indicator of suicide potential than hopelessness, they therefore note that the concept of coping, especially in relation to students’ adjustment to university life, deserves further attention.

In addition, O’Keefe(O’keeffe, 2013) states that student wellbeing can be seriously compromised if the university is unable to create a caring environment, develop a sense of belonging among students and provide adequate campus based counselling support.

The Huffington Post blog post ‘What Happened to Pastoral Care?’(“What Happened to Pastoral Care? | HuffPost UK”, n.d.) states that the term ‘pastoral care’ has been missing from many of the discussions on mental health in higher education, and asks if the loss of the term ‘pastoral care’ reflects that we no longer tend to hold universities responsible for student welfare?

The Universities UK Student mental wellbeing in higher education Good practice guide(Universities, 2015) states that there has been a very significant growth in the specialist support and guidance services provided for students in higher education.  This includes supported provided within faculties and teaching departments including personal tutoring and other pastoral systems.

With respect to duty of care, institutions have a general duty of care at common law to deliver their services to the standard of the ordinarily competent institution; and, in carrying out their services and functions as institutions, to act reasonably to protect the health, safety and welfare of their students.

The QAA report What students think of their higher education (“What-Students-Think-of-Their-Higher-Education.pdf”, n.d.) identifies that positive and supportive relationships with a personal tutor was essential to successful learners. However, inconsistencies in students’ experiences continued to be problematic with student comments including:

The personal tutor organisation has been really poor. After four years at […] I am now

on my seventh personal tutor, who doesn’t know anything about me and I don’t feel

very supported in my final (and very stressful!) year. I’m not very happy at the idea of

this person writing a reference for me for a future job as they will only have the basic

information that is on my student record.’

 Suggested improvements included greater and easier access to personal tutors through scheduled tutorials and that tutors should be contactable via email.

 A response to the inconsistencies in the approach to student advising and tutoring has been the establishment of UKAT, the UK Advising and Tutoring group. UKAT believes that personal tutoring and academic advising have not been given the attention they deserve in UK institutions and aims to redress this situation, offering professional development and training in this vital area, providing a forum for the exchange of ideas and working to ensure that tutors and advisers receive the respect they deserve (“About Us”, n.d.).

References:

Mental health and mental wellbeing has become a key issue both nationally, with the independent Mental Health Task Force launched in 2015(“NHS England » Mental Health Taskforce”, n.d.), and for Higher Education Institutions (HEIs) with the number of 1^st year UK domiciled students with a known mental health condition increasing 220% between 2010-11 and 2015-16(“Disability – Higher Education Funding Council for England”, n.d.).

Table 1: First year UK domiciled HE students with known mental health condition(“Students and graduates | HESA”, n.d.)

Year	Number of students (all levels)
2015-16	15395
2014-15	11915
2013-14	9610
2012-13	7960
2011-12	7315
2010-11	6055

Most full-time first year students in UK HEIs are aged less than 25yrs (2015/2016: (“Higher Education Statistics for the UK 2015/16 | HESA”, n.d.) Table 4a) and are in the age group 16-24yr olds.

However, this does not account for those students that develop a clinically-recognisable mental health issue whilst attending HE institutions or those that report facing difficulties or distress. A mental health poll discussed the All-Party Parliamentary Group on Students in December 2015 found that 78% of respondents believed they had experienced problems with their mental health in the last year(“Mental Health Poll November 15 – Summary – Mental-Health-Poll-November-15-Summary.pdf”, n.d.).

The HEFCE blog post ‘Accommodating mental health’(“Accommodating mental health | HEFCE blog”, n.d.) reported that in 2015, student support services saw a 150% increase in appointments. Also, that approximately 29% of students experience clinical levels of psychological distressed associated with increased associated with increased risk of anxiety, depression, substance abuse and personality disorder.

The 2014 Adult Psychiatric Morbidity Survey (APMS) ((mr) Web Master, 2016) found that 15.7% of adults surveyed were identified with symptoms of Common Mental Disorder (CMD), with an expectation that this would be between 14.7% and 16.7% (95% confidence interval) for the whole population.  Common Mental Disorders comprise different types of depression and anxiety.  Anxiety disorders include generalised anxiety disorder (GAD), panic disorder, phobias and obsessive compulsive disorder (OCD).

Additionally the APMS suggests that amongst 16-24 year olds, there has been a growing gap in rate of CMD symptoms between men and women. In 1993 the rates were 8.4% (men) and 19.2% (women), increasing to 9.1% (men) and 26.0% (women) in 2014.  In addition, anxiety disorders were found to be more common young women than any other age-sex group(“apms-2014-cmd.pdf”, n.d.).

With regard to ethnicity, CMD did not vary significantly by ethnic group in men, but did in women with CMDs more common in Black and Black British Women (29.3%), and less likely in non-British white women (15.6%) compared to White British women (20.9%).

This is particularly worrying as evidence suggests that those who have high levels of depression are less likely to seek help and that depressive symptoms in young people are linked with negative attitudes towards help-seeking for mental health difficulties(“How psychological resources mediate and perceived social support moderates the relationship between depressive symptoms and help-seeking intentions in college students – 03069885.2016.1190445”, n.d.).  The Institute for Public Policy Research(“not-by-degrees-summary-sept-2017-1-.pdf”, n.d.) found that just under half of students who report experiencing a mental health condition choose not to disclose it to their university.

The APMS also asks participants about suicidal thoughts, suicide attempts and self-harm(“apms-2014-suicide.pdf”, n.d.).  A fifth, 20.6%, of adults reported having suicidal thoughts, with the expectation that this would be between 19.5% and 21.7% for the wider population (95% confidence interval). This was more common in women than men. It was noted that although men were likely to commit suicide, women were more likely to report attempts to do so. Additionally, more women than men reported self-harm.

The difference in self-reporting rates for suicidal thoughts, suicide attempts and self-harm between men and women are extremely noticeable between 16 to 34 year olds in comparison to other age groups.

Figure 1: Chart showing % men (M) and women (W) self-reported suicidal thoughts, suicide attempts and self-harm(“APMS 2014: Chapter 12 – Suicidal Thought, Suicidal Attempts, and Self-Harm – Tables [.xls]”, n.d.)

The authors of the APMS report note that although they did not find any significant differences due to ethnic group they recognised that this may be due sample size limitations and might mask real differences.

Following a Freedom of Information request from Universities UK, data on student suicide in the England and Wales (among students aged 18 and over) for 2007 to 2011 was released. These data are shown in Table 1: Student suicides in England and Wales (ages 18+), 2007 to 2011.  A University of York report(“Student Mental Ill-health Task Group Report Mar 2016.pdf”, n.d.) noted that while the overall number of students increased across the period, the relative increase in suicides far outstripped the increase in student numbers.

Table 2: Student suicides in England and Wales (ages 18+), 2007 to 2011¹²³⁴

Year	2007	2008	2009	2010	2011
Male	57	74	76	90	78
Female	18	21	33	37	34

Source: Office of National Statistics

1. Figures for deaths registered in each calendar year
2. Data for England and Wales includes deaths of non-residents
3. Data relate to those classified as full-time students at death registration
4. Suicide defined using the International Classification of Diseases Tenth Revision (ICD10) codes X60-X84, Y10-Y34

Self-harming method and reasons for self-harming data were grouped slightly differently, the age categories were 16-34 years, 35-54 years and 55+ years.  Across all age groups, cutting themselves was the most prevalent form of self-harm. This was reported by 84.3% of 16-34 year olds.  This age group (16-34) were also more likely to self-harm in order to relieve unpleasant feelings (81.9%) (which included feelings of anger, tension, anxiety or depression), than reporting self-harm in order to draw attention to themselves (28.6%, for all ages this was 31%).

With regards to help-seeking behaviour, 16-34 year olds were more likely to seek support from friends and family (29.9%) or GP/family doctor (29.1%) after a recent suicide attempt than hospital/specialist medical or psychiatric service (20.8%) with 51.6% not seeking any help.  37.7% of people who self-harmed received medical or psychological help afterwards, but for 16-34 year olds this value drops to just 31.1%.

A University of York Student Mental Ill-health Task Group Report (March 2016)(“Student Mental Ill-health Task Group Report Mar 2016.pdf”, n.d.) notes that students’ experiences of higher education have changed over the previous 10 years, which may have an adverse impact on their mental health highlighting three specific factors.

These factors were:

1. The rapid withdrawal of financial support for home students and an increasing reliance on loans, and in consequence, an increase in student debt.
2. The current cohort of students faces a more difficult labour market than earlier generations of students. There is a higher risk of unemployment and insecure employment for those graduating with arts and humanities degree than those studying medicine and subject allied to medicine.
3. The electronic environment created by electronic communication technologies can expose young people and students to pressures that were avoided by previous generations. This includes cyberbullying and victimisation.

The Institute of Public Policy Research (IPPR) report Flexibility for Who? Summary (“flexibility-for-who-summary-july-2017.pdf”, n.d.) states that  younger workers in part-time and temporary work are more likely to experience poorer mental health and wellbeing, with 22% of younger graduates who are overqualified for their jobs report being anxious or depressed, compared to 16 per cent of those in professional/managerial jobs.  Younger workers who work part-time are 43% more likely to experience mental health problems that those who work full-time.

It has been reported that the most common type of mental health problems at any university or college are depression, anxiety, co-occurring substance problems, eating disorders, suicidal ideation, and self-injury(Anderson, 2015) echoing some of the findings of the 2014 Adult Psychiatric Morbidity Survey.

References

In September I presented at the Association for Learning Technology Conference. My talk was an argument that we should be doing more to support students, especially with the growing concerns around mental wellbeing, and my belief that learner analytics can and should be used to assist in this support.

This talk was recorded and can be viewed on YouTube: https://youtu.be/M9Fzdn943eE?t=48m9s

In addition to the talk, I would also like to share my preparatory work with you. This will be presented over a series of blog posts, of which this is the first.

Learning Analytics as a tool for supporting student wellbeing – Introduction

The last few years has seen a raise in awareness of mental health/wellbeing issues through campaigns such as Heads Together and of student mental health/wellbeing with a number of articles being published on this topic by The Guardian.

At the same time questions have been raised about the quality of support for students in our Higher Education Institutions, will many students complaining about the poor quality of personal tutoring and the lack of information their tutors have about them, the course they are studying and their progress.

Much of the data that would be useful in assisting the role of the personal tutor, is also useful in the analysis and improvement of teaching and learning. In the emerging field of learning analytics data is being leveraged predominantly with the aim of improving student retention and enhancing the student experience.  I therefore ask if we can apply learning (learner) analytics to better informing personal tutoring.

However, there is much debate around the ethical implications of learning analytics and the data protection and privacy rights of students. Due to the sensitive nature of mental health and the potential fatal consequences of mental ill-health this is further complicated.

Due to the scale of the mental wellbeing issue within our institutions and the duty of care to our students, should these outweigh other concerns?

These blog posts will discuss four key themes with regard to this question, these are: student mental health and mental wellbeing, the role and responsibilities of HE Institutions in supporting students’ mental wellbeing, how mental wellbeing issues can be identified and the role learning analytics could play in supporting student wellbeing.

I was invited by Skills Matter to give a talk on the 8th March 2017 as part of the InfiniteConf Bytes series of meet-ups.

InfiniteConf Bytes is a series of talks showcasing the speakers and topics for the upcoming InfiniteConf at Skills Matter in July (6th & 7th).  At which I will be giving a talk entitled: Ethical Conundrums of an Educational Data Scientist.

My talk on the 8th March was a general overview of what learning analytics is and what some of the difficulties faced by the field are.

A recording of the talk can be viewed at: https://skillsmatter.com/skillscasts/10055-learning-analytics-what-is-it-and-why-is-it-so-difficult-with-samantha-ahern and the slide deck from the talk is available on slide share (https://www.slideshare.net/SamanthaAhernMIET/learning-analytics-infiniteconfbytes).

I am currently preparing a session on Learning Analytics for the HEA Conference on 5th July, and have had my session proposal for September’s ALT conference accepted.  A summer of literature reviews and writing for me it seems.

In addition, I have found some interesting patterns in the laptop loans data that I am hoping to investigate further. Included below is a plot from Tableau showing the relative number of laptop loans by location from the 26th September 2016 until 24th April 2017. In the plot the larger the bubble, the greater the number of loaned laptops.