Engaging with research is an important part of understanding cognitive science and the science of learning.
It helps test how well current teaching and learning strategies and theories work and develop new ones; it helps support or contradict theories we might have about teaching and learning; and without it, we wouldn’t know which strategies are helpful or harmful to students.
But, if research forms the foundation of the strategies teachers need, why is it so difficult to understand? Well, there are a few basics to understand to make this less daunting. So, we’ve put together this guide to help you recognise a research paper that would be helpful for you and understand what it actually means…
Causation vs correlation
In research, relationships between factors are often described in one of two ways:
- Causation: where an action (X) directly causes an outcome (Y)
- Correlation: where action (A) is associated with another action (B)
So what might this look like in real life? Let’s say that a group of researchers find that the more people buy ice cream, the more they buy sunglasses. This is a correlational relationship, because although ice cream sales do not cause sunglasses sales, there is still an association between the two.
But let’s say that ice cream sales and sunglasses sales are both caused by sunny weather. The weather causes people to buy more ice cream and sunglasses; this is a causation.
Generally speaking, correlational research is not as reliable as causal research. An indicator of a good and valid study is when it is able to prove that X directly caused Y.
When you are reading research, consider whether the results are correlational or causal before you share the findings with you students. For example, if a study suggests that more time spent studying caused higher grades, you can tell your students that one way to improve their grades might be to study for longer. However, if the research only shows a correlation between hours spent studying and grades, then there is simply an association between the two. In this case, recommending that students study for longer periods of time to obtain a better grade may not be useful.
Find this graphic and many more in our latest book, Teaching & Learning Illuminated!
One factor that often varies between research studies is sample size, which is the number of participants in a study, usually represented by “n”.
Choosing a sample size is extremely important in research. Although they’re not always possible to obtain, large sample sizes make data more accurate and reliable, as researchers can assume that any of the study’s findings can be generalised of the rest of the population too.
For example, if 1,000 students perform better when interleaving revision instead of blocking, it is quite likely that other students will also benefit from the same strategy. You can therefore assume that the findings of the study will also be true for your students.
However, this doesn't make smaller studies unreliable or inapplicable – sometimes, these allow for more detailed findings. But more importantly, a large sample size with little or no diversity among subjects may not be applicable to your students if they don't fit the same criteria.
Research will often describe the results of their studies using a standardised effect size, which is the strength of the relationship between two variables.
Effect sizes are calculated using a statistical equation, and give a value between 0 and 1. The bigger the effect size, the stronger the relationship between the variables. For example, a strong relationship would have an effect size of around 0.8 or above, whereas a weaker relationship would have an effect size of around 0.2.
When reading research, you should aim to look for papers with an effect size of around 0.5 – but remember that a small effect size is better than no effect size at all. If a low-effort intervention has a small effect, it may still be worth your time.
Have a look at the p value (probability value) as well, a number between 0 and 1 which indicates how likely it is that the effect was the result of chance. Most of the time, researchers use a significance level of 0.05 or less, which means there is a less than 5% probability that chance caused the effect. The closer to 0, the more statistically significant the findings.
Cognitive scientists may choose to conduct their research in different ways, depending on what they are trying to find out. For example, if a researcher wants to investigate the use of spacing in the classroom, they may wish to conduct the research in a classroom setting. Let’s take a look at some types of study designs…
Research in a psych lab
A laboratory experiment is research that is conducted in a highly controlled environment. The researchers are able to manipulate the conditions of the experiment however they like. They decide where the experiment will take place, and a chosen set of circumstances.
Laboratory experiments are often researchers’ preferred study design; they are the only type of research that allows them to control for external variables that may interfere with the results of the study. For example, if researchers are investigating whether retrieval practice improves test performance, external factors such as the temperature of the room or how much sleep the participant got the night before may affect performance on the test as well. Lab experiments are able to control these factors, and therefore allow researchers to conclude causal relationships, which are highly reliable in the research world.
However, be careful when generalising the results of lab experiments to your students. One drawback for this type of study design is that the artificial conditions of the lab may cause participants to behave in a different way to how they would in real life. This makes it is more difficult to generalise results to real life.
Research in the classroom
While research that is conducted in a setting such as a classroom may not be as reliable in terms of causation, it is more likely to reflect your real-life experience, making results more applicable to the wider population. Generalising this type of research to your students is usually better than generalising the results of lab experiments.
However, using this study design means that researchers have less control over external variables that might affect how valid the research investigation is.
A meta-analysis is where researchers collate a large number of similar studies and use statistical analysis to determine how impactful a strategy might be. These studies do not conduct any experiments themselves – instead, they give an overview of the existing literature on a particular topic, which will give you a more complete view of a certain strategy.
Longitudinal studies investigate individuals or a group of people over a long period of time. This can range from years to decades, and looks at how a particular behaviour changes throughout the study. This study design allows researchers to compare participants to themselves rather than to other participants over time.
Longitudinal studies are useful when researchers want to investigate the effect that time has on a certain factor, such as the impact that consistently exercising has on student achievement.
Although it is important for testing old ideas and developing new ones, cognitive science research is not intended to replace teacher judgement. Instead, it exists to help inform it. One study cannot give a definitive answer, but taken as part of a collection, it can help paint a picture to provide guidelines for how that might work best.