Correlational Research
2.5 Correlational Research
After the ethical collection of data, researchers must determine how to analyze and explore the information.
The approach chosen depends on the study design and the research question being investigated.
Observations, case studies, and surveys are common methods aimed at identifying relationships between two or more variables.
Correlation is one method to quantify relationships.
The focus is on identifying how one variable changes in relation to another.
Scatterplots:
A scatterplot is a graph that displays the relationship between two variables.
It features one variable on the x-axis (horizontal) and another variable on the y-axis (vertical).
Provides a visual representation of the relationships between variables.
Strong relationships are depicted by data points clustering closely around a straight line.
A correlation statistic, denoted as r, describes both the direction and strength of the relationship.
2.5.1 Direction of Correlation
Correlations can exhibit positive, negative, or zero directionality:
Positive Correlation:
Indicates that the variables move in the same direction.
Example: Height and weight; as height increases, weight also generally increases.
Negative Correlation:
Represents an inverse relationship: an increase in one variable leads to a decrease in another.
Example: Usage of computers during lectures negatively affects grades (Sana et al., 2013).
Zero Correlation:
Indicates no apparent relationship between variables.
Example: No correlation exists between vaccination early in life and autism development (Jain et al., 2015).
Scatterplots effectively illustrate directionality:
Positively Correlated Variables: Data points cluster from the bottom left to the top right.
Negatively Correlated Variables: Data points cluster from the top left to the bottom right.
Zero Correlation: Data points appear random, with no discernible pattern.
2.5.2 Strength of Correlation
Positive and negative values indicate direction, but do not reflect how closely variables are related.
The strength of a correlation is assessed using a correlation coefficient:
The correlation coefficient (denoted as r) ranges from to .
Positive and negative signs indicate directionality.
The absolute value of r indicates the magnitude or strength of the correlation.
A stronger correlation means the points are closer to the line of best fit.
As the correlation coefficient approaches (perfect positive) or (perfect negative), the relationship’s strength increases.
When (or close to zero), it indicates no relationship between the variables.
2.5.3 Correlations Can Be Misleading
A critical element of correlation is that it does not imply causation: "Correlations are not causation."
The existence of a correlation does not clarify the reason behind the relationship between two variables.
Example of Misleading Correlation:
Ice cream sales rising concurrently with homicide rates does not imply that ice cream consumption is causing violence.
A confounding variable may affect both variables.
Confounding Variable: A third variable that can influence the correlation seen between two other variables.
In this example, temperature acts as a confounding variable influencing both ice cream sales (people eat ice cream to cool down) and homicides (increased outdoor interactions in hot weather).
Understanding correlations requires scrutiny beyond mere numerical relationships;
Many curious or humorous correlations exist, such as:
"Eating more margarine lowers divorce rates."