Assignment 6 2004-2005: Computing Correlations

Date Assigned: January 11/13
Title of Assignment: Computing Correlations
Due Date: January 18/20
Points: 10

Using a dependent variable of interest, your task is to uncover three interesting relationships with independent variables as summarized by Pearson correlation coefficients. When you finish this assignment you should have three substantively meaningful correlations that help to explain variation in a single dependent variable of your choice (which may be an index composed of more than one basic variable). Your correlations should approach or exceed .20. Also keep in mind basic issues of statistical significance.

Given that you must end up with three relationships of some interest, you should probably start working with more than three independent variables. And be careful to do any necessary recoding of your variables and to declare any missing values before calculating any correlations.

To calculate the correlations between your dependent variable and your independent variables, use the Correlations command in the syntax window. In completing your assignment, create a correlation matrix including all three independent variables and your dependent variable.

Submit the correlation matrix, and using a few sentences each, please describe the three relationships you have found in terms of strength, direction and significance. Before describing the direction of each relationship, make sure to look carefully at the coding of the variables. Then try to distinguish among the independent variables in terms of their explanatory power (i.e., through the use of r-square: how much variation in the dependent variable is being explained byt the independent variable). Finally, briefly summarize the substantive meaning of the relationships you have uncovered (i.e., what are the relationships telling us?).

How you interpret the relationships you find between your dependent variable and your independent variables will depend in part on the interrelations among the independent variables. Make sure that they are not just slightly different indicators of the same underlying dimension of variation. To do so you should consider their interrelations by examining the portion of the correlation matrix formed by the pairings of the independent variables.

If you think it is appropriate, and you are working with SPSS for Windows, you may also wish to look at the scattergram results. Be careful in interpreting plots of data at less than the interval level for they can be very misleading.

 Happy searching!

Assignments