I’m trying out this glossary thing. It’s very nascent, but imagine it would contain all the technical terms I use. Do let me know if it’s helpful! :)
A 2x2 design is one of the classics of psychological experimentation. The basic research is about correlation or prediction, and involves one predicting variable (or independent variable) and an outcome. The next step is looking at two predicting variables and how they affect the outcome. For example, let’s say you want to know how the number of crayons and the presence of art on the walls influence the child’s production of drawings. In a 2x2 design, you take these two factors and manipulate them. For instance, you can give some of the children 3 crayons and some of the children 24 crayons. You can also sit some of the kids in a room with no art on the walls and some of the kids in a room with lots of art on the walls. If you cross these two you get a 2x2 design, with 4 groups: group A would have 3 crayons and no art on the walls; group B would have 3 crayons and lots of art on the walls; group C would have 24 crayons and no art on the walls; and group D would have 24 crayons and lots of art on the walls. If you count the number of drawings each of the groups produced, you can examine the effects of each of the factors, but you can also look at the interaction between the two factors.
Here is an example of a study that used a 2x2 design: Museum learning study. There are many studies that used a 2x2 design, obviously, but none that I’ve referred to yet.
This is a statistical term that sounds complicated, but really is not. When we have two variables and a third variable explains the relation between these two variables, we call the third variable a mediator. So, for example, we know that parents from lower socioeconomic status tend to resort to more harsh parenting behaviours. So, without looking any closer, it could appear that parents from lower SES just choose the wrong behaviours–and the road to parenting classes is quite short from there. However, the variable that mediates or explains this link is psychological distress: parents from lower socioeconomic status experience more psychological distress, and parents with more psychological distress tend to exhibit more harsh parenting behaviours. As soon we enter psychological distress into the equation, it explains the link between low socioeconomic status and harsh parenting behaviours. So, in other words, once psychological distress is in the equation, there is no longer a statistical link between SES and parenting behaviours. Parents from lower SES do not need parenting classes (any more than parents from high SES do), they need less psychological distress.
A study of studies. Basically, the scientist collects published studies and runs statistics on all the data that was used in all the studies. There are certain limitations to this method, but it’s a powerful tool because it allows us to see whether a certain effect is consistent across studies. You can also divide the studies along certain factors (for instance, what country were they conducted, the age of the children in the sample, etc.) and see if these factors influence the results.
Here are some examples of meta-analyses:
- The cool nature-nurture study
- The parents and bullies study
- The staying-at-home vs. working study
- The popularity study
A pre-post design is very simply a design in which you measure a target behaviour before and after the intervention. Think of the before and after pictures of people who have done the magic diet of the day. The problem with this design is that it does not account for other variables. For example, let’s say you are measuring the effects of giving your child a candy after every time she goes potty. You measure the number of times she goes potty in a day before you begin (the pre-intervention measure, or baseline), then you run the intervention (say, 2 weeks of candy every single time she goes potty). You then measure the number of times she goes potty in a day again (the post-intervention measure), and see whether it increased. The problem with this design is that the passage of time can account for some of the change (your girl is 2 weeks older, and that is a lot of time when you are 2 or 3 years old). Just by doing a pre-post measure you cannot rule out that the child is just a bit older, more mature, or whatever. So, you would have to have a control group (participants that do not get the candy intervention)–and preferably randomly assign them into the groups–in order to reach any conclusions about the effectiveness of the intervention.
This is when you assign participants in your study into groups. The groups differ along what you are trying to study. For instance, if you want to study the effect of colour on mood, you would assign your participants randomly (that is, not based on any of their traits such as age, gender, or personality) into three rooms: the blue room, the red room, and the green room. Randomly assigning people to the different groups means that the groups should be equal on other factors (such as age, gender, and personality), and, therefore, you don’t really need to worry about these other factors influencing your findings. You would then measure your outcome variable (mood), and if, for instance, people who have spent an hour in the green room are happier than people who have spent an hour in the blue room, you have evidence to support the idea that colour influences mood.
Here are some excellent studies that have used random assignment, and are therefore awesome (generally speaking):
- The “helpful babies” study
- The “sign language yes or no” study
- The “reliable marshmallow task” study
When something is reliable it works in the same way for a long(ish) period of time. In psychological testing, reliability is the tendency of people to get similar scores on the test. Think about a personality test, and assume that personality is a stable trait (i.e., we don’t change our personality from day to day). A good personality test would provide us with the same results regardless of when and how many times we take it. So, to test for reliability, researchers typically have participants do the same test twice, with a certain period of time in between. If the scores are similar between the first time you take the test and the second time you take the test, the researchers conclude the test is reliable.
Socioeconomic Status (or SES) is an index that is typically composed of things such as people’s education level, income level, and sometimes the neighbourhood they live in. In developmental research, parental education is usually the forefront of these variables, as it has a significant impact on children’s outcomes.
Transference is a technical term in the field of learning. It refers to the ability to generalise from learning a skill in one context to solving a different task with the same skill. Researchers are interested in transference because it’s a strong test for interventions. Let’s say I have an intervention that is supposed to improve children’s memory. Ideally, I would have a task that requires memory but is not very similar to the intervention itself. I would give children in the control and experimental group this task before and after the intervention period, and if children in the experimental group (who actually got the intervention) improved on this task more than the children in the control group (who, ideally, did something else for the same amount of time, something called active control), I have support for the idea that children can transfer the skill they learned in the intervention to other contexts. Researchers typically talk about near-transference and far-transference. So near-transference would be generalising to a context that is fairly close to the context of the intervention, like in the example I gave just now. Far-transference is the more interesting and practical issue: this would be transference to areas that are not as close to the intervention but are of interest and related to the intervention area. For instance, if the children in my experimental group also showed an increase in school tests scores following the intervention, that would be far-transference and much more interesting than improvement on a task in the lab.