(Worth up to 1 po!nt"# OB$E%TI&ES : 1. Formulate the null and alternative hypotheses for a given problem. 2. Determine the appropriate test based on mean or proportion, sample size, and known and unknown. 3. ele!t " based on the !onte#t of the problem. $. %erform an hypothesis test for a mean and a proportion. &. 'ompute test statisti!s and p()alues using *egatat. +. ,nalyze test statisti!s, make a de!ision, and formulate a !on!lusion in the !onte#t of the problem. P'O%E()'E: 'omplete the hypothesis tests in this -ab using the & steps in .ypothesis /esting0 1. Des!ribe the problem to be tested stating the 1ull and ,lternative .ypothesis 2.3 and . a 4. 2. pe!ify the de!ision rule using both the !riti!al value and p(value approa!hes. 3. 'olle!t the data and test the hypothesis using *egatat. $. *ake a de!ision0 5ased on your test results, either re6e!t the null or fail to re6e!t the null. &. 7rite a !on!lusion and8or a!tion to be taken based on the results of your hypothesis test. You *u"t +,p-.!n /our conc-u"!on 0!th!n th+ cont+,t o1 th+ pro2-+*3 !n oth+r 0or4"3 5u"t 0r!t!ng '+5+ct Ho or 6.!- to '+5+ct Ho !" not +nough 1or 1u-- cr+4!t7 9999999999999999999999999999999999999999999999999999999999999999999999999 :!on 138*gt 13 -ab ; %age 1 NOTE: To +n"ur+ /our "ucc+""3 /our !n"tructor "trong-/ r+co**+n4" th.t /ou 4r.0 . gr.ph 1or +.ch h/poth+"!" t+"t to 2+ p+r1or*+47 <se the *egatat summary statisti!s from the (ES%'IPTI&E STATISTI%S if ne!essary to derive the data re=uired to !omplete the hypothesis test for the mean or proportion. /o test the hypothesis, you will use *egatat, .ypothesis /est, and *ean vs .ypothesized )alue or %roportion vs .ypothesized )alue. /o determine whether you should a!!ept or re6e!t the null hypothesis, the =uestions will ask you to use the 'riti!al )alue method and8or the %()alue method. .owever, /our 4+c!"!on ru-+ "hou-4 "t.t+ 2oth .ppro.ch+". %r!t!c.- &.-u+ (%&# M+tho4: 'ompare the test statisti! to the ') for the given alpha. >f the test statisti! is e=ual to or greater than the '), you re6e!t the null. >f the test statisti! is less than the '), you fail to re6e!t the null. 6or th+ p-&.-u+ M+tho4: 'ompare the p()alue 2whi!h is the tail area4 to the given alpha. >f the %()alue is less than or e=ual to alpha, you re6e!t the null. >f the p()alue is greater than alpha, you fail to re6e!t the null hypothesis. /he p(value is the !han!e, !omputed under the assumption that .3 is true, of getting the observed value and more e#treme values. &E'Y IMPO'TANT: IN 'EAL WO'L( APPLI%ATIONS3 YO) ESTABLISH YO)' (E%ISION ')LE BEFORE YOU COLLECT THE DATA3 TH)S3 6O' O)' APPLI%ATIONS3 'EMEMBE' YOU WOULD NOT HAVE THE STATISTICAL DATA WHEN YOU STATE THE NULL AND ALTERNATIVE HYPOTHESES AND WHEN YOU FORMULATE THE DECISION RULE7 <se the & steps shown above for ea!h hypothesis test. *ake sure that your null is the status =uo or the belief or the !laim. /he alternative or resear!h hypothesis is what we are most interested in testing. ,fter determining the null and the alternative, you must sele!t a value for " whi!h is the probability of re6e!ting the null when it is a!tually true or !ommitting a /ype > error. /he probability of a /ype >> error, ?, means that you fail to re6e!t the null when it is a!tually false and should be re6e!ted. @e6e!ting a null hypothesis when it is false is what every 9999999999999999999999999999999999999999999999999999999999999999999999999 :!on 138*gt 13 -ab ; %age 2 good hypothesis test should do. /he Apower of the testB is the measure of how good a test is. >t is the probability that the test will re6e!t .3 when in fa!t it is false. 7e often talk about alpha 24 and beta 24 using the language of ChigherC and Clower.C For instan!e, we might talk about the advantages of a higher or lower (level in a study. Dou have to be !areful about interpreting the meaning of these terms. >f we use higher (levels, we are increasing the !han!e of a /ype > :rror. /herefore, a lower (level a!tually means that we are !ondu!ting a more rigorous test. For your referen!e, here are ways to interpret alpha0 the lower the , the lower the powerE the higher the , the higher the power the lower the , the less likely it is that you will make a /ype > :rror 2i.e., re6e!t the null when itFs true4 the lower the , the more CrigorousC the test an of .31 2!ompared with .3& or .134 means the resear!her is being relatively !areful, s8he is only willing to risk being wrong 1 in a 133 times in re6e!ting the null when itFs true 2i.e., saying thereFs an effe!t when there really isnFt4 an of .31 2!ompared with .3& or .134 means that both your statisti!al power and the !han!es of making a /ype > :rror are lower an of .31 means you have a ;;G !han!e of saying there is no differen!e when there in fa!t is no differen!e in!reasing 2i.e.., from .31 to .3& or .134 in!reases the !han!es of making a /ype > :rror(( saying there is a differen!e when there is notE de!reases the !han!es of making a /ype >> :rror(( saying there is no differen!e when there is and de!reases the rigor of the test in!reasing 2i.e., from .31 to .3& or .134 in!reases the power of the test be!ause one will be re6e!ting the null more often and, !onse=uently, when the alternative is true, there is a greater !han!e of a!!epting it DonFt feel overwhelmed after reading the ways to interpret alpha, this information is for your referen!e. 'an you think of some situations where you would want to use " H 3.31 and why you would want to use that level of signifi!an!eI Jne more item0 ,fter reading an hypothesis problem and you are trying to determine the null and alternative hypothesis, assume the alternative hypothesis is K unless there is something in the problem that suggests that the resear!her is interested in proving L or M. 9999999999999999999999999999999999999999999999999999999999999999999999999 :!on 138*gt 13 -ab ; %age 3 T/p+ I: '+5+ct th+ nu-- 0h+n !t !" tru+3 8 T/p+ II: (o not r+5+ct th+ nu-- 0h+n !t !" 1.-"+3 9 Po0+r o1 th+ t+"t: '+5+ct th+ nu-- 0h+n !t !" 1.-"+ (1 : 9# Nood lu!k this, !hapter is very !hallenging and is e#tremely important. E;AMPLE 9 HYPOTHESIS TEST O6 THE MEAN3 < <se the following set of sample data points, to test the alternative hypothesis 2.a4 that the population mean 2<4 is larger 2M4 than =707 A""u*+ th.t th+ nu-- h/poth+"!" !" 2<4 H 3.3. :nter your data into 'olumn , in an :#!el spreadsheet. 3.$, $.1, 2.O, +.1, 3.3, 1.$, 2.;, 3.3, 3.P, $.2, 3.1 No to *egatat, (+"cr!pt!>+ St.t!"t!c"3 !li!k JQ, enter your input range by highlighting your data and then !li!k on CJQ.C Dour input range would be ,1 to ,11 2if your data is listed in 'olumn ,4. For your output position, enter any !onvenient !ell and !li!k JQ. :#!el returns the sample mean of 3.$&$& along with the other des!riptive statisti!s and a ;&G !onfiden!e interval. STEP 1 H0: < ? =703 H . : < @ =70 A!>+n th.t B ("!g*.# !" unCno0n .n4 th+ ".*p-+ "!D+ (n ? 11# !" "*.--3 0+ u"+ th+ t 4!"tr!2ut!on. in!e there is no information about these data, then we will use the !ustomary alpha of .3&. 5efore we !omplete step 2, we need to find the !riti!al t value to !ompare to our t test statisti!. No to the 'riti!al )alues of t table at the end of your te#tbook. <se alpha 2upper tail area4 of .3& with 13 degrees of freedom 2d.f. is n(14 to find the !riti!al t value. For our data you should find t !riti!al value of 1.O12&. STEP E (+c!"!on ru-+--I1 th+ t+"t "t.t!"t!c !" +Fu.- to or gr+.t+r th.n our cr!t!c.- t >.-u+ o1 17G1E3 r+5+ct H0 or !1 th+ p-&.-u+ !" -+"" th.n or +Fu.- to 703 r+5+ct th+ nu--7 Niven that this is a one(tailed greater(than test, you should draw a graph showing the mean, re6e!t region in the upper right, and the !riti!al t value. 9999999999999999999999999999999999999999999999999999999999999999999999999 :!on 138*gt 13 -ab ; %age $
RH3.3 t )alues0 3 1.O12&!v /o !ondu!t the a!tual hypothesis test !omparison of this sample mean 23.$&$&4 against R, from any !ell of your worksheet in :#!el, go to ,dd(ins, *egatat, .ypothesis /est, and sele!t *ean versus .ypothesized )alue. ele!t data input and highlight your !olumn of data. :nter the hypothesized mean 3.3, sele!t greater than for the alternative, !li!k on display ;&G !onfiden!e interval, t test, and !li!k JQ. /he dialogue bo# is shown below. STEP = W+ con4uct our h/poth+"!" t+"t o1 th+ *+.n o2t.!n!ng th+ r+"u-t": Hypothesis Test: Mean vs. Hypothesized Value 3.0000 hypothesized value 3.4545 mean Data 1.1553 std. dev. 0.3483 std. error 11 n 10 d 1.30 t .110! p"value #one"tailed$ upper% &.!'84 (oniden(e interval )5.* lo+er 4.&30' (oniden(e interval )5.* upper 0.''!1 mar,in o error 'ompare our results to the de!ision rules shown above. 9999999999999999999999999999999999999999999999999999999999999999999999999 :!on 138*gt 13 -ab ; %age & STEP H S!nc+ our t t+"t "t.t!"t!c 17=0 !" -+"" th.n our t cr!t!c.- >.-u+ 17G1E .n4 our p- &.-u+ o1 7110I !" gr+.t+r th.n .-ph.3 0+ FAIL TO REJECT HO. /his result indi!ates that our sample mean is lo!ated within the region of a!!eptan!e as shown on the graph below.
RH3.3 t )alues0 3 1.33ts 1.O12&!v 7e !annot !omplete step &, be!ause no information was given about this data set from whi!h we !ould draw !on!lusions and make re!ommendations. -astly, not!c+ th.t th+ 9J con1!4+nc+ !nt+r>.- o1 E7IKGH L < L H7E=0K contains our sa!"# #an o1 =7HH 0h!ch con1!r*" th+ .ccur.c/ o1 our 4+c!"!on to $ai" to r#%#ct t&# nu"" h/poth+"!" th.t < ? =707 HYPOTHESIS TEST 6O' A P'OPO'TION3 M /o perform an hypothesis test of a sample proportion, p, for a population proportion, S, we follow the same & steps for hypothesis testing. Jne simplifi!ation !ompared to hypothesis tests for the mean, R, is that we always use the T distribution for a proportion. .ere we will !omplete a two(tailed test for a proportion. Failure to meet payments on student loans guaranteed by the < government has been a ma6or problem for both banks and the government. ,ppro#imately &3G of all student loans guaranteed by the government are in default. , random sample of 3&3 loans to !ollege students in one region of the < indi!ates that 1$P are in default. %erform the hypothesis test for the population proportion. 9999999999999999999999999999999999999999999999999999999999999999999999999 :!on 138*gt 13 -ab ; %age + STEP 1 H0: M ? 703 H. N 70 5efore we !omplete step 2, we need to find the !riti!al T value to !ompare to our t test statisti!. 5elow is a table showing 'ommon T !ores that you might find helpful as we !omplete this and the ne#t !hapter. %o**on O Scor+" -evel of ignifi!an!e 2-evel of 'onfiden!e4 13G 2;3G4 &G 2;&G4 1G 2;;G4 .&G 2;;.&G4 .2G 2;;.OG4 Jne(/ailed 2T
4 U1.2O U1.+$& U2.33 U2.&P& U2.OO
/wo(/ailed 2T /2 4 U1.+$& U1.;+ U2.&P& U2.O1 U3.3O in!e we are doing a two(tailed test and using alpha of &G, there is a probability of .32&3 in ea!h tail and our T !riti!al value is U 1.;+. /he graph is shown below. S H .&3 T )alues (1.;+!v 3 V1.;+!v 9999999999999999999999999999999999999999999999999999999999999999999999999 :!on 138*gt 13 -ab ; %age P STEP E (+c!"!on ru-+ !" !1 th+ O t+"t "t.t!"t!c !" +Fu.- to or gr+.t+r th.n our cr!t!c.- >.-u+ o1 P179I3 0+ r+5+ct H0 or !1 th+ p-&.-u+ !" -+"" th.n or +Fu.- to 703 r+5+ct th+ nu--7 For this problem, our sample proportion, p H 1$P83&3 H 3.$2. Jur !hallenge is to transform .$2 into a standardized T value to see if our sample proportion lies in the a!!eptan!e or the re6e!tion region. >t is obvious that .$2 lies to the left of .&3, but we need to determine if the T s!ore for .$2 is greater than, e=ual to, or less than (1.;+. /o !ondu!t the hypothesis test of the sample proportion, p H 3.$2 against the population proportion S H .&, from any !ell of your worksheet in :#!el, go to ,dd(ins, *egatat, .ypothesis /est, and sele!t %roportion versus .ypothesized )alue. :nter the observed proportions, the hypothesized proportion, sele!t not e=ual to, display ;&G !onfiden!e interval and !li!k JQ. /he dialogue bo# is shown below. 9999999999999999999999999999999999999999999999999999999999999999999999999 :!on 138*gt 13 -ab ; %age O STEP = W+ con4uct our h/poth+"!" t+"t o1 th+ proport!on g!>!ng th+ r+"u-t": Hypothesis test or proportion vs hypothesized value Observed Hypothesized 0.4& 0.5 p #as de(imal% 14'-350 1'5-350 p #as ra(tion% 14'. 1'5. . 350 350 n 0.0&!' std. error "&.)) z .00&8 p"value #t+o"tailed% 0.3!83 (oniden(e interval )5.* lo+er 0.4'1' (oniden(e interval )5.* upper 0.051' mar,in o error 'ompare our results to the de!ision rules shown above. STEP H S!nc+ our O t+"t "t.t!"t!c -E799 !" gr+.t+r th.n our O cr!t!c.- >.-u+ o1 -179I .n4 our p-&.-u+ o1 700EG !" -+"" th.n .-ph.3 0+ REJECT HO. 9999999999999999999999999999999999999999999999999999999999999999999999999 :!on 138*gt 13 -ab ; %age ; P700EG /his result indi!ates that our sample mean is lo!ated within the region of re6e!tion as shown on the graph below. S H .&3 T )alues (2.;;T (1.;+!v 3 V1.;+!v STEP Our ".*p-+ r+"u-t" !n4!c.t+ th+r+ !" o'#r(&#"in) #'i*#nc# th.t th+ tru+ proport!on o1 "tu4+nt -o.n" !n 4+1.u-t !" -+"" th.n 0J7 Finally, not!c+ th.t th+ 9J con1!4+nc+ !nt+r>.- o1 07=IG= L M L 07HK1K *o#s not contain our ".*p-+ proport!on o1 *+.n o1 =7HH 0h!ch con1!r*" th+ .ccur.c/ o1 our 4+c!"!on to r#%#ct t&# nu"" h/poth+"!" th.t M ? 077 9999999999999999999999999999999999999999999999999999999999999999999999999 :!on 138*gt 13 -ab ; %age 13 LAB 9 ASSIANMENT : B+1or+ "t.rt!ng th!" .""!gn*+nt3 /ou "hou-4 h.>+ r+p-!c.t+4 th+ E,.*p-+" .n4 ".>+4 th+ 0orC"h++t" ." /our St.t" E,.*p-+ 9 0orC2ooC7 P+r1or* th+ H/poth+"!" T+"t!ng St+p" .n4 u"+ M+g.St.t to "o->+: During a national promotion to in!rease sales of @alphFs %rivate ele!tion brand !ola, the national sales manager noti!ed that the average number of !ases sold to !ustomers nationwide during this sale was 3.P !ases. .owever, the >rvine area sales manager believes that this sale was even more su!!essful in his region. , random sele!tion of 1P store patrons in the area pur!hased the following number of !ases of !ola0 3 $ & 2 $ + 3 & O O & 2 2 P 3 + & <sing a 13G level of signifi!an!e 2"H.134 and both the ') and %(value methods, !an the lo!al sales manager support his !laim that this sales !ampaign was more su!!essful in >rvine than it might have been nationwideI During the same sales !ampaign, the >rvine area manager asked a random sample of 233 people if they preferred the storeWs %rivate ele!tion brand over another supermarket brand. /he results were that 113 of those responding said they preferred %rivate ele!tion over the !ompeting !ola brand. Jverall, the manager believes that more than &3G of the >rvine population would prefer his %rivate ele!tion brand over the !ompeting !ola brand. <sing alpha H .3&, does this sample provide eviden!e to support the managerWs belief use both the !riti!al value from the z table and the p(value from your :#!el output to support your !on!lusion. , bottling ma!hine is set to automati!ally fill ea!h bottle with 3&& ml. of soft drink. 9999999999999999999999999999999999999999999999999999999999999999999999999 :!on 138*gt 13 -ab ; %age 11 3$& 3&O 3&P 3&+ 3$& 3&O 3&P 3&; 3&& 3&; 3$P 3&P 3&3 3&+ 3&2 3&O 3&3 3$2 3$O 3&1 . /o !he!k whether the ma!hine needs to be read6usted, a =uality !ontrol inspe!tor e#amines a random sample of newly filled bottles. <se alpha of &G and the ') method to support your de!ision. tudies indi!ate that P&G of those attending 7orld eries games en6oy at least one beer during the game. @esear!hers found that 12$ of 1&; of guests at Name 1 of this yearFs 7orld eries drank at least one beer. >s there eviden!e that the Name 1 guests differ from the population proportionI ave your file as tats -ab ;. 9999999999999999999999999999999999999999999999999999999999999999999999999 :!on 138*gt 13 -ab ; %age 12 :!on 138*gt 13 /<D:1/ 1,*:0 LAB 9 ANSWE' SHEET 'EA( THESE INST')%TIONS %A'E6)LLY P+r1or* th+ r+Fu!r+4 h/poth+"+" t+"t" u"!ng th+ "t+p" 4+"cr!2+4 !n th+ -.2 0h!ch !nc-u4+" 1or*u-.t!ng . 4+c!"!on ru-+ u"!ng 2oth %& .n4 p-&.-u+ *+tho4" .n4 0r!t!ng . co*p-+t+ conc-u"!on !n th+ cont+,t o1 th+ pro2-+*7
6O' TEST STATISTI% AN( %'ITI%AL &AL)ES YO) WILL ENTE' THE SIAN O6 THE ANSWE' AN( )SE = (E%IMAL PLA%ES7 6or +,.*p-+3 . corr+ct t+"t "t.t!"t!c or cr!t!c.- >.-u+ 0ou-4 2+ Q17KI or -07=HG7 An !ncorr+ct t+"t "t.t!"t!c or cr!t!c.- >.-u+ 0ou-4 2+ 17KI or 7=HG7 NOTE: YO) MAY NEE( TO )SE THE SIAN: Q/- 6O' %'ITI%AL &AL)ES WHEN YO) A'E (OINA A TWO TAIL TEST7 6or +,.*p-+3 . corr+ct cr!t!c.- >.-u+ .n"0+r 1or . t0o t.!- t+"t 0ou-4 2+ Q/-17IH7 An !ncorr+ct .n"0+r 0ou-4 2+ Q17IH or -7IH7 From the @alphFs %rivate ele!tion brand !ola0 1. 7hat is the sign of .aI a. K b. M !. L d. H 2. 7hat is the !riti!al value 2use sign2s4 of answer and 3 de!imal pla!es4I 3. 7hat is the test statisti! 2use sign2s4 of answer and 3 de!imal pla!es4I $. De!isionI a. Fail to @e6e!t .o b. @e6e!t .o &. 'an the lo!al sales manager support his !laim that this sales !ampaign was more su!!essful in >rvine than it might have been nationwideI 9999999999999999999999999999999999999999999999999999999999999999999999999 :!on 138*gt 13 -ab ; %age 13 a. D: b. 1J From the preferred the storeWs %rivate ele!tion brand over another supermarket brand0 +. 7hat is the sign of .aI a. K b. M !. L d. H P. 7hat is the !riti!al value 2use sign2s4 of answer and 3 de!imal pla!es4I O. 7hat is the test statisti! 2use sign2s4 of answer and 3 de!imal pla!es4I ;. 7hat is the p(value 2use $ de!imal pla!es4I 13. De!isionI a. Fail to @e6e!t .o b. @e6e!t .o From the bottling ma!hine0 11. 7hat is the !riti!al value 2use the sign2s4 of the answer and 3 de!imal pla!es4I 12. 7hat is the test statisti! 2use the sign2s4 of the answer and 3 de!imal pla!es4I 13. Does the ma!hine need to be read6ustedI a. D: b. 1J From the 7orld eries Name 10 1$. 7hat is the sign of .aI a. K b. M 9999999999999999999999999999999999999999999999999999999999999999999999999 :!on 138*gt 13 -ab ; %age 1$ !. L d. H 1&. >s there eviden!e that the Name 1 guests differ from the population proportionI a. D: b. 1J ESTIMATE( TIME TO %OMPLETE E;AMPLES AN( LAB: 9999999999999999999999999999999999999999999999999999999999999999999999999 :!on 138*gt 13 -ab ; %age 1&