Correlation, Causation, and Extraneous Variables

​Correlation vs Causation

Analyzing Data

Correlation does NOT mean Causation.

• We are wired to see patterns, so when two variables come together on a chart your human instinct tells you to assume they are related.

Analyzing Data Continued

Correlation does NOT mean Causation.

• Confounding variables can make it seem like two things are related, when in fact there could just be a coincidence (spurious correlations).

• Understanding their differences will help you evaluate and interpret your research data.

​https://www.tylervigen.com/spurious-correlations

What's the difference?

Definition: an association between variables: when one variable changes, so does the other.

• Example: activity level is positively correlated with self-esteem.

Correlation

What's the difference?

Definition: changes in one variable brings about changes in the other (there is a direct cause-and-effect relationship between variables).

• Example: increasing the temperature of a system causes the molecules to speed up.

Causation

A correlation doesn’t imply causation, but causation always implies correlation.

Which one do you have?

• You’ll need to use an appropriate research design to distinguish between correlational and causal relationships.

  • Non-Experimental research designs can only demonstrate correlational links between variables, while Experimental research can test causation.

    • Observations = test for correlation

    • Labs = test for causation

Experimental Research

This type of research involves controlled experiments where the researcher manipulates the independent variable to observe the effect it has on the dependent variable.

Experimental Research

This type of research involves controlled experiments where the researcher manipulates the independent variable to observe the effect it has on the dependent variable.

• Independent Variable = changed or manipulated to test the effects

• Dependent Variable = being observed/measured in the experiment

• Extraneous Variable = not being investigated, but can potentially affect the outcomes of your data

  • Controls = extraneous variables the researcher has identified and chosen to keep constant (controlled during the experiment).

Experimental Research

This type of research involves controlled experiments where the researcher manipulates the independent variable to observe the effect it has on the dependent variable.

• In a controlled experiment extraneous variables are controlled as much as possible to minimize their influence on the variable you are interested in tracking.

RECALL - Non-Experimental

In research a confounding variable is an unmeasured third variable that influences the supposed cause and effect.

• Example: You collect data on sunburns and ice cream consumption. You find that higher ice cream consumption is associated with a higher probability of sunburn. Does that mean ice cream consumption causes sunburn?

Answer...

• NO! Here, the confounding variable is temperature

  • when it gets hotter outside more people eat ice cream and because the sun it out people spend more time outside (which results in more sunburns).

What should you do?

It is important to consider confounding variables before starting your research and account for them in your design to be sure that your results are valid.

• If you know your design type you can accurately conclude the type of data you've collected and draw correct conclusions about the data.