Just to be clear, the data are reported in the picture attached.
We set the difference in graffiti incidents between before (B) and after (A) as:
d = A - B
If the number of incidents has indeed reduced, then d will be negative.
In order to choose the hypothesis, we need to recall that the null hypothesis H₀ represents the statement that we want to prove wrong, while the alternative hypothesis H₁ represents the statement that we want to prove right.
Therefore, we want μd to be strictly negative (= 0 would mean that the number of incidents stayed constant).
Hence:
H₀ : μd ≥ 0
H₁ : μd < 0
