Correlations
Correlations
- A correlation is not a research method as such, rather it is an analysis of the relationship between co-variables
- In correlational research, the variables are not manipulated and instead two co-variables are measured and compared to look for a relationship - one or both co-variables may be pre-existing e.g. school attendance and number of GCSEs achieved
- Each participant has two scores e.g. number of cups of caffeine and number of hours sleep. This is plotted as one point on a scattergraph
- Scattergraphs will show one of three outcomes:
- Positive correlation: One co-variable increases as does the other (but not necessarily at the same rate) e.g. calories consumed and weight gained
- Negative correlation: One co-variable increases while the other co-variable decreases (but not necessarily at the same rate) e.g. hours spent watching television and level of fitness
- Zero correlation: There is no relationship between the variables e.g. hair colour and IQ
- Analysing the relationship between co-variables can be done either with a scattergraph visually or by calculating the correlation coefficient which is expressed as a numerical value
- The correlation coefficient represents both the direction and strength of the relationship between the co-variables, expressed as a value between -1 and +1
- A perfect positive correlation would be expressed as +1
- A perfect negative correlation would be expressed as -1
- No relationship would be expressed as 0
- Both positive and negative coefficient correlations can be described as weak, moderate or strong e.g. a correlation coefficient of 0.03 is a weak positive correlation; a correlation coefficient of - 0.08 is a strong negative correlation
Strengths of Correlational Analysis |
Limitations of Correlational Analysis |
They are able to show relationships between variables | They do not show causation i.e. they are unable to show which variable impacts the other |
The data may be easily available for researchers to quickly analyse |
Extraneous relationships with other variables may affect the co-variables and the outcome e.g. number of days absence from school may be due to illness rather than choice |
Allows predictions to be made when looking at the relationships between co-variables |
Correlations work well for linear relationships e.g. height and shoe size, but are less successful when dealing with non-linear relationships e.g. number of hours worked and level of happiness |
Due to the data being readily available, there are unlikely to be any ethical issues |
The difference between correlations & experiments
- Experiments involve deliberate manipulation the independent variable (IV) to measure the impact on the dependent variable (DV)
- This means we can establish a cause and effect relationship
- With a correlation, there is no deliberate manipulation of any variables
- Both variables may impact each other and we measure the degree to which this happens
- There is therefore no way to establish cause and effect, even if there is a strong positive correlation
- This means that whilst we can understand how one variable will impact another, we cannot establish a cause and effect relationship because correlation doesn't always mean causation
- Neither is better or worse as it is dependent on the context and situation being investigated