- Widely used, quick and easy to calculate.
- Requires data on ordinal (ranked) scale.
- Non-parametric (not normally distributed) data.
- Shows correlation between two variables with a degree of statistical accuracy.
- Requires a sample of not less than 7 observations.
- It tests for linear relationships only and would give an answer of 0 for curvilinear relationships.
- Scale is important. Two variables might give a strong positive correlation but if a sub set of these variables is considered a value of 0 or a totally random result might be obtained.
- Nonsense correlations might results - i.e. variables which are not related at all might show up as positive/negative correlations, e.g. height of people and IQ!
- A scatter graph can be drawn using the ranked data - this would show the relationships betwen the variables clearly.
- Because only ranked data are used, a more accurate measure of the degree of association between two variables is Pearson's Product Moment Correlation Co-efficent.
Spearman's Rank Correlation Co-efficent
No replies to this topic
1 user(s) are reading this topic
0 members, 1 guests, 0 anonymous users