CORRELATIONS

• CORRELATIONS:

  • The extent to which 2 variables (must be quantifiable) are related (association)
  • They DO NOT demonstrate cause & effect (no manipulation of variables)
• Positive Correlation:

  • Increase in one variable tends to be associated with an increase in the other
• Negative Correlation:

  • Decrease in one variable tends to be associated with an increase in the other
• Zero Correlation:

  • No relationship between variables
• Uses of correlations:

  • Predictions: if there is a relationship between 2 variables we can make predictions about one from another
  • Validity: Concurrent (correlation between new / established measure?)
  • Reliability: Test-retest (are measures consistent?) Inter-rater (are observers consistent?)
  • Theory Verification: Predictive validity
• Correlation Co-efficient:

  • Numerical value between -1 & +1 (Used instead of drawing a scattergraph)
  • Extent to which the pairs of numbers for the 2 variables lie on a straight line
  • Values over 0- positive correlation, Values under 0- negative correlation
  • Correlation of -1  = perfect negative correlation, Correlation of +1 = perfect positive correlation 
• There will be a positive / negative relationship -> directional

  There will be a relationship -> non-directional

  There will be no relationship -> null
• STRENGTHS OF CORRELATIONS:

  • Provide quantitative data (easy to analyse)
  • Can research sensitive topics (would otherwise be unethical) as there is no manipulation of variables
  • Preliminary research technique- allows researchers to identify a link that can be explored through controlled research 
• WEAKNESSES OF CORRELATIONS

  • Correlations only show relationships between variables, not cause & effect
  • Third-variable problem: If 2 variables are positively / negatively correlated, this doesn’t mean that 1 causes another. The random / coincidental relationship may be due to another unmeasured / unintended variable
• EXPERIMENTS VS CORRELATIONS:

  • Experiments- assess the effect of 1 variable (IV) on another measured variable (DV)
  • So this means that data is discrete / separate, and the effect of this is something else that is being measured
  • Correlations- do not use separate / discrete conditions, assess how much of a relationship exists between 2 related variables