Hi wta2k!
Once again, it's me answering your questions. I've already worked on
your questions #185196 and #185198, so please refer to those first,
since I will draw on some concepts I outlined in those answers.
As in the previous exercises, what you want to find here is the
p-value of the test; in this case, what is the probability of
observing 12 defective items (that's 4% of 300) or more if the claim
that no more than 2% of the items is defective is true. Thus, first we
define the null and alternative hypothesis:
Call 'p' the proportion of defective items
Null hypothesis: p<=0.02
Alternative hypothesis: p>0.02
Second, note that the number of defective items follows a binomial
distribution. Check the following page, which describes what a
binomial experiment looks like to see why is it so.
http://stat.tamu.edu/stat30x/notes/node66.html
Now that you know the distribution, it's easy to calculate the
probability you're looking for. Define X to be "number of defective
items observed". You're looking for
Prob( X >= 12 )
given p=0.02 and n=300 (n is number of observations). Now, it's
difficult to calculate this probability using the binomial
distribution. Fortunately, for a large n, we can approximate the
binomial by using a normal distribution, with mean equal to n*p and
standard deviation (SD) equal to sqrt( n*p*(1-p) )(recall that sqrt
means square root). Thus, we can say that X has approximately a normal
distribution with
Mean = 300*0.02 = 6
SD = sqrt(300*0.02*0.98) = 2.42
And now, we calculate the probability just like in the other
questions. We have to substract the mean and divide by the SD to have
a *standard* normal distribution in order to use the table.
Prob( X >= 12 )
=Prob( X-6 >= 12-6 )
=Prob( (X-6)/2.42 >= (12-6)/2.42 )
=Prob( (X-6)/2.42 >= (12-6)/2.42 )
=Prob( z >= 2.47 )
(z is (X-6)/2.42, i.e. a random variable with standard normal
distribution)
In order to find that probability, again, we use the table. You may
use the link I provided in question #185198. The value of this
probability is 0.006, and is also called the p-value of this test.
Based on the p-value, we could conclude that the supplier's claim is
not credible. If the true proportion of defective items was 2%, we
would observe 12 defective items in 300 with a probability of only
0.006. Thus we reject the null hypothesis. Even if we set the level of
significance as low as 0.01 (rejecting only p-values that are smaller
than 0.01), we still reject the null hypothesis.
I hope this explanation was clear to you. As in the previous
questions, please don't hesitate to request further clarification
before rating my answer.
Best wishes,
elmarto |