A conversation reported to me by my most reliable informant (my wife) suggests that schools in general are pretty hopeless at data protection and freedom of information. Educational data is her professional field, and the conversation was with a lawyer specialising in the same.
The use of data in schools has – well, it hardly needs me to say…. and while one would hope that, as public institutions they understand their obligations, I wonder to what extent this is generally true.
For example, the use of algorithms to derive pupils’ target grades might seem harmless enough. But when those targets are used to inform specific actions in relation to individual pupils, there is at least a ‘grey’ area regarding legality – and certainly what the pupil has the right to know. Every week I see cases where precisely this happens – teachers making decisions about expectations, intervention strategies, exam entries and more.
Section 12 of DPA covers automated decision making:
(1) An individual is entitled at any time, by notice in writing to any data controller (i.e. any organisation (not a private individual) holding their personal information), to require the data controller to ensure that no decision taken by or on behalf of the data controller which significantly affects that individual is based solely on the processing by automatic means of personal data in respect of which that individual is the data subject for the purpose of evaluating matters relating to him such as, for example, his performance at work, his creditworthiness, his reliability or his conduct.
(2) Where, in a case where no notice under subsection (1) has effect, a decision which significantly affects an individual is based solely on such processing as is mentioned in subsection (1)—
(a) the data controller must as soon as reasonably practicable notify the individual that the decision was taken on that basis, and
(b) the individual is entitled, within twenty-one days of receiving that notification from the data controller, by notice in writing to require the data controller to reconsider the decision or to take a new decision otherwise than on that basis.
The fact that automatically-derived data is being used to shape teachers’ decisions might therefore cause some concern. We should at least be aware of the above, which I will admit I was not until the chance conversation about it with my wife.
Other decisions are taken which are not always documented and this might cause further concern. For example, I have known of cases where raw targets were always rounded up rather than down simply on the whim of those doing it. While I think we can accept that education uses data for overwhelmingly benign reasons, the impact of specific decisions on the experience of individuals is not always positive, for example through the amount of pressure that the outcomes can apply.
The following (PDF) comes from the Open University’s ethical use of student data policy: See especially principle 3.
Analysis based on the characteristics of individual students at the start of their study must not be used to limit the University’s or the students’ expectations of what they can achieve.
- Predictive analytics reflect what has happened in the past, not the future. In their calculation of error rates, it is accepted that there will always be individuals whose behaviours do not follow the typical pattern.
- We should guard against stereotyping. Students who do not fall within any priority group may encounter difficulties during their study which become apparent as a result of learning analytics data, and subsequently benefit from targeted interventions.
- Caution needs to be exercised in the interpretation of data for a variety of reasons and guidance provided to staff will aim to support this. For example, individual members of staff may not have access to the full data set that is available to the University and may have an incomplete view of the student and their experience.
We might also consider the situation with respect to teachers. The data derived about their classes is a major factor in determining their performance. A lot hangs on this data being correct, particularly as the thing of the moment seems to be the judging of ‘progress over time’. Such judgements can only have significance if the baseline data are correct to begin with. Again, the consequences of getting it wrong can be severe.
At very least, the mechanisms used to derive data should be available for scrutiny, yet I’m far from certain this is always the case. While I haven’t gone over it with a fine toothcomb, as far as I can see there is nothing on the Fischer Family Trust website that provides public access to even an outline of their methods. Perhaps it is available behind the log-in, but I don’t intend to register to find out (even if it will allow a private individual to do so).
Some years ago, another reported conversation indicated that a County Council personnel professional was only too glad to see the back of schools as they were a nightmare on that front. The use of data is now so endemic that many teachers probably scarcely pause to consider the implications. Too often, data are treated – by schools, if not individual teachers – either with gay abandon or as the gospel truth, especially when it comes to predicting exam results. In doing so abuses may well be being perpetrated that potentially invalidate any judgements being made.
Here, we may have another example of the way the supposedly shiny corporate culture of modern schools is hiding unacceptable practice. At very least, I suspect that more training for schools and teachers might not go amiss.
With thanks to my wife for the D.P. input.