Using blended learning and problem-based learning approaches to embed quantitative methods into substantive modules

Using blended learning and problem-based learning approaches to embed quantitative methods into substantive modules

Raul Gomez, University of Liverpool

I do not have any hard evidence on this, but virtually every student I have talked to acknowledges that quantitative methods are a fundamental skill for Politics graduates. This, of course, is completely inconsistent with students’ apparent distaste for quantitative methods modules. I have seen this for years. Take Sociology, for example. Most Sociology programmes have at least one core mandatory methods module. And, most of the time, that is the module that receives the worst evaluations - usually combined with words such as “dry” and “boring”. The fact is that, whether they like statistics or not, many of our undergraduate students have not done any maths for years, so when confronted with standalone research methods modules, all they feel is fear and insecurity – combined with a lot of risk aversion.  

Embedding quantitative methods into substantive modules solves the dilemma confronted by students who are afraid of statistics but also understand the benefits of it. This is because embedding provides students with opportunities to enhance their quantitative skills while studying something else. Put another way, it turns quantitative methods into what they are supposed to be: a tool for understanding political phenomena. In practice, however, embedding quantitative methods into substantive modules is rather more complicated than it seems on paper. As soon as you start designing a module that contains both substantive and methodological topics you realise there are a number of questions that need to be addressed: What are the leaning goals? What level of statistics will students reach on this module? How and when will they acquire the statistical background necessary to understand certain methods?

As I see it, there are often two alternative ways of solving these questions. The first solution consists in asking students to take standalone methods modules as a prerequisite for any substantive module that incorporates statistical analysis. But that, I am afraid, is only viable for the very few students who would take research methods modules anyway. The second solution consists in lowering the level of the statistical content learned on the module, so that the lecturer can dedicate their attention to the substance without having to spend a substantial amount of their contact time explaining statistics. Students are generally happy with modules that do this, and they do always get something out of it. But there is an obvious limitation: while those modules work well for students who are unfamiliar with statistics, they do not guarantee a decent level of statistical proficiency.

To be sure, embedding can always be used to encourage students to become familiar with basic statistics in the hope that they will take standalone methods modules later. But I am not sure this is a great idea, not only because students will, of course, be aware that the level of statistics in standalone modules will be substantially higher, but also because it involves shifting our teaching philosophy from a learning-by-doing approach to a “let us lure students into the difficult stuff” approach.

Approaching the problem

I wanted to use embedding on a substantive module on voting behaviour, and I wanted students to be able to analyse survey data using inferential statistics. It was nothing complicated (mostly bivariate analysis), but I knew many of my students would have a very basic knowledge of statistics. The problem was that I had to figure out a way to offer training on specific methods (starting from basic statistical concepts) at the same time as I covered all the substantive topics. And I did not want the module to look as a methods module with a little bit of voting behaviour theory, because the purpose was precisely the opposite. So, my approach was to use blended-learning and problem-based learning to allow students to train themselves. This is what I did:

On my third-year module ‘Comparative Voting Behaviour’ students are asked to attend weekly lectures (1h) and PC sessions (1h). The lectures cover the substantive part of the module, so they follow a sort of traditional approach, only we spend a lot of time commenting on graphs and tables from research articles. Alongside the lectures, students are also invited to explore substantive topics further in PC sessions using statistical software.  Each PC session revolves around a particular country and a specific aspect of voting behaviour, for which I prepared 10 different datasets that are available to students in advance through Blackboard. The PC sessions are based on two active-learning strategies: flipped classrooms and problem-based learning.

Before the sessions: flipped classrooms

In advance of each session, students are asked to access two sets of materials (videos and handouts) that were specifically designed for this module. The videos provide a brief explanation of the statistical concepts or methods that students need for each session. The handouts provide more technical information about how to run different tests using the software (in this case, SPSS). Other materials that are provided in advance include “memo cards”, and images containing tips or summaries with useful information.

During the sessions: problem-based learning

Once in the PC Lab, students are confronted with substantive questions which they need to answer using real data (e.g. how did economic evaluations affect vote choice in the 2013 German election?). On my module, the questions are framed as a ‘problem’ brought up by a hypothetical client and contain brief background information about a specific election that students are asked to analyse using survey data. Problem-based learning means these sessions are not focused on methods but on answering substantive questions, which enables students to strategically develop quantitative skills by using statistics as a means to an end.

During the PC sessions, the role of tutors is essential. I supervise all the activities and provide assistance where needed. The flipped classroom format means that I can answer specific questions and help students out individually, so I do not need to spend time explaining the basics unless there is something that some student/s struggle to understand. Sometimes students come up with technical questions. Other times, their questions are related to the theory of the specific method they are using, or to the meaning of their findings. There are particular moments (for example, the first time that they deal with the concept of statistical significance) where you need to go through some of the concepts already explained in the videos because they find them difficult to grasp. But. in my experience, all this going back-and-forth really helps them understand what they are doing.

Importantly, students are not only encouraged to look at the materials in advance (which they normally do), but are also allowed to use them during the PC sessions. This is because the goal of these sessions is to enable students to learn while they explore the data, for which being allowed to watch the videos or read the handouts when and as needed is essential. This provides a context for students to apply what they have learned and reinforces the learning process by creating a link between theory and practice.

In the last 10 minutes, students are randomly chosen to report back on their findings and to show the other students what they have done and how. This introduces a nice element of peer learning and also gives them an opportunity to see how others have been able to solve the same problem in a different way.

What went well?

From my point of view, this strategy is a great success. Using flipped classrooms has many advantages. First, you do not lose all your contact time trying to explain statistics because you already do that in the videos and handouts. Instead, you can provide much more valuable individualised assistance to support students’ learning. Second, you can tailor the materials to meet the specific needs of different modules depending on year, topic, etc. And third, students can watch the videos whenever they want and as many times as they want – before, during, and after each session. This means that the learning process does not end (or start) in the classroom.

Most students achieved considerably high levels of statistical skills (as demonstrated by their assessed reports) and showed great enthusiasm with the subject and satisfaction with the module. Student feedback also points to high levels of personal development, with most students agreeing that the module makes them feel more ‘confident in tackling unfamiliar problems’ and that it had contributed to developing their ‘understanding of the topic’. They generally like the flipped-classroom and problem-based learning formats, and many of them become passionate about statistics to the point that some decide to apply for Data Analyst jobs or go on to do an MA in social or political research when they graduate.

Lessons learned

I think the first lesson learned is that we should trust students more. Instead of wasting our time explaining things that they can read about or watch outside the classroom, we should focus more on helping them understand while they learn in an applied fashion.

The second lesson learned is that creating the necessary materials for the first time is really time consuming. The bright side is that those materials can be adapted to the specific demands of other modules, so they can be used by different lecturers. In an ideal world, students should be exposed to embedding on at least one module per semester. This would not only allow them to become familiar with statistics without ever taking a standalone module on quantitative methods, but it would also enable lecturers to raise the difficulty of the methods they learn.