You are here

Correlational Research

15 February, 2016 - 10:55

Correlational research is designed to search for and test hypotheses about the relationships between two or more variables. In the simplest case, the correlation is between only two variables, such as that between similarity and liking, or between gender (male versus female) and helping.

In a correlational design, the research hypothesis is that there is an association (i.e., a correlation) between the variables that are being measured. For instance, many researchers have tested the research hypothesis that a positive correlation exists between the use of violent video games and the incidence of aggressive behavior, such that people who play violent video games more frequently would also display more aggressive behavior.

media/image10.png
Figure 1.10 Correlational Design. 
The research hypothesis that a positive correlation exists between the use of violent video games and the incidence of aggressive behavior 

A statistic known as the Pearson correlation coefficient (symbolized by the letter r) is normally used to summarize the association, or correlation, between two variables. The Pearson correlation coefficient can range from −1 (indicating a very strong negative relationship between the variables) to +1 (indicating a very strong positive relationship between the variables). Recent research has found that there is a positive correlation between the use of violent video games and the incidence of aggressive behavior and that the size of the correlation is about r = .30 (Bushman & Huesmann, 2010).

One advantage of correlational research designs is that, like observational research (and in comparison with experimental research designs in which the researcher frequently creates relatively artificial situations in a laboratory setting), they are often used to study people doing the things that they do every day. Correlational research designs also have the advantage of allowing prediction. When two or more variables are correlated, we can use our knowledge of a person’s score on one of the variables to predict his or her likely score on another variable. Because high-school grades are correlated with university grades, if we know a person’s high-school grades, we can predict his or her likely university grades. Similarly, if we know how many violent video games a child plays, we can predict how aggressively he or she will behave. These predictions will not be perfect, but they will allow us to make a better guess than we would have been able to if we had not known the person’s score on the first variable ahead of time.

Despite their advantages, correlational designs have a very important limitation. This limitation is that they cannot be used to draw conclusions about the causal relationships among the variables that have been measured. An observed correlation between two variables does not necessarily indicate that either one of the variables caused the other. Although many studies have found a correlation between the number of violent video games that people play and the amount of aggressive behaviors they engage in, this does not mean that viewing the video games necessarily caused the aggression. Although one possibility is that playing violent games increases aggression,

media/image11.png
Figure 1.11 Playing violent video games leads to aggressive behavior. 

another possibility is that the causal direction is exactly opposite to what has been hypothesized. Perhaps increased aggressiveness causes more interest in, and thus increased viewing of, violent games. Although this causal relationship might not seem as logical, there is no way to rule out the possibility of such reverse causation on the basis of the observed correlation.

media/image12.png
Figure 1.12 Increased aggressiveness causes more interest in, and thus increased viewing of, violent games. 

Still another possible explanation for the observed correlation is that it has been produced by the presence of another variable that was not measured in the research. Common-causal variables (also known as third variables) are variables that are not part of the research hypothesis but that cause both the predictor and the outcome variable and thus produce the observed correlation between them (Figure 1.13). It has been observed that students who sit in the front of a large class get better grades than those who sit in the back of the class. Although this could be because sitting in the front causes the student to take better notes or to understand the material better, the relationship could also be due to a common-causal variable, such as the interest or motivation of the students to do well in the class. Because a student’s interest in the class leads him or her to both get better grades and sit nearer to the teacher, seating position and class grade are correlated, even though neither one caused the other.

media/image13.png
Figure 1.13 Correlation and Causality.  
The correlation between where students sit in a large class and their grade in the class is likely caused by the influence of one or more common-causal variables.  

The possibility of common-causal variables must always be taken into account when considering correlational research designs. For instance, in a study that finds a correlation between playing violent video games and aggression, it is possible that a common-causal variable is producing the relationship. Some possibilities include the family background, diet, and hormone levels of the children. Any or all of these potential common-causal variables might be creating the observed correlation between playing violent video games and aggression. Higher levels of the male sex hormone testosterone, for instance, may cause children to both watch more violent TV and behave more aggressively.

You may think of common-causal variables in correlational research designs as “mystery” variables, since their presence and identity is usually unknown to the researcher because they have not been measured. Because it is not possible to measure every variable that could possibly cause both variables, it is always possible that there is an unknown common-causal variable. For this reason, we are left with the basic limitation of correlational research: correlation does not imply causation.