15 - Methods Of Correlation

Question 1

Correlation

Answer 1

• Degree of correspondance between variable.
• measures only linear relationships
• positive / negative / no correlation
• indicate direction and strength of the relationship
• confidence interval depend on N

As N increase the correlation value required to achieve significance decreases.

Question 2

Coefficient of determination

Answer 2

• r^2

- what percentage of the total variance of variable X is due to the variance of variable Y

Question 3

Assumptions for correlational analysis

Answer 3

• data must be obtained for related pairs
• ratio or Interval level of measurement
• each set of score should be normal
• linear relationship
• homoscedasticity

Question 4

homoscedasticity

Answer 4

Variability of scores for one variable should be roughly similar at all values of the other variable.

Question 5

Pearson correlation test

Answer 5

• r= 0.1 small effect size
• r= 0.3 medium effect size
• r=0.5 strong effect size

Question 6

Spearman’s rank order correlation test

Answer 6

• rho (correlation)
• ranked data or interval if ranked
• few subjects involved
• provide a lower estimate of correlation than r because data is degraded

Question 7

Errors in interpreting correlation coefficient

Answer 7

• causality
• 2 variables are related to a third
• mathematically correct but in reality meaningless
• if the values do not represent the full range of the distribution => careful in interpreting the correlation
• inherent relationship of variables may differ from population to population

Question 8

Other correlation techniques

Answer 8

• Cramer’s V (chi square)
• Kendall’s tau (<10 normal distribution but harder to calculate and lower r obtained)
• phi (when 2 variables are dummy variables like yes/no)

Question 9

Partialling out

Answer 9

Holding a variable constant to eliminate its influence on the correlation between two variables.