Correlational Research

When I was growing up, one of my favorite video games was Mortal Kombat.  One of the trademark features of the game are grisly, blood-soaked finishing moves, involving plenty of pixelated gore.  My parents were horrified, of course.  Would playing these games make us children grow up to be violent and aggressive? 

It should be apparent that we could test the effects of violent games through a simple experiment.  We would recruit a number of children, and randomly assign half of these children to play violent video games (like Grand Theft Auto) for several hours each day, while the other half would play only non-violent games (like Tetris).  Then, over a period of several years, we would measure whether the children in the "violent games" group show more signs of violence - like getting into fights at school.  If so, we would finally have solid evidence that violent video games are, in fact, harmful.

Of course, such an experiment has never been done, and never will be done.  Why not?  Well, if you were a parent, would you agree to let your child participate in this experiment, knowing that they might be transformed into a violent monster?  Of course not.  This study raises huge ethical dilemmas.  Imagine if one of the children in the "violent games" group seriously hurt or killed another child!  Whose fault would that be?

There are many situations where experiments are ethically or practically impossible to carry out.  In these situations, psychologists often turn to correlational studies - research that simply measures the association between variables, without actively manipulating (changing) any of the variables.

Video Activity

Watch the video below, describing a large-scale study on violent video games carried out in Singapore, and answer these questions.

1.  What variables were measured in this study?

2.  What association did the researchers find?

3.  Is this an experiment?  Why or why not?


The research study in video is a correlational study, because it measures the extent to which two variables are related.  In this case, the variables measured are time playing violent games, and aggressive thinking.  There are three possible outcomes:

  • An increase in one variable (X) tends to be associated with an increase in the second variable (Y).  This is called a positive correlation.  An example would be number of years of school (X) and income (Y), since highly educated people tend to earn more money.

  • An increase in one variable (X) tends to be associated with a decrease in the second variable (Y).  This is called a negative correlation.  An example would be physical activity (X) and blood pressure (Y), since higher levels of exercise is linked to lower blood pressure.

  • An increase in one variable (X) has no relation to a second variable (Y).  This is called no correlation.  An example would be height (X) and intelligence (Y), since there is no relationship between the two.

Graphing the two variables can show you what kind of correlation exists, as shown below:
Correlational studies can also compare people who naturally fall into one group or another, for instance comparing the musical intelligence of men versus women, or the punctuality of Japanese versus Spanish students.  This is also correlational data, except that one of the variables is a category (for instance, gender) rather than a numerical value.  For instance, the bar chart below shows a correlation between race & gender (variable X) and income (variable Y). 
At first glance, experiments and correlational studies seem to be quite similar.  However, there is a subtle (but very important) difference between them. 

In an experiment, the researchers manipulate (change) one of the variables (the IV), and measure the effects on another variable (DV), while keeping all other variables constant.  This demonstrates a clear, "cause and effect" relationship between the IV and the DV. 

On the other hand, in a correlational study, researchers do not manipulate (change) any variables.  They simply measure variables to see associations between them.  Understanding why there is a link is often less straightforward in a correlational study.  For instance, in the graphic above, its clear that gender and race are related to earnings.  But why Hispanic women make so much less money than White men cannot be answered from the graph alone.

In the video you saw earlier, on violent games in Singapore, the researchers asked teens to self-report how often they play violent games, and then measured aggressive thinking.  This is an example of a correlational study, as researchers simply measured two variables.  An experiment, on the other hand, would require researchers to randomly divide teens into two groups, with one group being instructed to play violent games, and the other group not - this involves manipulating one variable (time spent playing violent games) to see the effects on a second variable (violent behavior).
Try it Out

Can you tell the difference between an experiment and a correlational study?  Read the research synopses below, and identiy which research method is used. 

A)  A study was conducted comparing the efficacy of cognitive behavioral therapy (CBT) and anti-depressants (Prozac) in treatment of depression.  Participants were randomly assigned to either receive 3 months of CBT therapy, or 3 months of anti-depressant medication.  It was found that both treatments were equally effective.

B)  A new study finds a link between eating omega-3 fatty acids and reduced risk for cancer.  Participants filled out a questionnaire assessing their intake of foods high in omega-3 fatty acids, and were followed up for 5 years.  Participants who ate high levels of omega-3 fatty acids had a 17% lower chance of cancer over the 5 year period.

C)  According to the latest research, elderly people who travel frequently have a lower risk of dementia and Alzheimer's disease.  People aged 70 + who spend at least 2 weeks overseas each year had a 34% lower risk of dementia than those who did not travel at all.

D) A study on bilingual children found that speaking 2 or more languages fluently is associated with better problem solving and abstract reasoning skills, as measured by standardized tests of cognitive reasoning.

E) An investigation on the effects of mindfulness meditation found that it had positive effects on academic performance.  Participants were Grade 8 students, who were randomly assigned to do either 10 minutes of mindfulness meditation, or 10 minutes of relaxation exercises each day.  After one year, the students who did mindfulness meditation had higher test scores and fewer behavioral problems.

​Scroll to the bottom for the answers
Checklist

  • I can define a correlational study​

  • I can explain positive, negative, and zero correlations

  • I can explain the difference between an experiment and a correlational study

  • I can classify a research study as being either experimental or correlational
Quiz Yourself

1.  A correlational study examines:

(a) Whether a change in one variable causes a change in a second variable

(b) The association between two or more variables

(c) Why two variables are related

(d) A cause-and-effect relationship between variables


2.  A survey on coffee drinkers found that drinking more coffee per day is associated with fewer hours of sleep.  This is an example of a:

(a) Positive correlation

(b) Negative correlation

(c) Cause-and-effect relationship

(d) Case study


3.  A study which compares the spatial reasoning of men and women is ___

(a) Experimental, because gender is the independent variable

(b) Experimental, because all variables are carefully controlled

(c) Correlational, because no variable is manipulated

(d) Correlational, because there is a positive association between variables


​4.  Which of the following is not a strength of correlational research?

(a) It can help avoid ethical issues that would be present in an experiment

(b) It often can be carried out in less time than an experiment

(c) It can study a large number of participants

(d) It demonstrates a clear explanation for why variables are related





Answers

Try it Out

​A) Experiment.  The independent variable is the type of treatment, which is manipulated to see the effects on recovery from depression.

B) Correlational.  No variables are manipulated.   A negative correlation is found between consumption of omega-3 fatty acids and risk of cancer.

C) Correlational.  No variables are manipulated.  A negative correlation is found between frequency of travel and risk of dementia.

D) Corerlatoinal.  No variables are manipulated.  A positive correlation is found between bilingualism and cognitive skills.

E) Experiment.  The independent variable is type of activity (mindfulness meditation or relaxation), which is manipulated to see the effects on academic performance


Quiz

1 - B, 2 - B, 3 - C, 4 - D