I am testing the effects of traditional mental health counseling versus culture specific counseling. Therefore, there will be two groups: one group that receives culture specific treatment and one group that receives traditional treatment. Everyone will be given a pretest and a posttest, the Beck Anxiety Inventory, which involves a Likert scale. I want to compare the scores between the groups to see who experienced better treatment. Which statistical test should I use to do this? I have SPSS.
Assuming that you have randomized the assignment to groups in an appropriate manner such as calling a computer system prior once each patient has agreed to participate (otherwise this becomes a lot more difficult and you have to start thinking about things like propensity scores etc.), I would consider an analysis of covariance for the change from the pre-intervention value to the post-intervention value with the treatment group as a factor in the model and the pre-intervention BDI value as a covariate (plus a pre-specified list of further potentially obvious covariates that might influence the outcome as additional model terms). Any standard statistics software will be able to fit this type of model. The model assumes that the regression residuals are reasonably normally distributed, which is often plausible if the Likert scale has enough steps on the scale. It would also be logical to check this on the data of some previous study. It would be worthwhile to get some statistics support at your institution to help you write the protocol (and then the analysis plan) for the study.
I would suggest a 1-way ANOVA with the difference score (Post test - pre test) as the outcome variable. You'll want to make sure the groups were similar enough on the BDI at the beginning.