Each inspection could have 2 results: good (no error) or bad (error).
P(bad) = 0.2, P(good) = 1 - 0.2 = 0.8.
It is assumed that results of inspections are independent.
(1) P(first error will be for the 3rd order) = P(good, good, bad) = 0.8*0.8*0.2 = 0.128
It is easier to solve (3) the next:
(3) P(the inspector will examine at least seven orders) = P(the first 6 orders are good) = (0.8)6 = 0.262144
Now solve (2).
(2) Consider 2 random events:
A: the inspector will find error in the 1st order
B: the inspector will find the first error between the 2nd and 6th order (inclusive)
C: the inspector will find the first error after the 6th order.
These events are mutually exclusive and one of them must occur.
So, P(A) +P(B) + P(C) = 1
P(A) = 0.2
P(C) = P(the first 6 orders are good) = (0.8)6
P(B) = 1 -P(A) - P(C) = 1 - 0.2 - (0.8)6 = 0.537856
(4) P(the first error will be on the eighth order or later | the first four telephone orders contain no errors) =
= P(the 5th, 6th, and 7th orders are good | the first 4 orders are good)
Because we assume independence, results of following inspections do not depend on the previous results:
P(the 5th, 6th, and 7th orders are good | the first 4 orders are good) = P(the 5th, 6th, and 7th orders are good) = (0.8)3 = 0.512.