There are $n$ persons sitting around a circular table. Then, in how many different ways 3 persons can be selected if none of them are neighbours.
My approach:-
Let us pretend that we have already picked 3 out of n persons. Now we have to place them with remaining n-3 persons in such a way that none of them are sitting together.
This can be easily done by gap method. There are 4 gaps between 3 persons in a line, xoxoxox , but there are 3 gaps between 3 persons in a circle.
So there are $(n-3)$ gaps for $(n-3)$ persons in a circle. We have to place those chosen $3$ persons in these $(n-3)$ spots. This can be done in $n-3 \choose 3$ ways which is $\frac{1}{6}(n-3)(n-4)(n-5)$. Which should be the answer because we can again pick those three and they fulfil our conditions.
But, The actual answer is $\frac{1}{6}n(n-4)(n-5)$.
The question is How?
Other people have explained how to get the correct answer, so I will just focus on why your answer is wrong: there is no one-to-one correspondence between the ways to choose three gaps between $n-3$ people and the ways to choose three non-neighbor people out of $n$. Let's call the people in your gaps problem $A_1, \dots, A_{n-3}$, and the people in the original problem $B_1, \dots, B_n$. If you put three additional persons between $A$s, how would you choose which $B$s correspond to them? For example, you can always identify $A_1$ with $B_1$, but then you will only get the ways to choose three $B$s other than $B_1$. So you see that the answer to the gaps problem is smaller than the answer to the original problem.
By the way, since every person (for example, $B_1$) is selected in exactly $\frac 3n$ of all the ways to choose three non-neighboring $B$s, what you should get your way is $1-\frac 3n$ of the correct answer. And this is exactly what you get: $\frac 16(n-3)(n-4)(n-5) = (1-\frac 3n)\cdot \frac 16n(n-4)(n-5)$.