I'm quite new to SPSS so i hope I'm able to explain everything right...

I am currently conducting a case-control study and tried to match my sample in SPSS. I got two Datasets (one with all the cases and one with all the controls). I merged them all in to one new Dataset: I got one row for each case/ control and and one column with their matchID so I know which cases and controls are matched. For my further analysis I wanted to do an Wilcoxon-test (Data is not normally distributed) to see if there are any differences between cases and controls for different variables. If I try to perform this specific test it is not possible the cases and controls are not arranged in one line (like pairs). I also wanted to do Odds-Ratios and got the same problem. Hopefully I was able to describe the problem somewhat understandably. Thanks in advance.

Hello Tabea.  You said that after you merged the files, you had "one row for each case/ control and and one column with their matchID."   On the face of it, that sounds like there is ONE row per matchID.  But something you said later makes me think you actually have TWO rows per matchID.  Which is it? 

If you have two rows, you selected the ADD CASES option when merging.  To get one row per matchID, you need to select the ADD VARIABLES option.  (In terms of syntax, those options correspond to ADD FILES and MATCH FILES commands respectively)  Here are some tutorials from UCLA that should be helpful:

• https://stats.oarc.ucla.edu/spss/modules/match-merging-data-files/
• https://stats.oarc.ucla.edu/spss/modules/concatenating-data-files/

Regarding odds ratios, the conventional approach for a matched case-control design is to estimate a conditional logistic regression model.  AFAICT, SPSS still does not have a command for that.  However, as the following tech support note explains, you can use NOMREG to estimate the model when it is 1:1 matching.  (Another tech support note shows how to use COXREG in the case of 1:n matching.)

• https://www.ibm.com/support/pages/11-matched-case-control-studies-conditional-logistic-regression

An alternative to conditional logistic regression is to estimate a logistic regression model via GENLIN with generalized estimating equations (GEE) to account for the paired observations.  Once upon a time, I compared the two approaches on some data I had to analyze, and the results were very similar. 

By the way, the relevant coefficient shown in the output (B) will be the log of the OR you want, so you'll have to request Exp(B) in the output. 

I hope this helps.

NOTE to the powers that be at IBM SPSS:  Do you not think it is well past the time that SPSS have a command for conditional logistic regression?