One possible analysis to investigate the relationship between two ordinal variables (education levels) and job levels on an interval variable is to use a multiple regression model. A multiple regression model can estimate the effect of each predictor variable (education levels and job levels) on the outcome variable (interval variable) while controlling for the other predictor variable. The model can also test whether the interaction between the two ordinal variables is significant, which would indicate that the effect of one ordinal variable on the outcome depends on the level of the other ordinal variable. A multiple regression model can provide information about the direction, magnitude and significance of the relationships among the variables.

At a minimum, you should be able to convert the education levels into a linear variable by recoding it as "years of schooling."

Hello Jeanine,

You could:

1. Build a path model and estimate the terms via software that allows for ordinal values (e.g., lavaan, mplus);

2. Use Spearman correlations for each of the three relationships, noting that the association is between ranks of scores and not scores (for which there is not a one-to-one relationship as is most often the case with other transformations);

3. Convert your DV to an ordinal variable and use ordinal (logistic) regression; (This of course ignores information in the DV scores, so it's less desirable.)

4. The conversion of string variables to ordinal values should follow some defensible process.

Good luck with your work.