(OP, ignore this bit too!)
*You are assuming that retail and self employed are mutually exclusive. Which if it said X work in retail and Y work in farming they would be.
However as far as I can see you can work in retail and be self employed, eg running your own shop.
As the OP has worded it, I (obviously) think my interpretation is correct, though I accept I could be wrong and the question could have used different examples for improved clarity.
If it was
507 pupils in a school. 248 have a dog. 172 like spaghetti. What is the probability a child either has a dog or likes spaghetti, you could see these are independent things.*
I think what we are saying is that this question would be impossible to answer based on the given information, unless you assume that the two groups are mutually exclusive - which would be a silly assumption given the example of dogs/spaghetti, of course, but there is no other way to solve the problem. Given the actual question, with retail and self-employed, it is much more reasonable to assume they are likely to be mutually exclusive categories, particularly since you couldn't solve the problem any other way.
To use your dogs/spaghetti example - you couldn't know how many of the dog lovers liked spaghetti or vice versa; it could be all of them or none of them. There's no way to work out any probabilities in that case.
If the question had say "the probability of someone liking dogs is 248/540, and the probability of someone liking spaghetti is 172/540, what is the probability that a randomly selected person likes both/neither/one or the other", then that would be solvable - and in that case, yes, using tree diagrams could be helpful. Giving the probabilities of these events, however, is very different than giving the number of people in the groups as is the case here!
So it's impossible to solve it any other way than assuming the categories are mutually exclusive, and given the actual categories and the level that the OP's course is, that would really be the most sensible interpretation, even though it's poorly worded because it is possible that there could be people in both categories. While that would be an interesting question for slightly more advanced GCSE students, you would need further information to draw a Venn diagram or equivalent.
It's also poorly worded by not saying that the person in question is 'randomly selected'!