> i changed it already.
Do not change the question type, if you already have data. Otherwise, you will only see the number of options checked for all the "old" data. You can switch back later, but you still will have to merge your datasets.
> Can you please explain it a bit more detailed?
If you have a multiple choice matrix, this allows people to check one or more options. Each option is 1/2-encoded with 1=not checked, and 2=checked. This means that you need one variable for each response option. If the options 1 and 3 are checked, the result would be
 2 1 2 1 1
The leading  is the number of options checked, because SoSci Survey will add this information. And then 2 for "option 1 checked", 1 for "option 2 not checked" and so on.
This makes sense for some research designs, but not if you want a likert-like scale.
> Where can i add the value, change the deviation?
The part with the standard deviation was about your question how to identify straightliners. So, if you have a scale, the data for option 3 checked would simple be
and if you have 5 items, and each time the option 3 is checked, you have this:
3 3 3 3 3
So, the standard deviation of a straight line is always 0. And this is ehy you can easily identify straightlined scales with valueSD(). Nothing to change, just a little PHP code on the next page of your questionnaire.