Correlation & regression (MR, unit 10)

Correlation: the measurement of the nature and strength of association between two variables

Two variables might be...

• perfectly correlated - exact linear relationship
• partly correlated - tendency of association
• uncorrelated

Negative & positive correlationDetermine strength of relationships (N.B. scatter and linear can forecast too)

1) The scattergraph method: line of best fit through plotted data
2) Linear regression analysis (the least squares method)
3) The correlation coefficient (r): very complicated equation, basically if...
r = +1 perfectly positively correlated
r = -1 perfectly negatively correlated
r = 0 variables are uncorrelated
4) Spearman's rank correlation coefficient: If data is given in terms of order or rank rather than values, calculate Spearman's rank correlation coefficient (interpreted the same as above, just specific to rank) e.g. placings of students rather than their actual mark
5) Multivariate analysis: analysing the relationship between more than two variables (use statistical tool - way to complicated)