Tuesday, September 25, 2007

Linear Correlation

When researchers wish to explore/establish if two phenomena or variables are related, they use what is known as correlational technique.

Creating graphs and computing for correlation coefficients are frequenly done in such research projects.

The scattergram or scatter plot (or simply scatter) is the graph that is used in correlational studies.

Click the provided links to review how to interpret scatter plots.
Want to create your own scatter plot? Click here or here.

Please note that NOT all relationships are linear. Click here for scattergrams illustrating nonlinear relationship.

The most popular correlation coefficient is the Pearson’s r.

Please take note that one of the assumptions of Pearson’s r is linearity and there are variables that may have nonlinear relationship.

Another assumption is level of measurement.

But sometimes the only information is binary in nature (e.g., right or wrong answer, passed or failed the test, etc) Such data are definitely lower than interval level. We can use point biserial correlation coefficient or phi coefficient, both “descendants” of the Pearson’s r.

Click here to review the alternatives to Pearson's r when its assumptions could not be satisfied. Additional Links
Other Interesting Links
Posted by at