Lorna: Could you start just by talking a little bit about your own background and how you’ve come to be working with data?
Mark: I suppose my background’s different from a lot of people who do this kind of work. They often come from a maths, stats or science background, but I didn’t do A-level maths or anything like that. My first degree was politics – there was no maths in it at all.
After university I got funding to do a Master’s in social research and I found that I really enjoyed the quantitative side of that, so much so that I decided that it was the area in which I wanted to work.
I got a job working for HEFCE (Higher Education Funding Council for England – now the Office for Students) as an analyst. A lot of the people there had PhDs in maths and computer science, technical people. For many of them, they really enjoyed writing an elegant piece of code to extract and analyse data, but many were less interested in what that data could actually tell them. I could write the code too, but I was much more interested in what the data was telling us. I think that’s because of my social science background. I look at the data and think about what it’s telling us, what’s interesting about it and what we can learn from it.
After my time at HEFCE I went into strategic planning in universities. That typically involves working with the senior management of the university, understanding all the different data sources, extracting the data, and turning it into reports that are easily digestible for senior management.
Lorna: What tools were you using for your data analysis?
When I was doing my Master’s in social research I used SPSS for my dissertation research – most social science departments in universities tend to use SPSS. When I worked at HEFCE (now the Office for Students), they didn’t use SPSS. They used SAS which was much less user friendly, requiring writing code rather than pointing and clicking. It’s really just habit – it’s what they’ve always used.
In university strategic planning departments a lot of the time you’re just manipulating data in Excel because you’re not really doing statistics, you’re just generating graphs and tables. When I moved to my current role in Jisc I moved back more to the research side and so I’ve started using SPSS much more again.
Lorna: Can you tell me a bit about your role at Jisc and the types of analysis you’re doing there?
Mark: I’m responsible for carrying out surveys of students, teaching staff and professional services staff in universities, asking them about their digital experience and the reporting on the findings of those surveys. We’re interested in finding out about what technologies they use, their attitude to digital technology, whether they’re early adopters of technology or not, how they rank their organisation and the digital support that they get. There’s a whole range of questions on all aspects of their digital experience.
We use an online survey platform on which institutions run their own surveys and we provide them with guidance on how to do that. Then we harvest the data at the end, analyse it and write up a national report. Last year we surveyed 30,000 students as well as about 6,000 or 7,000 teaching staff so it’s quite a big undertaking. The data is extracted and put into Excel. Now, if it was a case of just doing graphs and tables for the reports then Excel does the job fine. The thing is, though, is you want to do statistical tests as well then SPSS does it a lot more easily and intuitively and offers a much wider range of tests than Excel. For example, I need to take the data into SPSS to test whether the differences we are finding in the survey data are likely to be true also in the sector as a whole. That’s the main reason we use SPSS.
Lorna: Can you tell me a bit about the questions that you’re asking? What are you interested in?
Mark: For example, we might ask respondents how effectively they feel their institution addresses issues of digital data privacy, or how well their institution manages digital teaching. We might then want to compare answers to a question like that between the HE and FE sectors. We’ll carry out statistical tests to find out whether the difference we’re seeing, which seems big, is likely to be actually true in the whole population that we’re looking at. Are the differences we’re seeing between HE and FE actually real? That’s what we’re using SPSS for.
Also, because we’ve got a student survey and a teaching staff survey, we want to see whether their answers match up within the same institution – do they share the same perceptions of the institution’s digital performance? In order to do that we need to get institutions to do both the student survey and the teaching survey and to generate a sufficient response rate from both groups to enable us to do the analysis.
We do a lot of correlation tests in SPSS. What we’re finding is that institutions where the teaching staff highly rate the digital support they get, are also the same institutions in which the students rate their digital experience highly. We can’t prove causation, but we can say that there’s definitely a relationship here. It’s quite likely that universities who are supporting their staff to use digital will have students who are more satisfied with their digital experience as well.
Essentially using SPSS enables us to take our analysis one step further, to be able to carry out statistical tests that not only describe what’s happening, but also tell us whether it has significance. I know there’s other packages out there, like R for example, but I don’t have experience of these yet and they’re not so user-friendly to get going with. Also, if you want to train up other members of the team then it’s quicker to do that with SPSS as it’s more intuitive to use.
Lorna: Can you explain a bit more about how your job works and what your main responsibilities are?
Mark: Jisc has its own survey package called Jisc Online Surveys – it used to be called Bristol Online Surveys – and I write a lot of guidance for institutions regarding how to use it. So for example that might be guidance about how to engage your students, how to run effective surveys, any aspect of how the survey system works.
I also work with external consultants to update the question sets every year as new things come in. We spend a lot of time testing new question sets. We’re also chasing institutions through the year, the ones that haven’t launched their surveys yet to say, “do you need any help here?” We’ve got a tracking system just to help institutions go through the complete pipeline, because it’s one thing for them to sign up, it’s another thing for them to actually do the survey. We’re helping them to get through.
At the end of the process the data is extracted into a CSV file for me. I’m one of the lead writers of the national reports. It’s like a ‘state of the nation’ report where we compare HE and FE. There’s one for students, one for teaching staff and one for professional services staff. Last year we also ran surveys in Australia and New Zealand with universities there. We’re piloting a researcher survey this year as well. There’s lots of different reports that need to be written.
Lorna: Do you see any additional opportunities for using data in other ways, things that you’d like to be doing with it?
Mark: Yes, there’s plenty of interesting stuff in there. For example, we want to be able to look at the differences between campus-based and online HE students to see what’s going on there. When we ask demographic questions, it’d be nice if we had time to break that down and maybe do some kind of regression modeling to see what kind of factors influence the scores. That’s something we just don’t have the time to do, basically. There’s a lot more to get out the data, it’s the constraints of time that limit what we’re doing at the moment.
Lorna: What advice you would give for people who are thinking about a career in research or analytics or data?
Mark: Be curious. You can teach technical skills, but curiosity , that’s something you cannot teach. If you’re looking at data, you should be saying to yourself, “That looks interesting. Why is that happening? Why am I getting these results? What are the confounding factors that may be explaining it? How can I test my hypothesis?” You can teach the technical skills but it’s much harder to teach curiosity.
Because SPSS is relatively easy to use, sometimes you find that people just press the buttons without any real understanding of what they’re doing or why. You don’t need to be a statistician to use SPSS, but you do need to understand at a certain level what the tests are doing and why. If you know that, then that’s enough then to start using SPSS.
People from a pure maths or science background are not always necessarily the best people for these types of roles. You almost want a social science background – someone who’s curious about society and to whom you can teach the technical skills. Those kinds of people are actually quite hard to get hold of.
Having that social science background gives something quite interesting to the mix, so I think people shouldn’t be put off doing data-related jobs just because they don’t come from a data background. It’s not enough just to understand the statistics. You also need to understand where the data came from, how is was collected and what it tells you. There’s no value in just coming up with lots of correlations without being able to explain what it means.
If people want to make themselves more career-proof in the future, it’s not always about developing more technical skills. It’s actually more people skills. There’s lots of software out there now which can bring different data sources together and replicate much of what data scientists are doing. We might not need people to do some of those technical things in the future but we’ll still need people to interpret the findings. Computers aren’t curious – people are.