First, you must collect your data. You can choose X for independent variable, for single parents, race, family income level, or educational attainment of parents. It seems to me you are to find equations for each of them.
Next choose Y for dependent variable; that is the high school drop outs.
Make table: First line for X, and the next line for Y. Plot the scatter diagrams and see what they would be, see whether it's a straight line or parabola, or higher degree polynomials.
Make another table for the variables you find in you normal equations. If it's a straight line, you will have two equations and two unknowns. If it's a parabola, you will have 3 x 3, etc.
Once you got the coefficient of regression, you can test the goodness of fit by computing the coefficient of correlation. If it's good, you can move the next stage. If not, continue till you get what you want.
The next stage is to do sample correlation and find the limits of 95% or 99% confidence limits.
If you want to make things interesting, you can test all you equations to find the dependency of variable; that's to find out the relationship of all your information using multiple correlation. When it's done, you will know whether race, income, unmarried parents contributes to the failure to complete high school, etc.
There is a lot of work, and a lot of formula involved that I cannot possible list them. You need to read and try out some examples before you apply all the above.
Perhaps, at the end of your research, you will find my prediction to be accurate. Here is my prediction: Those high school drop outs are just plain fools.