Method 2 is right.
Consider an example, of flipping two fair coins at two different places.
Bh : Head appeared on the coin flipped at Bangalore.
Bt : Tail appeared on the coin flipped at Bangalore.
Ph : Head appeared on the coin flipped at Pune.
Pt : Tail appeared on the coin flipped at Pune.
As the both coins being used at Pune & Bangalore are fair,
P(Bh) = P(Bt) = 1/2 and
P(Ph) = P(Pt) = 1/2.
The probability of getting head at both the places =
P(Bh ∩ Ph).
But since the result of coin flip at on location is not going to affect the result at other location, they are independent.
so P(Bh ∩ Ph) = P(Bh)xP(Ph) = (1/2)x(1/2) = 1/4.
but here it can be observed drom the diagram below that, these two events (Bh & Ph) are having overlapping sample spaces:
Hence,
"Independent events are events whose sample space of happening never overlaps."
This is not a correct argument.
Disjoint sample spaces of the two events tells us that the events are mutually exclusive.
And if any two events are mutually exclusive then they are highly dependent because occurrence of any one event immediately tells us the non occurrence of other event.
Also if A & B are two mutually exclusive events,
then A ∩ B = null.
so P(A ∩ B) will always be 0.
but if both P(A) & P(B) are non zero then P(A) x P(B) will never be equal to 0.
So A & B are not independent, if they are mutually exclusive and both of them having non zero probabilities.