Correlation Coefficients and Their Interpretation

Correlation Coefficients

• Definition: A numerical index reflecting the relationship between two variables.
• Range: Values range from $-1$ to $+1$.
• Bivariate Correlation: Relationship between two variables.
• Pearson Product-Moment Correlation: Examines the relationship between two continuous variables (e.g., height, age).

Types of Correlations

• Direct (Positive) Correlation: Variables change in the same direction (e.g., as X increases, Y increases). Value is positive ($.00$ to $+1.00$).
• Indirect (Negative) Correlation: Variables change in opposite directions (e.g., as X increases, Y decreases). Value is negative ($-1.00$ to $.00$).

Key Principles of Correlation

• Strength: The absolute value of the coefficient reflects strength; $-.70$ is stronger than $+.50$.
• Data Points: Requires at least two data points (variables) per case.
• Variability: Must have variability in both variables; if one variable does not change, the correlation is zero.
• Constrained Range: Restricting the range of a variable reduces the observed correlation.
• Notation: Represented by $\text{r}_{xy}$ for variables X and Y.

Computational Formula for Pearson Correlation Coefficient ($\text{r}_{xy}$)

• $r_{xy} = \frac{n\Sigma XY - (\Sigma X)(\Sigma Y)}{\sqrt{[n\Sigma X^2 - (\Sigma X)^2][n\Sigma Y^2 - (\Sigma Y)^2]}}$
  • $n$ = sample size
  • $X$, $Y$ = individual scores
  • $\Sigma XY$ = sum of products of X and Y
  • $\Sigma X^2, \Sigma Y^2$ = sum of squared individual X and Y scores

Visual Representation: Scatterplots

• Definition: A plot of each set of scores on separate axes ($X$ on horizontal, $Y$ on vertical).
• Interpretation: The general shape indicates direction and strength.
  • Positive Slope: Data points cluster from lower-left to upper-right (direct/positive correlation).
  • Negative Slope: Data points cluster from upper-left to lower-right (indirect/negative correlation).
  • Perfect Correlation ($\pm 1.00$): Data points align along a straight line.

Correlation Matrix

• A table showing correlation coefficients for all pairs of multiple variables.
• Diagonal values are $1.00$ (variable correlated with itself).
• Symmetrical: $\text{r}<em>{AB}$ is the same as $\text{r}</em>{BA}$.

Interpreting Significance of Correlation Coefficient

• General Interpretation (Rule of Thumb):
  • $.8 \text{ to } 1.0$: Very strong
  • $.6 \text{ to } .8$: Strong
  • $.4 \text{ to } .6$: Moderate
  • $.2 \text{ to } .4$: Weak
  • $.0 \text{ to } .2$: Weak or no relationship

Coefficient of Determination ( $r^2$)

• Definition: The percentage of variance in one variable accounted for by the variance in the other variable.
• Computation: Square the correlation coefficient ($r^2$).
• Example: If $\text{r} = .70$, then $\text{r}^2 = .49$, meaning $49\%$ of variance is explained.
• Coefficient of Alienation (or Nondetermination): The percentage of unexplained variance ($1 - r^2$).

Correlation vs. Causality

• Association: Correlations express an association between variables.
• No Causality: Correlation does NOT imply causation. A third variable (confounder) might explain the relationship (e.g., ice cream sales and crime rates are both influenced by temperature).

Other Correlation Coefficients

• Different techniques exist for variables at different levels of measurement (e.g., point-biserial for nominal-interval, Spearman rank for ordinal-ordinal, Phi coefficient for nominal-nominal).