Shuffling one column of pandas df with sample

Question

When I do the following I seem to wind up with the original df,

train['col1'] = train['col1'].sample(frac=1)

While if I use.values I get the shuffle I was after:

train['col1'] = train['col1'].sample(frac=1).values

I'm unclear why this happens - any insights?

Answer 1

Because the version without .values keeps the index and the column preserves its order.