As mentioned above, you can get full marks for things like the correlation coefficient and the least squares regression line with no working. The flipside of this is that if you have the wrong answer (easy to do, anyone can key in an incorrect value), then you get nothing. It's always a good idea to show some working - at the very least, you should do the calculation twice, entering all of the data again, to make sure that you get the same answer.
From the most recent S1 mark scheme;
No Method Shown
Where the question specifically requires a particular method to be used, we must usually see evidence of use of this method for any marks to be awarded.
Where the answer can be reasonably obtained without showing working and it is very unlikely that the correct answer can be obtained by using an incorrect method, we must award full marks. However, the obvious penalty to candidates showing no working is that incorrect answers, however close, earn no marks.
Where a question asks the candidate to state or write down a result, no method need be shown for full marks.
Where the permitted calculator has functions which reasonably allow the solution of the question directly, the correct answer without working earns full marks, unless it is given to less than the degree of accuracy accepted in the mark scheme, when it gains no marks.
Otherwise we require evidence of a correct method for any marks to be awarded.