Correlation does not always imply causation

25 Mar

There are many instances in psychological research when results seem to show significance between one variable and another, this is called a correlation. However, it is often wondered whether these correlations also show causation e.g. variable A caused variable B to happen. Sometimes results show significance between variables and it is assumed that causation has occurred, even though this is not always the case. However, how can one be sure about whether variables show causation or whether there is something else causing it? The name for assuming that one variable causes another is called cum hoc ergo propter hoc meaning “with this, therefore because of this” is commonly assumed when two things occur together that one must have caused the other, however this is not necessarily the case.

The cum hoc ergo propter hoc can be explained in many different ways;

For example,

  • If variable A causes a correlation in variable B, therefore A causes B.
  • A may be the cause of B.
  • B may be the cause of A.
  • There could be an unknown variable C, that causes A and B.
  • There may be relationships between all above variables in ways unexpected from the hypothesis.

However, despite all the possibilities there is so way to make a conclusion regarding correlation and causation, these issues need further research to determine if the correlation does mean that there is a causation.

When it comes to psychological research, it is always difficult to determine causation. Even if results are significant, it does not necessarily mean that there is causation between your variables. Psychologists are only able to present what potentially could be happening and not necessarily what is happening. However, as this is down to science these hypothesis can always be tested further, and if a psychology was really unsure of their results further testing can always help provide more answers.

 

http://en.wikipedia.org/wiki/Correlation_does_not_imply_causation

http://stats.stackexchange.com/questions/36/correlation-does-not-mean-causation

Advertisements

10 Responses to “Correlation does not always imply causation”

  1. laurencedown 03/29/2012 at 14:08 #

    It doesn’t matter so much that correlation doesn’t mean causation does it? I mean, if you are controlling for as many confounding variables as possible and you have good control over your independent variables, an extremely strong correlation could well predict causation. Although technically correlation cannot show causation, I reckon it can give us pretty much enough to go on to make up our minds!

  2. Psucc0 04/08/2012 at 13:16 #

    Although I completely agree with you, Correlation isn’t all bad its good at hinting at further research. You can compare the variables to have a look and see if there is anything worth investigating between them, I guess that’s the main point of corelational research. Although the media always seem to miss the point!
    paper http://www.criticalthinking.org.uk/smokingparents/ – the antisocial teenagers and smoking parent’s example

    http://www.experiment-resources.com/correlation-and-causation.html

    Research for the behavioural sciences – Gravetter and Forzano

  3. psucd8 04/17/2012 at 18:48 #

    I think that although correlation cannot directly show causation between variables, it can often be the closest estimate a researcher can obtain. For example, in psychology there are many instances where it is unethical, or just not possible to control confounding variables therefore it is not possible to identify causation so correlation gives the best idea of the relationship between variables. However, Wright (1921) suggested that there can be correlation between variables they may actually act independently of each other but be acted upon by another variable. He also suggests that if a mathematical relationship can be identified between variables then correlation can be used to find the degree of determination of variables and thus calculate causation (using lots of long and complicated formulas).

    Wright (1921):
    http://www.ssc.wisc.edu/soc/class/soc952/Wright/Wright_Correlation%20and%20Causation.pdf

  4. psuc97 04/18/2012 at 16:38 #

    Helloo 🙂
    I agree that it is very difficult to conclude that there is a definite cause and relationship between two variables in psychology as it it almost impossible to control all extraneous or confounding variables. For example Loftus and Palmer conducted two studies looking at how information after an event effected the accuracy of recall of said event by having participants watch a video clip of a car crash in a controlled laboratory environment. They concluded the information after the event did effect memory recall, however there was no way of ensuring that all participants were in the exact same mood state or in perfect health. Also there may be other variables that would come into question if the participants actually viewed a real car crash such as shock which may have a bigger or different impact on memory.
    In addition, correlation studies are conducted in a way that means the variables being observed and doing so in there natural environment which means there is an even lower level of control that if an experiment looking for a cause and effect relationship was being conducted.
    Correlation studies aren’t pointless though, they allow researchers to narrow their studies and bring to light the areas that need further researching.:)

  5. poeywycheung 04/18/2012 at 23:18 #

    I agree with you that Correlation does not imply causation.
    To me, correlation is more like suggesting a relationship between the 2 variables whereas causality is talking about the distinct/absolute relationship between the 2 . Just because two events/variables correlate does not mean that one has caused the other.
    And the wrong thinking(correlation=causation) cause a problem with quantitative data and research. Although correlation shows significance,it cannot tell the reason/why in between. Therefore we have to remind ourselves to be more empirical, scientific whenever we have to deal with these kind of researches.

Leave a Reply

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out / Change )

Twitter picture

You are commenting using your Twitter account. Log Out / Change )

Facebook photo

You are commenting using your Facebook account. Log Out / Change )

Google+ photo

You are commenting using your Google+ account. Log Out / Change )

Connecting to %s

%d bloggers like this: