There are many types of correlational research. The commonality among all types of correlational research is that they explore relationships between variables. Where descriptive research only described what was going on, correlational research talks about the link between different things. It is important to understand that correlational research does NOT tell us that variable A caused Variable B, but rather that they are somehow related.
Correlational Study Example
For example, if I told you that there was a correlation between domestic violence (violence between a family members) and bowling you would look at me strangely. But there is a relationship between the variables (variable 1- domestic violence, and variable 2- bowling). As more people bowl in the US, more domestic violence occurs, which is correlational research.
Does that mean in this correlational study example that bowling causes domestic violence- like you had bad game and take it out on a loved one. Or domestic violence causes bowling- like you fight with a sibling and feel the need to take it out on some pins. As you have already guessed- one does not cause the other to occur, but they are related- for every time people bowl, I can predict that domestic violence will go up, and every time domestic violence goes down I should be able to find a lane at the local bowling alley. There is a hidden variable that links both of them together. In this case it is winter time. In the winter more people bowl and more people stay in their homes (which increases the chances of domestic violence).
Direction of a Correlation
Before we examine the different types of correlational research methods, understand that correlations can go in two directions; positive and negative.
Positive Correlation: when two variables go in the SAME direction. For example, domestic violence and bowling. When bowling goes up, so does domestic violence. When domestic violence decreases, so does bowling.
Negative Correlation: here the two variables go in DIFFERENT directions. For example, consumption of garlic and dating (now I am making this one up). The less garlic you eat, the more you date. The more garlic you eat, the less the date. One variable going in one direction can be used to predict the other variable going in the opposite direction.
Scientists measure the strength of a correlation by using a number called a correlational coefficient. Now you do not have to know how they get the number, but you should know what it means when you see it. The number range from -1 to +1. If two variables (like studying and grades) have a correlation above zero (like +.76) then you have a positive correlation and the more you study, the better grades you have. The the number is below zero (like -.42) then you have a negative correlation and when one variable goes up the other goes down (like garlic and dating). If two variables have a correlation of zero then they have NO relationship with each other. The closer the numbers go to either +1 or -1, the stronger the correlation. The strength has nothing to do with whether the number is positive of negative. A correlation of -.88 is stronger than one that is +.56. the closer the number gets to zero (whether positive or negative), the weaker the correlation.
Types of Correlational Studies
There are many different ways to show a correlation between two variables. Let’s discuss some of the more popular ways; the survey method and naturalistic observation.
The Survey Method
Perhaps the most common type of research around is survey research. Every time you receive a letter in the mail asking you to take a minute and answer a few questions, or get a phone call begging for ten minutes of your time to speak about how you feel about ??????, you are experiencing the survey method of research. All surveys have one thing in common, they ask questions.
Now there are good and bad things about surveys in research. The good- no matter how you do it, internet, mail, phone, in person- they are fairly cheap. You can cover large populations of people easily if you use the phone or internet. The bad aspects of surveys is that 1. the response rate is REALLY low (for every 100 mailing you send out, you will be lucky to get one back). Second, people can lie on the survey so you can always question the validity of your data.
Let’s break down the survey method as a tool of correlational study. Pretend our hypothesis was the more garlic people eat, the less they date. First, we have to come up with some survey questions (pretend they ask about the amount of garlic one has eaten in the past 6 months and how much they have dated in the past sixth months). Hopefully, when people answer the survey, we will see that people who have stated that they have eaten a lot of garlic have also answered that they have dated less (a negative correlation).
But who are we going to give the survey to? As with ALL types of studies (except some case studies) we must choose a sample of people to take the survey (a sample is just a group of subjects). We have to first identify a population of people from which we are going to get the sample. The population includes anyone who can possibly be chosen to be part of the sample. If we are studying anorexic women and their dating habits we would choose a sample from a population of anorexic women (asking a chubby dude like me would not make sense for an anorexic study so I would NOT be a part of the population). In the case of garlic and dating, I am going to limit my population to single men and women between the ages of 18-25 from the Westchester area (if I do not limit my population, then I would have to start contacting people from all around the world).
Now, how do I pick people to be a part of my sample. Do I call all my single buddies in the Westchester area and give them the survey? That would not be a very fair way of doing it. To make the survey valid I MUST randomly select a sample from the population. Random selection means that every person in my population has an equal chance of being selected for the survey. If I can do this, then my sample has a greatly likelihood of actually representing the larger population I am studying. How do I randomly sample my population- I can randomly pick names out of a phonebook (but in a way that is unfair to single people in Westchester who do not have phones)- in other words, finding a truly random sample is not easy.
Another correlational research method is called naturalistic observation (although you can also use it as a descriptive research tool as well). Naturalistic observation is when a researcher attempts to observe their subjects in their natural habitats without interacting with them at all. Pretend I had a hypothesis; marijuana increases hunger (munchies). If I wanted to use naturalistic observation I would find a bunch of pot users and watch them. I would follow them around to parties, watch them smoke, and then see if they eat.
I would never interact with them- but just watch. If I see that every time a pot user smokes they eat, I could claim that smoking and eating are related, but I would NEVER know if the smoking caused the eating (it could be one of a million other things). Once again, at most these types of studies show correlation. The pinnacle of all science if is prove causation.
What Is Correlational Research Used For?
Although they cannot resolve causality, correlational studies are pretty useful. Some variables, such as a disability or mental illness, cannot be ethically manipulated. You can’t ethically give someone depression, for example, even if the intent is to help them get better – this is simply not unethical or moral. Other variables that cannot be manipulated are birth, sex, and age.
Correlations are also useful for making predictions. Once a psychologist recognizes that two variables, A and B, are connected to each other, he or she can make a more precise measure of one from the other. Knowing how much of a change in variable B is produced by a change in variable A allows the psychologist to predict the change in B just by knowing the value of A.
Thirdly, when ethically and morally appropriate, correlational evidence can lead to hypotheses and experiments. A psychologist may want to figure out if a third variable, C, is involved or if A causes B. The opposite could also be true: B might be causing A. had the correlational research not been done, the relationship would not have been detected. This type of research is helpful in many ways.
Final Words on Correlational Research
Just like every other research method, correlational research has its pros and cons. There is plenty of hypotheses that this type is useful for and there is a time and a place to apply it. Understanding what this type is all about can help you to understand it’s application in everyday life, as well.