Binomial distribution can be used because the situation satisfies all the following conditions:
1. Number of trials is known and remains constant (n)
2. Each trial is Bernoulli (i.e. exactly two possible outcomes) (success/failure)
3. Probability is known and remains constant throughout the trials (p)
4. All trials are random and independent of the others
The number of successes, x, is then given by

where

Here given
p=0.8 (success=non-defective)
n=3
x=0 or 1
Then






Answer: probability of at most one non-defective out of a random sample of three is 0.104