Covary
Things that change together
Correlational research
Determine wether variables are related to another
Reasons for correlational rather than experimental research?
- Ethically impossible for us to do an experimental stduy
2. Usueful in measuring many variables and assessing relationships between them
Correlation coefficient
Indicates the strength(magnitute) and direction(- of +) of the relationship between two variables
Pearson correlation is a valuse between ___ and ___.
-1 and 1
Positicve correlation
Whent he score on 1 variable goes up, the score ont he other variable goes up
Negative correlation
When the score on 1 variable goes up the other variable goes down
Curvilinear relationships
Have a corelation of zero but there is a clear disctince correlation between variables present
Correlation can only be used on _____
Linear relationships
Statistical significance
The two variables are related in some meaningful way
3 factors that effect statistical significance
- Sample size
- Magnitude of correlation
- Researcher “conservativness) (ie 95% CI)
Sample size as effecting significance
Larger sample size= more likely to find sig
Magnitute of correlation as effecting sig
Closer number is to 1= higher chance of sig
Directional hypothesis
Predicting direction of correlation
Non directional hypothesis
Not sure which direction but know they vary in some way
Factors that distort correlation coefficients
- Restricted range
- Outliers
- Reliability of measures
Restricted range
Scores that come form a subset of the overal data set
Outliers
Data points that are 3 of more standard deviations form the mean
Ways to improve reliability
- Standardize the measure
- Clear opperational definitions
- Trained observers
3 factors that must be met to determin causality
- covariation
- directionality
- Elimination of alternative explanations
Directionality
-time order relationsips(ie manipulsting 1 variable caused the other to change)
Spurious correlation
Correlation between 2 variables that seem to move int he smae direction but are not actually related to one another
Partial correlation
Correlation between 2 variables with the influence of 1 or more other variables statistically removed
