Choosing a university course, the geeky way – and why it matters

In today’s UK, around 50% of our young people are likely go on to higher education at some point before they turn 30. But how well are they choosing which courses to study?

In some ways, you could argue it doesn’t matter too much: people with higher education are more satisfied with their lives, even after controlling for their higher incomes. However, there are clearly issues, too. Almost 40% of people in the UK work in a field that is different to their original studies. This is the third highest mismatch (after Greece and Ireland) among OECD countries. According to a CIPD survey, nearly 30% of people report that they are over-qualified for their roles. Such mismatches can be bad for productivity, wages and career satisfaction. There is also an issue for public finances: the Institute of Fiscal Studies forecasts that more than 80% of graduates will not repay their student loans in full.

So why do young people choose university subjects that might not pay off for them? Rather than provide an analytical answer (and apologies for any disappointment amongst readers that that may cause), I’m going to share a personal case study: helping my daughter choose which universities and courses to apply for. You’ll find out how a true geek goes about it and about all the fascinating data that is available to inform applicants’ choices. At the end of the blog, I also share a further reflection on why this matters: unless girls start choosing different university courses, we may never close the gender pay gap.

Big life decision? Bring out the spreadsheet

I have a coffee mug which says “I love spreadsheets”, and it is not an exaggeration (even though Tableau has been gradually overtaking Excel as my go-to tool). Ask anyone in my family and they’ll tell you that I’ll happily spend a few hours on a Saturday night “torturing data until it confesses” – on topics ranging from drug efficacy to weather in different capital cities to GDP’s correlation with various outcome metrics to the UK’s productivity puzzle. Major life decisions (e.g., which flat to buy or which job to take) have been informed by multi-criteria analysis using both objective and subjective information.

It is no surprise, then, that as my daughter contemplated her university choices, I turned to fact-based analysis. I was not discouraged by the fact that on previous occasions, such an approach had sometimes produced “the wrong answer” (i.e., an answer I did not like). That’s why the weights for the various criteria can and should be tweaked. Indeed, the process of listing the criteria, reviewing the data, and understanding the choices is probably more valuable than the output itself – and even more so when this leads to useful discussions with family members.

There is a lot of data about universities and courses out there

However, gathering the necessary data does take some serious commitment. Some data is delightfully easily available in a user friendly (and Excel- and Tableau-ready) format, such as the government’s statistics on graduate outcomes (see resulting heatmap in the header picture for this post, also available here). As Exhibit 1 shows, these outcomes vary dramatically between different courses and different universities. This therefore seems to be a critical piece of information to incorporate into any analysis about choices.

Various websites provide league tables and student satisfaction scores, even though these are rarely formatted for downloading and often have inadequate metadata about methodology and context. On student satisfaction, the National Student Survey results on the HEFCE website appear the most comprehensive. As for league tables, I found the Complete University Guide the most meaningful, even though I now see that the Guardian has a column on “value added”, which is to be applauded, and The Times Higher Education World University Rankings is helpful for providing an international perspective.

There are also complications in compiling this data, especially if one is keen to automate the process. Take the issue of university names. Where the World University Rankings refer to “University of Newcastle”, the HEFCE uses “University of Newcastle upon Tyne”. The graduate outcomes data refers to “The University of Edinburgh”, and HEFCE to “University of Edinburgh”, and so on. Such differences are not a hurdle for humans but make working with spreadsheets a bit more labour intensive. I actually gave up on integrating student satisfaction scores into the analysis for this reason. Indeed, for many data items, I ended up adding them manually after creating a shortlist of 20 or so courses. For this task I found the Which? University website to be particularly relevant and easy to use.

In contrast, the user-friendliness of universities’ own websites leaves a lot to be desired. I would have thought that most prospective students will have been told by their secondary school that there are a number of standard things to check out before making your choices. In particular, you want to know about entry requirements, but also the teaching and learning style (e.g., lectures vs. tutorials vs. independent work) and assessment approach (e.g., final exams vs. course work). Here, I have to give a shout-out to the University of Bath, whose website is quick and easy to navigate and which provides very clear graphical information on learning style and assessment (Exhibit 2).

Crunching the data to get to a short-list of courses

With much of the key data gathered, I then needed to come up with a prioritisation algorithm. These are the criteria I and my daughter ended up using: fit with preferred learning style, fit with preferred assessment style, earnings 5 years after graduation, ranking in university league tables, and a subjective element. (I had forgotten about the data on life satisfaction by occupation, shown in Exhibit 3, which probably should also have fed into the analysis and possibly have been a separate criterion.)

The subjective criterion mentioned above was essentially our “gut feel” of how good a fit the course would be after having read the main description of it on the university’s website (which, of course, by now I had copied into the spreadsheet). To arrive at scores for learning style and assessment style, I created a two mini-surveys, asking my daughter to rank the different options (e.g., lectures vs. tutorials, exams vs. course work), so I could allocate points to each course based on the combination of her preferences and the course characteristics. I also asked her to give a relative weighting for each of the criteria.

With a little bit of Excel wizardry, I then created a summary sheet with a ranked list of courses (alongside other relevant information, such as typical entry requirements and acceptance rates), which would dynamically change if we changed any of the weights, preferences, scores, or data. This was enormously helpful in creating a narrower shortlist and ultimately homing in on the five courses my daughter applied for.

Was it worth it?

If you’ve read this far, you are probably a fellow geek. However, you might still be wondering if it was worth all the time and effort on a choice that most people presumably make on the basis of a bit of googling and talking to friends, relatives and acquaintances. The answer has to be that “it depends on who you are”. While frustrating at times, I personally found the overall process very rewarding. First, I love spreadsheets (you knew that already), and my time was certainly better spent on this than playing Candy Crush. Second, I learned a few new Excel tricks, which is always uplifting. Third, I got real satisfaction from knowing that we had researched such an important decision thoroughly. Finally, I got to have meaningful conversations with my daughter about what she was looking for.

Reflections on the bigger picture

Along the way, I also found many interesting patterns in the data. The one I have ended up referring to the most is the gender balance of different courses (Exhibit 4). As we point out in our McKinsey Global Institute discussion paper, “The future of women at work in the United Kingdom”, the gender pay gap starts early. Certainly by the time people make university choices, women disproportionately choose subjects that just don’t pay very well. There are several reasons for this, and one might be that we are currently under-valuing some of the female-dominated careers, as I argue in a previous blog.

I do also wonder how many of the girls have access to the future earnings data, and how many of them might use it in their decision making. In no way am I suggesting that money is all that matters. We know, for example, that income matters less for individuals’ life satisfaction than mental and physical health (Exhibit 5). But I still think that the data should be more easily accessible, and something that parents, teachers, career councillors and others talk to kids about. This would contribute to more informed choices and possibly fairer outcomes. (Not everyone’s mother is an Excel and Tableau enthusiast.)