Research methods

Qualitative data is non-numerical data, most generally words. Qualitative research gathers qualitative data in the form of text, video, audio, etc… This is often done via interview, focus groups, or open-ended questions in surveys. Qualitative data is extremely useful to get an in-depth understanding of a situation or of the experience of a group of people. It can also be used to identify areas needing more research. However, it is important to remember that qualitative data is subjective, and it may not me generalisable to other groups of people or other situations. 

Quantitative data is numerical data. Anything that can be counted (for example the number of correct responses in a test), or measured (for example brain activity), or scored (for example the level of satisfaction when using a service), can be used as quantitative data.  Quantitative data is useful to compare groups, conditions, situations, points in time, or to predict things, using statistical analysis. We can use standardised methods to collect the data, which means that other researchers can then repeat the same method to verify the findings. However, it is important to remember that quantitative data generally focuses on very specific questions, and measures things as if they existed on their own, while in reality they exist in a given context or a given environment.

A cross-sectional study looks at data from a single point in time, like a snapshot. This type of study is descriptive, meaning that researchers can “describe” the things they are interested in, but they can’t manipulate them, and they can’t precisely understand how they are related to each other.

A longitudinal study looks at how data changes over time. Depending on what the researcher is interested in, they can measure the things they are interested in over several days, weeks, months, or even years. These studies can be very long and extremely expensive, but they can allow researchers to see how different things relate to and interact with each other, over time.

In an observational study (as opposed to intervention studies, detailed below), the researcher only “observes” the thing being researched, but does not manipulate it. For example, a researcher can measure a mental ability in one or several groups of people, only once (cross-sectional design) or several times over a long period (longitudinal design), and they can describe and compare groups or time points (case-control studies or cohort studies), but they can’t control or manipulate the groups, and they can’t give different interventions to each group.

A prospective study is a specific type of longitudinal study, where we are not simply observing how data evolve in time, but where we are actually on the look out for the apparition of specific thing (or event, or characteristic, depending on the field of research). Once this has appeared, we can go back to the data gathered during the study to identify what could have predicted that the thing was going to happen. A retrospective study looks into the future: we collect data about causes, and we wait for the outcome. Unlike the retrospective study, all the data collected during the study is specifically designed for the study, and gathers all the information the researchers will need, in a very consistent and standardised way.

This is used for example in the field of research related to early signs of autism. In these studies, researchers follow a large cohort of babies and toddlers, gathering data at several points during their development. When the children are old enough for an autism diagnosis assessment, clinicians check whether children in the group are autistic. Researchers can then look back on the data gathered during the study to try and identify whether some of the things measured could have suggested autism earlier in development.

A retrospective study is a specific type of cross-sectional study that is not only interested in how things are in the present, but instead at asking people about their past as well. Contrary to the prospective study, a retrospective study looks into the past: we already know the outcome, and we try and find the past cause in existing data (such as administrative or clinical registers). This type of study is often used in clinical research: for example, we know which patients at the hospital developed a particular disease, and we look back into their medical files to see if there were earlier hints that this disease would happen. The downside of this type of study is that researcher can only look back at already existing data that, most of the time, was not collected for the purpose of research at all. Researchers then have to use data that is often incomplete and inconsistent.

In an intervention studies (unlike observational studies), the researcher tests the efficacy of a particular intervention, therapy or treatment, by manipulating the groups of participants. Clinical trials are a type of intervention studies used for clinical purposes. You can read more about certain types of clinical trial, randomised controlled trials and double-blind randomised controlled trials, below.

A randomised controlled trial (RCT) is a type of study used to assess the usefulness of a new treatment, intervention, or drug in a rigorous way. In this type of study, researchers recruit participants who are broadly similar, and randomly assign them to 2 (or more) groups. One group is the experimental group, and receives the new intervention. The other group is the control group, and receives a dummy intervention or placebo. We then follow up the groups to see whether the new intervention is better than the dummy intervention to achieve the goal. It is important that both groups follow exactly the same procedure, with the only difference being the treatment or intervention tested. That way, we can be sure that any difference between the groups at the end of the study is due to the treatment or intervention. The “random” assignation to each group is extremely important, as it is the best way to make sure that the results of the trial are not biased by the way the participants are selected.

For example, when scientists develop a new medication against allergies, they want to know if the new medication is better than no medication at all, and better than all the best available medication. They can split participants in 3 groups, one group takes the new medication, one group takes a fake medication that doesn’t contain any active drug at all (a placebo), and one group takes the best existing medication. It is important to have the group that takes the placebo, because sometimes, just thinking that you have taken a medication can make you feel better, even if in reality you haven’t. This is called the placebo effect. At the end, if the group with the new medication feels better than both other groups, then it means that the new medication is indeed better.

A double-blind randomised controlled trial follows the same strategy than the randomised controlled trial, but with an added step to make sure the results are robust. In this type of study, the participants are again randomly assigned to groups, with one group receiving the new treatment or intervention that is being tested, and the other group receiving a dummy treatment or intervention (placebo). However, in this type of study, neither the participants nor the researchers know who is receiving which treatment, the new one or the dummy one. For example, participants only know they are receiving a treatment called A or B, but they don’t know whether it is the new treatment. In turn, researchers analysing the data will just know that they have a group A and a group B, but they will run the entire analysis without knowing which is which. They will be able to get this information from another person involved in the study only at the very end. That way, they are not biased towards analysing the data in a way that favours the new drug, and they are forced to remain completely objective.

This type of study can be upsetting for participants who want to make sure they are receiving the new treatment, not the dummy one, and researchers understand this. However, this double blinding (blinding of the participant and the researcher) is extremely important to make sure the new treatment really works. It is the best way to make sure that everybody, participants and researchers alike, are completely objective when judging the efficacity of the treatment.

In a study measuring the usefulness of a new treatment or intervention (like a clinical trial) on something (a skill, a characteristic, or a symptom, for example), researchers measure the skill (or characteristic, or symptom) before the treatment (or intervention), and after. This first measurement before the treatment is called the baseline.

“Blinding” is what happens for example in a blind taste test where, for example, a participant has to taste two pieces of brownies, one homemade by a friend and one store-bought, but doesn’t know which is which, and has to decide which tastes better. If the participant knew which was homemade by their friend, they would probably think this piece tastes better, even unintentionally. Whether we are aware of not, knowing things (like which cake is homemade) influences the way we perceive other things (like the taste of the cake), even when we try our best to remain objective. This is why sometimes in research, not knowing is the best way to make sure we remain objective. Blinding can be used for participants (who then are not told if they are doing the experiment A or B), and for researchers (who, when analysing the data, don’t know which group of participants did which experiment, and are only told at the end by another researcher). You can read more about this in the section “Double-blind randomised controlled trials”.

A control group is the standard, or benchmark, used to compare the results of an experiment. In clinical trials, one group, the experimental group, receives the new treatment, while the control group receives a dummy treatment. By comparing both groups at the end of the study we can make sure that any difference is only due to the new treatment, but not to other aspects of the study, such as time (because things like symptoms can change through time), taking part in a study (because people who take part in studies receive loads of attention, which can have some effect), or believing taking a treatment (because even taking a dummy treatment can have some effect, this is called the placebo effect). We also need control groups in non-clinical research. For example, if a researcher wants to know whether listening to classical music the night before a school exam influences the exam grade, the study would include a group that listens to classical music, and a group that does not: the control group.

Control measures are used to take into account factors that might influence the results of the experiment. Let’s take again the example of the researcher who wants to know whether listening to classical music the night before a school exam influences the exam grade (see “Control group” above). This study would include a group that listens to classical music, and a group that does not: the control group. However, there are many things that can also influence the exam grade, and all these also have to be measured and accounted for in the study. For example, the researcher will also have to measure the number of hours of sleep the student had, their level of stress regarding the exam, and how well they have prepared for the exam. Only by measuring all these confounding factors can the researcher make sure that the effect found at the end is due to listening to classical music. In clinical or psychological research, this can mean measuring memory abilities, IQ, or weight, depending on what is relevant for the study. It takes some more time during the appointment with the participant, but these measures are essential to make sense of the results.

The outcome is what is used to evaluate the results and success of a study. Outcomes always depend on the study. For example, a clinical trial for a new drug against headaches will have as outcome “intensity of headaches”.  A study interested in measuring things influencing exam grades will have “exam grades” as outcome.

Neuroimaging is a technique to capture, measure, and visualise how the brain and the rest of the nervous system work. Structural imaging is used to study the anatomy of the brain (its structure), while functional imaging is used to study how the brain activates (and deactivates), reacts, and functions. There are many types of neuroimaging techniques, but at the Patrick Wild Centre we mostly use MRI and EEG.

Magnetic resonance imaging (MRI) uses magnetic fields and radiowaves (but no X-rays and no radioactive tracers) to produce two- or three-dimensional images of the brain. Functional magnetic resonance imaging (fMRI) builds up on MRI images, and uses the changes in oxygenation of the brain blood vessels to show which parts of the brain activate and deactivate one the person is doing one action compared to another action.

Electroencephalography (EEG) is a method to record the electrical activity of the brain with a very high precision in time, but a lower precision in space. It is generally non-invasive, meaning that to do an EEG the participant does not need to receive any injection. It uses electrodes (small electricity sensors) able to sense the electric changes that everybody has on their skin all the time. To conduct an EEG, we place on the participant’s head a cap containing these electrodes, that will record the electric changes on the scalp (reflecting the electric changes at the surface of the brain).