Unformatted text preview: size of the sample is not a matter at issue here since all requirements for
use of the Student t have been met. 3. This real example is very similar to one that was performed in inpatient
pharmacy at Tripler several years ago. So, here goes: The Medassist
Pharmaceutical Company uses a machine to pour cold medicine into bottles in
such a way that the standard deviation of the weights is 0.15 oz. A new machine is
tested on '1’] bottles, and the standard deviation for this sample is 0.12 oz. The
Dayton Machine Company, which manufactures the new machine, claims that it
ﬁlls bottles with lower variation. At the 0.05 signiﬁcance level, test the claim
made by the Dayton Machine Company. If Dayton’s machine is being used on a
trial basis, should its purchase be considered? Solution: Here we want to test variances, so we must use the Chisquared statistic
and the Chisquared distribution. a. The claim is that the Dayton machine has a fill variance less than 0.15, so
the counter claim is that the Dayton machine has a fill variance of greater
than or equal to 0.15. The null hypothesis is then H0 :0“ = 0.15. The alternative hypothesis comes in this case from the original claim (the other has the equality) so
HI :0' < 0.15 . The hypothesis shows we need a left sided one tail test. The distribution for variances is the Chisquared distribution.
The dof is one less than the sample size or 70.
2 2 The Chi square statistic is given by 12 = ("—1215— = (19(0—122) or (0.15)
For 70 degrees of freedom, and using the ChiSquare table for a 5% level
of signiﬁcance we have a bit of a problem While the normal and the
Student t are symmetric, the Chisquare is not, and the table values are
read from the right, not the left. So since we want a left handed test, we
much leave 95% of the area to the right of the critical value. We look up
the value, then for 95%, not 5%: This critical value is 51.739. Now since
44.8 is to the left of 5 1 .739, our test statistic is in the rejection area. Hence
we reject the null and accept the alternative: they do better than a standard
deviation of less than 0.15. So this supports the procurement of the new machine. = 44.3 ...
View
Full Document
 Spring '10
 Serpa
 Statistics

Click to edit the document details