Research Methodology in Social Sciences
Dr. David Sichinava
November 30, 2018
How do we assess relationship between two variables?
- The simplest way to examine relationship between two variables is to check whether they covary,
- That is, whether they change simultaneously
- Covariance coefficient can be measured as follows:
- \( cov(x, y) = \frac{\sum(x_{i}-\bar{x})(y_{i}-\bar{y})}{N-1} \)
How do we assess relationship between two variables?
- As you might notice, covariance measurement depends on the scale of variables, therefore we should standardize coefficients
- Here we should include standard deviations for both variables as follows:
- \( cov(x, y) = \frac{\sum(x_{i}-\bar{x})(y_{i}-\bar{y})}{(N-1)s_{x}s_{y}} \)
Types of correlations
- Bivariate: examines the relationship between two variables
- Partial: examines relationship between two variables while controlling for the effect of other variable
- In this case, we have to assess the impact moderating variable has on correlation coefficients
Types of correlation: distribution is crucial
- Spearman's rho
- Kendall's tau
- Small population with multiple ranked variables
Correlation vs. Causation
Correlation vs. Causation
Factor analysis
- Often we have to deal with latent variables, that is phenomena which cannot be measured directly
- Put it simple, calculating correlations between variables might yield clusters of variables
- Therefore, variables which are highly correlated which might be measuring same latent concept
Factor analysis
- It is not neccessary to retain all factors to our analysis
- Examine so called scree plot
- In order to improve the quality of your factor analysis by rotating factors