No, you have a problem. I would say your levels were incorrect for the test. Play around with the input output levels until the graph doesn't oscillate like that.
Here's my soundcard graph using the same scale axis as you used.

What do you mean when you say run the measurement again. Running the soundcard measurement again will only produce another soundcard measurement file.
But, if you have a proper soundcard file it will make response measurements accurate because the results will remove the response anomolies caused by the soundcard. If you want to see that this system is working, you can take a normal sweep measurement of the piece of cable shorting line-in to line-out. It should be flat.
Mmm, I don't really see how you could do this?
brucek