Correlations

Correlations

1. a correlation is a systematic association between two continuous variables

2. studies using correlational analysis are known as correlational studies

3. the results can be displayed on a scattergram where each point plotted shows the values of the two variables for one indiviudal

4. correlational hypothesis

   • states the expected association between co-variables
   • positive correlation - as one increases, the other increases
   • negative correlation - as one increases the other decreases
   • zero correlation - no association
   • a correlation - unspecified direction (non-directional hypothesis)

5. correlation coefficient

   • number between -1 & +1 which tells us how closely the co-variables are related, a perfect positive correlation would be + 1 & a perfect negative correlation would be - 1
   • strong correlation has a correlation of coefficient closer to +1 or -1 & weak correlation is nearer 0
   • the statistical significance of the correlation depends on the number of participants, this can be analysed using tables of significance

6. Evaluation

   • can be used when not possible to manipulate a variable
     • this contrasts with an experiment, where an investigator manipulates the IV in order to observe effect on the DV
     • correlations can therefore be used to investigate trends in data
     • but this means that no conclusion can be made about one co-variable causing another in a correlational study
   • correlation does not imply causation
     • there may be intervening variables that explain the association between the co-variables. Alternatively the causal relationship may be the opposite of what it seems
     • e.g. correlation between attendance & exam results may have other explanations e.g. personal circumstances or motivation to succeed. we cannot conclude that increasing attendance would improve results
     • this means that further investigation would need to be carried out to find out if the relationship is causal
   • validity should be considered in a correlational study
     • e.g. generalisability of sample, the operationalisation of variables. These issues would affect external validiity & internal validity
     • correlational studies can often be easily replicated
     • so findings can be confirmed & causality can be explored further using experimental or longitudinal methods
   • Linear & Curvilinear correlation
     • linear correlations produce scattergram with values lying on a straight line
     • however, curvilinear correlation is also possible
     • this isi a consistent relationship between two variables which changes as one variable increases e.g. relationship between stress & performance
     • this requires a different sort of statistical significance testing, to see how closely the data fit a curve rather than a straight line
   • A curvilinear correlation
     • e.g. The Yerkes-Dodson effect - arousal is positively correlated with performance at low levels of arousal but negatively correlated at higher levels of arousal
       • ~~>

Images and files in the map