You are on page 1of 8

Farmasi UMS Semester Genap 2012/2013 Tutorial: Hypothesis Testing

Conceptual uestions 1. In hypothesis testing, the hypothesis tentatively assumed to be true is a. the alternative hypothesis b. the null hypothesis c. either the null or the alternative d. All of the above answers are correct. e. None of the above answers is correct. !"S#$%: & 2. In hypothesis testing if the null hypothesis is rejected, a. no conclusions can be drawn from the test b. the alternative hypothesis must also be rejected c. the data must have been accumulated incorrectly d. the sample size has been too small e. None of the above answers is correct. !"S#$%: e . !he level of significance is the a. ma"imum allowable probability of !ype II error b. ma"imum allowable probability of !ype I error c. same as the confidence coefficient d. same as the p#value e. None of the above answers is correct. !"S#$%: & $. !he level of significance in hypothesis testing is the probability of a. accepting a true null hypothesis b. accepting a false null hypothesis c. rejecting a true null hypothesis d. could be any of the above, depending on the situation e. None of the above answers is correct. !"S#$%: c %. !he probability of ma&ing a !ype I error is denoted by a. b.

c. 1# d. 1# e. None of the above answers is correct. !"S#$%: a '. (hen the following hypotheses are being tested at a level of significance of )*+ 1** )a+ , 1** the null hypothesis will be rejected if the test statistic - is a. . - b. . - c. , #- d. , 1** e. None of the above answers is correct. !"S#$%: c /. (hen the p#value is used for hypothesis testing, the null hypothesis is rejected if a. p#value , b. , 0#value c. p#value . d. p#value 1 e. None of the above answers is correct. !"S#$%: a 2. In order to test the following hypotheses at level of significance )*+ 1** )a+ . 1** the null hypothesis will be rejected if the test statistic - is a. . - b. , - c. , #- d. , 1** e. None of the above answers is correct. !"S#$%: a 3. (hich of the following does not need to be &nown in order to compute the p# value4

a. &nowledge of whether the test is one#tailed or two#tailed b. the value of the test statistic c. the level of significance d. All of the above are needed. e. None of the above answers is correct. !"S#$%: c 1*. In the hypothesis testing procedure, is a. the level of significance b. the critical value c. the confidence level d. 1 # level of significance e. None of the above answers is correct. !"S#$%: a 11. A soft drin& filling machine, when in perfect adjustment, fills the bottles with 12 ounces of soft drin&. Any overfilling or under filling results in the shutdown and readjustment of the machine. !o determine whether or not the machine is properly adjusted, the correct set of hypotheses is a. )*+ , 12 )a+ 12 b. )*+ 12 )a+ . 12 c. )*+ 12 )a+ 1 12 d. )*+ 1 12 )a+ 12 e. None of the above answers is correct. !"S#$%: '

(ro&lem 1: !he Naphtha reformer unit in a refinery produces an octane number of 3* under an e"isting 5old6 catalyst. A new catalyst has been developed and used in the unit. An average octane number of 32 has been estimated using a sample of $ data points using this new catalyst. )as the octane number changed significantly4 7se a hypothesis test
at the %8 significance level

9rom past e"perience, we &now that the standard deviation for octane number 5in octane units6 is 1.% and we are willing to assume that the variance hasn:t changed. Solution: )ypothesis test # 1. Null hypothesis )*+ 1 3* 2. Alternate hypothesis )a+ 3* . !est statistic+
X 3* A n

with substitution of the sample average132, n1$ and 11.% for our e"ample gives z 12.'/ $. ;ejection criteria+
X * > z A 2 A n

Is z 1 2.'/ inside the range of values for - that contains the middle 3%8 of the probability4 N<, z 1 2.'/ isn:t between #1.3' and =1.3'. 5>ee yellow table for rejection criteria6. !herefore, it doesn:t seem to ma&e sense that 1 3*. (e say that ?we reject the null hypothesis@ and we conclude, at the 3%8 confidence level, that 3* using the new catalyst.

(ro&lem 2: !he yield of a chemical reactor is being studied. 0revious e"perience has indicated
that the standard deviation of yield for the reactor is 8. !he past five days of plant operation have provided yields of 31.'8, 22./%8, 3*.28, 23.3%8 and 31. 8. 5a6 !he previous yield was 3*8. )as the yield of this reactor changed under current operation4 !est at the %8 significance level. 5b6 (hat is the type II error probability for this e"periment if the true mean yield is 3284

Solution:
In this Buestion, we are studying chemical reactor yield. (e are given a standard deviation value of ## the language in the Buestion indicates that this is considered to be a CN<(N value 5it is based on e"tensive previous operating e"perience, and no indication is given of the number of data points from which it might have been computed6. !hus, we should be using hypothesis tests or confidence intervals for means, with CN<(N variance. !he average yield computed from the data is 3*.$28 . )ypothesis test # a 2#sided test is reBuired 5D)as the yield of this reactor changed under current operation4D6. !his is a test of a value of the mean, with &nown variance. )*+ mean 1 3* )a+ mean not eBual to 3* !he test statistic is
X 3*

= *. '

since sigma is , and n1%. !his absolute value of this test statistic is compared against z*.*2% 1 1.3', and the null hypothesis is not rejected, i.e., we conclude that the yield has not changed. Eonfidence interval approach F !he confidence interval is
X z *.*2% A n < < X + z *.*2% A n

which produces 52/.2%, 3 .116. >ince this interval contains 3%8, we conclude that the yield has not changed. 5b6 (hat is the type II error probability for this e"periment if the true mean yield is 3284 !he acceptance region for hypothesis in part 5a6 is

z*.*2%

X 3* z*.*2% A n

3* z*.*2% A n X 3* + z*.*2% A n 2/. / X 32.'


Note here, the acceptance region is a region centred at 3*, not the confidence interval that we got in part 5a6. In order to calculate the type II error, we need to calculate the area under the normal curve with mean 32 and standard deviation , between the interval 52/. /, 32.' 6. !he new statistic is

X 32 A n
Eonvert the interval 52/. /, 32.' 6 to standard normal fence values as

2/. / 32 = .$% A % 32.' 32 = *.$/ A %

and

!hen the type II error is

= P ( .$% Z *.$/ )

= P ( Z *.$/ ) P ( Z .$% ) = *.'2*222 *.***22* = *.'2

!his is the same if we calculate a ?shift@ in the original standard normal fence values z*.*2% and

z*.*2% . !his shift is

* = 1.$3 n

>o the new fence values of standard normal distribution is

z*.*2% = 1.3' 1.$3 = .$% z*.*2% = 1.3' 1.$3 = *.$/

!he new fence values are the same as what we got above, so the type II error is

= P ( .$% Z *.$/ ) = *.'2

(ro&lem 3: !he sugar content in the syrup of canned peaches has been found to be Normally
distributed. !he variance is thought to be 12 mg2. A random sample of 1* cans has been selected, and has yielded a sample standard deviation of $.2 mg. Goes this observation support the conclusion that the population variance is 12 mg24 7se a hypothesis test at the %8 significance level.

Solution:
(e are interested in testing whether the sugar content variance is 12 mg 2. (e can assess this using a hypothesis test on variance. !he estimated variance is $.22 1 2 .*$. !he hypotheses are as follows+ )*+ variance 1 nominal variance of 12 )a+ variance not eBual to nominal variance of 12 !he test statistic is+

5 n 16 s 2
2 *

>ubstituting in the values, we obtain 352 .*$6A12 1 11.%2 (e need to compare this to two fences with an outer tail area of %8 5I have decided to test at the %8 significance level6. !he values are 2 3 ,*.3/% = 2./*
2 3 ,*.*2% =13.*2

>ince the test statistic lies in between these two fences, we do not reject the null hypothesis, and conclude that the sugar content is plausibly 12 mg2. Eonfidence interval approach F !he appropriate form of the confidence interval is+
s 2 5 n 16
2 n 1, *.*2%

< 2 <

s 2 5 n 16
2 n 1, *.3/%

!he chi#sBuared values are 2./* and 13 for the upper tail areas of *.3/% and *.*2% respectively. !he confidence interval is 51*.31, /'.26. >ince this interval contains 12, we conclude that it is plausible that the variance of the syrup sugar concentration is 12.

(ro&lem ): !he wall thic&ness of 2% glass 2 H bottles is being monitored by a Buality control
engineer. !he average of the bottles has been computed to be $.*% mm, and the sample standard deviation of the bottles is *.*2 mm. In order to confirm that Buality specifications are being met, we must demonstrate that the mean wall thic&ness is greater than $ mm. 7se a hypothesis test at the %8 significance level to assess whether this condition is being met.

Solution: In order to assess whether the wall thic&ness is greater than $ mm, use the following null and alternate hypotheses+ )*+ mean 1 $ mm )a+ mean . $ mm !he test statistic in this case is
X $ sA 2% = $.*% $ *.*2 A 2% = .12%

because the nominal value of the mean is $ mm, we have estimated the average using 2% data points, and we are estimating the variance from the data.
(e now have to compare this test statistic against the value from the >tudent:s t#distribution with 2$ degrees of freedom 5n#112%#112$6, and an upper tail area of %8 5since this is a one#sided test6. !he corresponding t#value is 1./11. >ince the test statistic value is greater than the fence, we reject the null hypothesis, and conclude that the wall thic&ness is greater than $ mm.

You might also like