Breaking-up is hard to do… exploring the limits of school data

One of the more fanciful claims surrounding the datafication of schools is the idea that (nearly) everything can be known through data. We are told that comprehensive data collection and analysis can lead to fresh educational insights. New patterns can be identified, trends spotted, predictions made, and teachers can better ‘know their students’. The underlying promise is that increased datafication will lead to prescient decision-making and more informed ways of working.

Of course, things are rarely this simple!  The scope of data that can be generated amidst the hurly-burly of the school day remains limited. There are substantial blind-spots in every school data system – including many factors that might well be of educational significance but (to quote Donald Rumsfeld) remain ‘known unknowns’.

Our ongoing research into school datafication addresses a relatively straightforward question: how are increased levels of data generation influencing what is ‘known’ (and what is not) within school communities? One interesting area of contention emerging from our investigations so far is the delicate matter of family break-ups.

The knowledge that a student’s parents are in the midst of separating or getting divorced can be useful information for schools. Parental separation can be an immediate and/or delayed impediment to a student’s wellbeing and/or learning performance, as well as an underlying factor in behavioural change. Moreover, from a simple administrative point of view, changes in parenting arrangements and responsibilities can lead to logistical and communication complications.

As such, number of our conversations around our research schools have touched upon the benefits of having reliable data on parental separation. This has particularly been the case in our Catholic school – Brookdale High School.

First, from an administrative point of view, lack of data on family status had led to a few practical problems for Brookdale staff – most of which relate to the school’s automated systems being configured to presume that all students are part of nuclear families. This is considered by the school to be the logical default position. As Brookdale’s Director of IT put it, “you’ve got two parents and they’re together.  Like, it’s a family”.

This led to some technical glitches. For example, unless a specific flag had been added, Brookdale’s school management system is configured to presume that any new contact details for one parent will also apply to the other parent by default. While Brookdale’s main online administrative system is capable of holding different addresses, its default is to provide both parents with all addresses and contact details associated with the student. As Brookdale’s office staff recounted, this has proven problematic where injunctions or restraining orders have been served on one parent, or where new partners were also now listed as guardians. Office staff reckoned that having reliable data on family relationship status could avoid similar mistakes.

The second area of interest in parental separation data was more analytically-focused. Here, some of Brookdale’s more data-savvy staff raised the possibility of factoring-in data on family break-up into the school’s data modelling and analysis. Indeed, during our first year in the school, staff were in the early stages of procuring a relatively sophisticated ‘Business Intelligence’ system that they (ambitiously) hoped could be a ‘one-stop-shop’ for school data analytics. This led to various discussions about the value of having data relating to students’ family backgrounds included in these analytics. As the school’s Director of IT speculated:

“So if we clearly can see a trend that there’s been 10 students that have all had family break-ups and nine out of 10 scenarios have all seen a drop-in performance …  so the BI tool might be able to predict for another kid where this happens then this is also going to happen.  And it can create a red flag for where a teacher may not be attentive to the problem.  Then the teacher gets a notification, they become attentive …. I think you could do some of that based on data”


Yet while Brookdale is proactively gathering all sorts of family and household information, at the moment this does not extend to collecting comprehensive data on parental relationship status. There are obvious legal and ethical reasons why schools do not (and cannot) demand this information outright from all parents. Moreover, this is not information that is easily reductive to a few discrete categories (as illustrated in Facebook’s decision to include a catch-all ‘It’s Complicated’ option for its ‘Relationship Status’ tag). This raises an interesting case-in-point regarding the possibilities and limitations of school data – in short, what can be known about parents’ current relationship status within the datafied school? Moreover, how is this knowledge being constructed?

Brookdale is adopting a few different strategies to generate data relating to students’ family backgrounds. First – as with most schools – administrative staff are primarily relying on parents to inform the school of any changes in circumstance. However, as a Catholic school, Brookdale families are often reluctant to disclose such information. As such, the school’s administrative team is also making use of a couple of ‘proxy’ indicators from other existing data to flag the likelihood of parental separation. These are both practical workarounds that used data generated by parents’ use of the school’s online systems to create additional indicators of family break-up.

First, is a flag triggered by only one parent logging into the school’s ‘Parental Permission’ system to change their contact details:so Mum could go in and change her details.  Then we’re thinking have they split?” (IT systems manager). The other tell-tale instance is a request for alterations to the parents’ payment details: “[one] way to know if they are split is that they’re at different addresses for accounts.  The LMS has got a symbol on there if they’re splitting the bills”

These might seem like relatively trivial aspects of how digital data is implicated in the day-to-day running of the school. Nevertheless, even these minor instances highlight some practical limitations of the datafied school. For example:

  • There are significant factors within students’ ‘non-academic’ backgrounds that do not lend themselves easily to quantification. There are many different forms of parental separation and post-separation family arrangements. Moreover, these situations can change quickly, and any data generated is likely to be post hoc and of limited predictive value.
  • School-wide engagement with data is clearly shaped by prevailing cultures and dominant values within schools and their wider communities. In Brookdale’s case, any administrative and analytical benefits of generating data-points on family separation are off-set by the Catholic values that pervade the school and many of its families.
  • School staff are understandably wary (and uncertain) of privacy issues. As one Brookdale administrator described, “you have to be so careful with addresses, phone numbers, family details …. So there are ways of knowing [about changed family circumstances], but yeah … the privacy [issues] keeps stepping up”.
  • Family break-up is an issue fraught with emotion and anxiety. There is inevitable tension between efforts to objectively ‘measure’ and ‘record’ this aspect of family background, and the traumatic nature of the incident being recorded. School staff and parents involved in the generation of this data therefore face considerable emotional demands. As one of Brookdale’s IT staff put it, the presumptions being made over split payments and change of addresses as proxies for a family breakup, “gets a bit awkward”.


At this point, then, perhaps the obvious concluding question to ask is simply whether parental separation is an appropriate area of students’ lives to be attempting to generate comprehensive data about? If not, then what other aspects of schooling are similarly inappropriate and might best be considered out-of-bounds? What impact might this have on prevailing hopes that data can ‘reveal all’ about the nuances of how schools work?