I saw this morning an advert from a company specialising in big data for universities – how to join together the data that universities already hold, turn it into useful information and get value from it in making a university more effective. There were some very impressive applications on display, enough to make any university management green with envy. All very good – but the client list is the big beasts of the higher education jungle: University of Michigan (income $3.4bn, 60k students), Oxford University (income £1bn, 22k students), Cornell University (income $3.1bn, 22k students), Brown University (income $700m, 8k students), Texas A&M (income $4.1bn, 53k students), Berkeley (income $2.1bn, 35k students). It can take big bucks to get big data.
In many UK universities budgets are tighter, and investment in the databases and analytics software that frees up big data isn’t this year’s (or next’s) priority. But it isn’t a lost cause: here are seven ideas which can make a real difference.
1. Know what data you have. Universities will have systems to record information and transactions about admissions, enrolments, exams, staff, space, finance, timetabling, learning resources, alumni, donors, research and more. Some of these systems may only be a spreadsheet, or paper-based files stored in one place, but knowing what is there can make a real difference. University management teams will see the possibilities of combining information; planning professionals will want to know what is there, make sure that its meaning is understood, and what the limits are on sharing the data.
2. A focus on data quality can be a real help. Look at where errors are creeping in to your data. Are you double-entering data because systems are not set up to be compatible? Have you got good documentation – with clear, unambiguous and relevant definitions of data fields, and good guidance for users – for all of the IT systems which you use to manage your background processes? A data quality policy will get you a tick from the governing body when it comes to the annual return to the funding council, and it can help you identify where you need to address problems.
3. Use the expertise you have. Universities have plenty of people who understand data and statistics – within the professional services, but also amongst the academic staff. Often these people will be only too pleased to be involved in making the data work better for their university. For staff in a professional services team, being part of a wider group looking at data can be a way to get a glimpse beyond the silo of their current role; and for academic staff, the chance to contribute on an institution-wide basis can be good for career development and professional recognition.
4. Get in training. Train people in what data you have – sharing this knowledge opens up possibilities. Train people in using the functionality of spreadsheet software – there’s power in these tools, for analysis and for presentation, which might surprise you. And train people in numerical reasoning – we all know an otherwise-high-performing-professional who has a real block with numbers, and overcoming this can be very empowering for them and for you.
5. Use the data you have. It’s always possible to want better quality data, in different formats, and bringing together data sets which don’t match. And there are some questions where you do need real accuracy. But the data you have is good enough to help answer an awful lot of questions: focus on what you can say, rather than what you can’t, and don’t let the quest for perfect data get in the way of effective use of data. Read ‘How to measure anything’ by Douglas Hubbard to get a sense of what is possible. And think about letting a postdoc scientist loose on the data – it’s their capacity to see and understand the numbers that matters, not their knowledge of the underlying business. You’ll be surprised at what a data scientist can do!
6. Look for bottlenecks in your systems. Do you have a colleague whose job it is to manage data requests, or is it a little bit of many people’s jobs? Is the data team in IT and disconnected with users, making prioritisation difficult? Sometimes sorting out one or two little problems can have a dramatic effect on how data can be made available and shared.
7. Spring clean your reports. Many data systems have reporting functions which require knowledge of SQL, for instance, to generate a report. Is the library of reports which have been coded a manageable size, and they reports which you still need? Find out what reports have already been written, remove duplicates, specify what you need now, and share the menu with others. Manage requests for new reports – if there’s real value in a new report, then it’s worth coding, but sometimes a colleague can happily use what already exists.
These seven tips won’t give you big data – you’ll still be casting longing glances at the analytics some universities use – but they will help you make an impact. And once the management team gets an appetite for data, who knows where that will go?
No comments:
Post a Comment