I am working on a paper on unemployment on psychological health of youth. Please in what order can input the following predictors variables using age as the control variable?
Perseverance
Irrational behaviour
Low self esteem
Depression and
Suicidal ideation
It probably depends on your hypotheses and your design. I'd look at inter correlations between them, to see if multicollinearity might become a problem, but in the end theory and predictions should dictate the order.
I assume that you are asking about the order in which variables are to be entered into a hierarchical regression-type model (ie not related to the imputation of missing variables).
Firstly, unless you have a very small sample, I would not recommend that you proceed by entering one variable at a time. I would suggest entering all of the potential control variables simultaneously into the model, and then refining the model by the removal of variables not contributing sufficiently to the model. (This is sometimes referred to as the "backward-elimination" procedure. Of course it is desirable to examine the VIFs (variance inflation factors) of each variable to make sure that there is not too high level of multicollinearity. (What is an acceptable value would depend on the sample size.)
An alternative would be to use the so-called "step-wise" procedure, which reduces the number of variables present simultaneously in the model. (This successively introduces and rejects variables in the model, depending on p-values or other criteria.) Both of these procedures can be done automatically by selecting the appropriate option in SPSS, but can be done manually.
- Badges
- Science method
- Similar topics
- Methodology
- Research Methodology
- Qualitative Research