Correlations & meta-analysis

definition of correlations

a statistical technique that measures strength and direction of a relationship between two co-variables

positive correlation

as one co-variable increases, the other increases

negative correlation

as one increases, the other decreases

zero correlation

no relationship

perfect negative

-1

none

0

perfect positive

+1

strength

1. weak

2. moderate

3. strong

1. 0-0.3

2. 0.4-0.6

3. 0.7-1.0

scattergram

graphical plot of each pair of scores

correlation does not imply…

causation

what might also explain the relationship

a third, unmeasured variable (cofounding variable)

• the direction of causality is unclear

strengths of correlations

useful when experiments are unethical or impractical; can identify patterns for further research; relatively quick.

limitation of correlation

cannot establish cause and effect; may be affected by outliers; possibility of spurious correlations.

what is a meta-analysis

A research method that combines the results of multiple studies on the same topic to produce an overall conclusion (often calculates an overall effect size).

process of a meta-analysis

1. Identify all relevant studies (published and unpublished).

2. Calculate an effect size for each.

3. Combine effect sizes statistically.

strengths of meta-analysis

increases statistical power (larger N); can resolve conflicting findings; more reliable and generalisable than single studies

weakness of meta-analysis

publication bias (only positive studies may be available); ‘apples and oranges’ problem (studies with different methods); researcher bias in selecting studies; file drawer problem (unpublished null results missing).