You are here:

Psychology/MA in Psychology


QUESTION: Dear Sir/Madam,
I want to ask about correlational method. Here is the copy and pasted paragraph from my textbook of Psychology.
"One of the most important points in understanding the result of correlation research is that finding a correlation between two variables does not in any way imply that two are linked causally.We can take the example of the possible relationship between television violence and viewer aggression. Because in most cases it is difficult to control adult viewers viewing habits, researcher must carry out correlation studies in which the aggressive content of television programs viewed by an individual is compared with the degree of aggressive behavior that person carries out.
Suppose the results are supportive of the hypothesis that high aggressive content is associated with high viewer aggression and that low aggressive content is associated with low viewer aggression. Drawing the conclusion that aggressive behavior caused the aggression would be inappropriate and quite possibly
It follows, then that although the use of correlation techniques allow us to learn what associations exist between two variables, it does not inform us about causality."
Please explain it in simple terms. I am unable to understand that  when the correlation coefficient of two variables would be positive it means there is relation between the two variables then how is it possible that it is not causative? Similarly, how drawing the conclusion that aggressive behavior caused the aggression would be inappropriate and inaccurate? Please help me.

ANSWER: In terms of the scientific method and statistical analysis of research results:  Causation can only be demonstrated by an experimental design in which the Independent variable is varied, all other variables are held constant, and the dependent variable is measured.  For example,  to test the effect of watching TV aggression on behavioral aggression:  the independent variable is TV aggression, the dependent variable would be behavioral aggression.  The experiment would require that you have two groups:  A control group that only watches nonviolent tv, an experimental group that watched violet tv.  All other variables must be controlled (held constant).  So, both groups must be matched by age, gender, educational level.  Both groups watch tv for the same amount of time, and watch no other tv.  Access to other violent influence must be controlled - for example, participants cannot read violent books, or go to violent sporting events during the experimental period.  Let's say you do this for 6 months, then observe behavior for 6 months.  The behavioral observation needs to be tightly controlled as well - all participants will be observed for 5 hours per week for 12 weeks in the same environment.    If, after conducting an experiment like this, the people who watched violent tv are more aggressive than those who didn't, you can conclude that violent TV impacted their behavior.

In a correlation, you are simply observing that two phenomenon co-vary together.  Because you did not control all the extraneous and confounding variables, you cannot attribute causation.  One example, I read an article that found that women who walk 8 miles per week live 20% longer than women who walk less.  Does walking increase their life span?   You cannot make that conclusion.  It is very possible that the women who walk more are healthier to begin with - that is why they are able to walk 10 miles per day.

I hope this helps you understand this.  If you are earning an MA in psych, it is extremely important that you understand that correlation is not causation.  This error is made in newspaper reports all the time, and it distorts scientific research, leading people to inaccurate conclusions.

[an error occurred while processing this directive]---------- FOLLOW-UP ----------

QUESTION: A Correlational Study of the connection between income and happiness shows earning more money is associated with greater happiness (positive correlation). Isn't in this case earning a lot of money(IV) also causally affecting the happiness(DV)? Can it be not said that some times correlation is also causal? Please help me.
Thanks and Regards

no, it cannot be said that sometimes the correlation is also causal without doing an experiment in which you control the variables.  You cannot make a statement about causation based ONLY on a correlation.    In the example that you cite, isn't it possible that people who are happy earn more money because they go to work every day, whereas people who are sad and depressed stay in bed all day and don't go to work?    You are making an assumption that money causes happiness, not that happiness causes professional success.   Indeed, there may be no causality at all.  For example, between 1970 and 1980 average SAT scores rose in the US.  At the same time, daily consumption of soda rose.   So, they are correlated - they VARY together, but I doubt that there is any causal relationship between soda drinking and SAT scores.  A good scientist looks at the logical flaws in the arguments and uses the scientific method to draw conclusions.  You can only conclude a causal relationship when all other variables have been ruled out - that is done in an experiment, not a correlational study.


All Answers

Answers by Expert:

Ask Experts


Katherine ONeill


I can answer academic questions about psychology. I am not a clinician (therapist), I am a research psychologist with expertise in biopsychology, general psychology, cognitive psychology, research methods and psychopharmacology.


I have 25 years experience as a researcher in health behavior, biopsychology and psychopharmacology.

Healthcare Businesswomen's Association

Applications of Market Research for Small Business UMBC Activate Program, March 2008 HIV/AIDS: An assessment of Need in the Continuum of Care. Optum Health Education., 12/2008 Maximizing the online medium for market research: Best practices. Market Research for Pharmaceuticals Conference, 12/06/2006 O誰eill, K.A. APD, ADD, ADHD and AD/HD: Personal and scientific reflections. Audiology Online, 6/6/2005. O誰eill, K.A. et al, Hyperactivity induced by NMDA injections into the nucleus accumbens. Pharmacology, Biochemistry and Behavior 34(4), Dec 1989, 739-745. O誰eill, K.A. and Liebman, J.M. Unique behavioral effects of the NMDA antagonist, CPP, upon injection into the medial prefrontal cortex of rats. Brain Research, 435(1-2), Dec 1987, 371-376. O誰eill, K.A. and Gertner, S.B. Effects of centally administered H2 antagonists on motor activity. Pharmacology, Biochemistry and Behavior. 264, 1987, 683-686. O誰eill, K.A. and Gertner, S.B. Effects of centrally administered H2 antagonists in the behavioral despair test. 90(2), 1986, 190-192. O誰eill, K.A. Chronic desipramine attenuates morphine analgesia. Pharmacology, Biochemistry and Behavior. 24(1), Jan 1986, 155 158. O誰eill, K.A. and Valentino, D. Escapability and generalization: Effect on 礎ehavioral despair. European Journal of Pharmacology 78(3), March 1982, 379-80. O誰eill, K.A. et al, An automated high capacity method for measuring jumping latencies on a hot plate. Journal of Pharmacological Methods, 10(1), Aug 1983, 13-18. O誰eill, K.A., Scott, C. and Weissman, A. Naloxone enhances nociceptive responding. Society for Neuroscience, Abstract 9: 274, 1983.

Ph.D. Experimental Psychology, University of Rhode Island, 1983. Post doctoral fellow dept of psychiatry, New York University Medical Center, 1983-1984. Post doctoral fellow, dept of pharmacology, University of Medicine and Dentistry of New Jersey, 1984-1985.

©2017 All rights reserved.

[an error occurred while processing this directive]