You are on page 1of 7

MRA-1.

Population Mean Hypotheses


Each of the following paragraphs calls for a statistical test about a population mean m. State the null hypothesis H0
an the alternati!e hypothesis Ha in each case.
"a#$he iameter of a spinle in a small motor is suppose to be % mm. &f the spinle is either too small or too large'
the motor will not wor( properly. $he manufacturer measures the iameter in a sample of motors to etermine
whether the mean iameter has mo!e away from the target.
H0: =5 Ha: 5
"b#)ensus *ureau ata show that the mean househol income in the area ser!e by a shopping mall is +,-'%00 per
year. A mar(et research firm .uestions shoppers at the mall. $he researchers suspect the mean househol income
of mall shoppers is higher than that of the general population.
H0: =42,500$ Ha: >42,500$
/$he e0aminations in a large accounting class are scale after graing so that the mean score is %0. $he professor
thin(s that one teaching assistant is a poor teacher an suspects that his stuents ha!e a lower mean score than the
class as a whole. $he $A1s stuents this semester can be consiere a sample from the population of all stuents in
the course' so the professor compares their mean score with %0.
H0: =50 Ha: <50
Testing a Hypothesis
Mar(et pioneers' companies that are among the first to e!elop a new prouct or ser!ice' ten to ha!e higher
mar(et shares than latecomers to the mar(et. 2hat accounts for this a!antage3 Here is an e0cerpt from the
conclusions of a stuy of a sample of 1-04 manufacturers of inustrial goos5
6)an patent protection e0plain pioneer share a!antages3 7nly -18 of the pioneers claim a significant benefit
from either a prouct patent or a trae secret. $hough their a!erage share is two points higher than that of pioneers
without this benefit' the increase is not statistically significant "9 : 1.1;#. $hus' at least in mature inustrial
mar(ets' prouct patents an trae secrets ha!e little connection to pioneer share a!antages.6
<in the p-!alue for the gi!en 9. $hen e0plain to someone who (nows no statistics what 6not statistically
significant6 in the stuy1s conclusion means. 2hy oes the author conclue that patents an trae secrets on1t
help' e!en though they contribute - percentage points to a!erage mar(et share3
To find the p-value we look up =!"!# in the no$%al ta&le" 'ow the (han(e) to )ee ou$ eviden(e o$ an even %o$e
(o%pellin* eviden(e with >!"!# e+ual) to the a$ea to the $i*ht of !"!#" That i) (lo)e to !#," -o a))u%in* that the$e wa)
no $eal diffe$en(e &etween the %a$ket )ha$e) &etween (o%panie) with p$odu(t patent) and tho)e without the%, we )till have
!#, (han(e of )eein* the !20. (o%panie) inve)ti*ated doin* that %u(h &ette$ than the othe$)" The$efo$e it i) not unlikel/
at all that the)e (o%panie) we$e doin* &ette$ )i%pl/ &/ (han(e of )ele(tion and not &e(au)e the/ had a p$odu(t patent" -o a
2, point diffe$en( in %a$ket )ha$e (an o((u$ within )a%ple) of !20. +uite ea)il/ and we do not have the &a)i) fo$ $e0e(tin*
the po))i&ilit/ that t$ade )e($et) and p$odu(t patent) %ake no diffe$en(e"
M*S-,. 7ne-year Returns on Electric =tility Stoc(s
$he one-year rate of return to shareholers in 144% was calculate for each in a sample of >; electric utility stoc(s.
$he ata' e0tracte from $he 2all Street ?ournal are recore in the attache ataset.
1. Specify the null an alternati!e hypotheses teste for etermining whether the true mean one-year rate of return
for electric utility stoc(s e0ceee ;08.
Ho: ="# Ha: >"#
-.)alculate the obser!e significance le!el of the test.
1o to 2al(ulate3Te)t (hoo)e t-te)t fo$ individual %ean) and Ho: ="#, Ha:>0"#" The p-value i) $epo$ted to &e le)) than "
000! whi(h i) indeed )upe$ )%all"
;.&nterpret the result in @- in the conte0t of the problem.
4e have )t$on* eviden(e that %ake) u) &elieve that the ave$a*e one /ea$ $etu$n fo$ ele(t$i( utilit/ )to(k) e5(eed) #0," 4e
will $e0e(t H0 and a((ept the alte$native h/pothe)i)"
M)S-,. )ontaminate Soil an p-!alues
En!ironmental Science A $echnology "7ct. 144;# reporte on a stuy of contaminate soil in $he Betherlans.
Se!enty-two6n=728 ,00-gram soil specimens were sample' rie' an analy9e for the contaminant cyanie. $he
cyanie concentration Cin milligrams per (ilogram "mgD(g# of soilE of each soil specimen was etermine using an
infrare microscopic metho. $he sample resulte in a mean cyanie le!el of 5-&a$ = 94%*3k* an a stanar
e!iation of ) = 90 %*3k*"
$est the hypothesis that the true mean cyanie le!el in soil in $he Betherlans e0cees 100 mgD(g. =se a : .10.
Ho: =!00 %*3k*
Ha: >!00 %*3k*
:ou have to (al(ulate what i) the p$o&a&ilit/ of a )a%ple to have a %ean 94 o$ la$*e$ if the population %ean t$ul/ wa) !00"
-o /ou )et the %ean=!00, n=72 , )=90 and find the a$ea to the $i*ht of 94" ;nd we find that we have a .5"55, (han(e to
)ee )u(h a )a%ple if it (a%e f$o% a population with %ean !00" The$efo$e the$e i) no )i*nifi(ant eviden(e to &elieve that the
<ean (onta%ination i) hi*he$ than !00" =f an/thin* we would tend to &elieve &a)ed on thi) eviden(e that the %ean
(onta%ination wa) lowe$ than !00"
2oul you reach the same conclusion as in part a using a : .0%3 =sing a :.013 " ;t alpha= "!0, "05 o$"0! we would
e+uall/ would not &e a&le to $e0e(t Ho )in(e ou$ p-value i) wa/ a&ove an/ of the)e alpha level)""
2hy can the conclusion of a test change when the !alue of p:alpha is change3 =t (ould (han*e if the p-value would
&e )a/ "0#" ;t alpha -"! o$ "05 it i) (on)ide$ed )%all enou*h to lead u) to $e0e(tion &ut at alpha="0! level it i) not
)i*nifi(ant &e(au)e a )%alle$ alpha level $e+ui$e) eviden(e that ha) le)) than "0! p$o&a&ilit/ of o((u$$in*"
The$efo$e at the "0! level we (ould not $e0e(t H0 &a)ed on thi) eviden(e with p-value="0#"
M)S->. Fefecti!e Proucts
A large mail-orer company has place an orer for %'000 electric can openers with a supplier on the conition that
no more than -8 of the e!ices will be efecti!e. $o chec( the shipment' the company tests a ranom sample of
,006n=4008of the can openers an fins that 11 6)a%ple p$opo$tion=!!3400="02758are efecti!e.
1.Foes this pro!ie sufficient e!ience to inicate that the proportion of efecti!e can openers in the shipment
e0cees -83 $est using a : .0%.
!" H0:p="02
Ha: p>"02
2" )a%ple p$opo$tion i) "0275 and n=400"
#">)e den)it/ tool (li(k on the &o5 and (hoo)e p$opo$tion" ?nte$ p="02, n=400 and %ove the fla* to find the a$ea to the
$i*ht of "0275" the a$ea i) "!47 %eanin* that thi) )a%ple had a !4"7, (han(e to (o%e f$o% thi) di)t$i&ution whe$e the
%ean i) in fa(t 2," ;t alpha ="05 level the$e i) no *$ound to $e0e(t the h/pothe)i) that the %ean i) in fa(t 2,"
<in an interpret the obser!e significance le!el for the hypothesis test.
@="!47 the$e i) a !4"7, (han(e that we have )een thi) )li*htl/ hi*he$ value in thi) )hip%ent of 400 (an opene$) &/ (han(e"
$RE-;. Garbage
$hese ata are the total weights of garbage iscare by househols in one wee( "base on ata collecte as part of
the Garbage ProHect at the =ni!ersity of Ari9ona#. At the 0.01 le!el of significance' test the claim of the city of
Pro!ience super!isor that the mean weight of all garbage iscare by househols each wee( is less than ;% lb.' the
amount that can be hanle by the town. *ase on the result' is there any cause for concern that there might be too
much garbage to hanle3
2al(ulate3te)t" 2hoo)e individual t-te)t, =#5 Ha: <#5" -et alpha="00!" 4e *ot p<"000! )o we have )i*nifi(ant eviden(e
to $e0e(t Ho and a((ept the )upe$vi)o$) (lai% that in fa(t the ave$a*e hou)ehold t$a)h i) le)) than #5 l&" the$e i) no (au)e
fo$ (on(e$n &a)ed on the eviden(e"
2EB-,. Raio 7wnership
$he Raio A!ertising *ureau of Bew Ior( reports in Raio <acts' that in 144, the mean number of raios per =.S.
househol was %.>. A ranom sample of ,% =.S. househols ta(en this year yiels the ata recore in Fata Fes( on
number of raios owne.
Foes the ata pro!ie sufficient e!ience to conclue that this year1s mean number of raios per =.S. househols
has change from the 144, mean of %.>3 =se the following steps to answer the .uestion.
1.State the null an alternati!e hypothesis. H05 =5"A, Ha: 5"A
-.Fiscuss the logic of conucting the hypothesis test.
4e will u)e the )tati)ti( fo$% the )a%ple to (al(ulate how likel/ it i) to )ee a value of a )a%ple %ean diffe$ f$o% the
population %ean thi) %u(h o$ %o$e eithe$ &ein* that %u(h la$*e$ o$ that %u(h )%alle$" The p-value thu) will a((ount fo$
po))i&le (han(e diffe$en(e) of thi) %a*nitude in &oth tail) o$ di$e(tion)"
-.&entify the istribution of the !ariable 0' that is' the sampling istribution of the mean for samples of si9e ,%. the
)a%plin* di)t$i&ution will &e no$%al )in(e &elow the population )tanda$d deviation i) *iven" The %ean fo$ the )a%plin*
di)t$i&ution i) 5"A and itB) )tnd)$d deviation i) !".3)+ua$e $oot of 45"
;. 7btain a precise criterion for eciing whether to reHect the null hypothesis in fa!or of the alternati!e hypothesis.
Cet u) )et alpha="05" The$e i) no pa$ti(ula$ dan*e$ a))o(iated with a t/pe = e$$o$ )o alpha (an &e that la$*e"
Apply the criterion in the pre!ious .uestion to the sample ata an state your conclusion. Assume the population
stanar e!iation of the year1s number of raios per =.S. househols is 1.4.
2al(ulate 3te)t u)e -inte$val), )pe(if/ )i*%a and %u" :ou *et p="#0 and fail to $e0e(t H0 o&viou)l/" 'o eviden(e to &elieve
that the$e i) an/ diffe$en(e in the %ean nu%&e$ of $adio) pe$ hou)ehold"
2EB->. Hospital )osts
$he American Hospital Association reports in Hospital Stat that the mean cost to community hospitals per patient
per ay in =.S. hospitals was +4;1 in 144,. &n that same year' a ranom sample of ;0 aily costs in Massachusetts
hospitals yiele a mean of +11;1. Assuming a population stanar e!iation of +;;; for Massachusetts hospitals'
o the ata pro!ie sufficient e!ience to conclue that in 144, the mean cost in Massachusetts hospitals e0ceee
the national mean of +4;13 Perform the re.uire hypothesis test at the %8 significance le!el.
!"H0: =.#!
Ha: >.#!
2" 5&a$-!!#!, n=#0, )i*%a=###
#" >)e den)it/ tool" 2hoo)e %ean" ?nte$ <ean=.#!, )=###, n=#0 if /ou %ove /ou$ fla* it leave) the )($een at !!!#
an/thin* a&ove that value ha) a p-value le)) then "000!in pa$ti(ula$ ou$ )a%ple %ean !!#! ha) p-value le)) than"000!
(on)e+uentl/" 4e will $e0e(t H0 at 5, level" DEini)h =nte$p$etationF"
2EB-J. $he <ocus Fatabase
$he <ocus atabase contains information on %00 ranomly selecte Ari9ona State =ni!ersity sophomores. Se!en
!ariables are consiere for each stuent5 se0' high-school GPA' SA$ math score' cumulati!e GPA' SA$ !erbal
score' age an total hours. =se a statistical software pac(age sol!e the following problems.
1.&n 144% the "no-recentere# mean SA$ math score was ,J- nationally. Fo the <ocus ata on SA$ math scores
pro!ie sufficient e!ience to conclue that the mean SA$ math score of Ari9ona State =ni!ersity sophomores
e0cees the 144% national mean3 =se alpha e.ual to 0.0%.
/e), p<0"00! )o we will $e0e(t H0 at 55 level"
-. &n 144% the "no-recentere# mean SA$ !erbal score was ,J- nationally. Fo the <ocus ata on SA$ !erbal scores
pro!ie sufficient e!ience to conclue that the mean SA$ !erbal score of Ari9ona State =ni!ersity sophomores
e0cees the 144% national mean3 =se alpha e.ual to 0.0%.
fail to $e0e(t, p=!"oo thi) low )(o$e of 45A doe) not p$ovide eviden(e o&viou)l/ fo$ the ave$a*e &ein* hi*he$ than 492"
;. ?ustify the use of the hypothesis-testing proceures that you employe in parts 1 an -. -(o$e) in the )a%ple) we$e
no$%all/ di)t$i&uted, no outlie$) that %ake) ou$ a))u%ption that the population i) no$%al plau)i&le.
IMM1.
&n a iscussion of the eucation le!el of the American 2or(force' someone says 6$he a!erage young person can1t
e!en balance a chec(boo(.6 $he Bational Assessment of Eucational Progress "BAEP# says that a score of -K% or
higher on its .uantitati!e test reflects the s(ill neee to balance a chec(boo(. $he BAEP ranom sample of J,0
young men ha a mean score of 0-bar : -K-' a bit below the chec(boo(-balancing le!el. Assume that sigma : >0'
choose an alpha le!el' an perform a hypothesis test to etermine whether this is goo e!ience to conclue that the
mean score for all young men is less than -K%.
!"H0: =275 Ha: <275
-et alpha="05
-. 5&a$=272, )i*%a=A0, n=940 )i*%a of 5&a$=A03$oot940=2
#"p-value=@65&a$<2728=@6<-!"58="0AA9
4" -in(e the p-value i) hi*he$ than %/ )et alpha level = will not $e0e(t the null h/pothe)i)" 4e have no )i*nifi(ant eviden(e
to &elieve that /oun* %en (anBt &alan(e thei$ (he(k&ook) &a)ed on thi) )a%ple"
Type I , Type II error
IMM-;. $esting for ESP
A researcher loo(ing for e!ience of e0trasensory perception "ESP# tests %00 subHects. <our of these subHects o
significantly better "P less than .01# than ranom guessing.
*ase on this' is it proper to conclue that these four people ha!e ESP3 E0plain.
2hat shoul the researcher o now to test whether any of these four subHects ha!e ESP3
; te)t at !, level ha) a full !, (han(e to $e0e(t H0 in favo$ of Ha" -o if /ou $epeated the te)t 500 ti%e) /ou e5pe(t thi)
t/pe = e$$o$ to o((u$ a&out 5 out of the 500 ti%e)" -o )eein* $e0e(tion 4 out of 500 ti%e) (ould &e e5plained &/ the)e &ein*
the in(ident) fo$ a T/pe = e$$o$ and the$efo$e 4 out of 500 $e0e(tion) will *ive no eviden(e fo$ ?-@"
IMM-,. Schi9ophrenics
A group of psychologists once measure KK !ariables on a sample of schi9ophrenic people an a sample of people
who were not schi9ophrenic. $hey compare the two samples using KK separate significance tests. $wo of these tests
were significant at the %8 le!el. Suppose that there is in fact no ifference on any of the KK !ariables between people
who are an people who are not schi9ophrenic in the ault population. $hen all KK null hypotheses are true.
1.2hat is the probability that one specific test shows a ifference significant at the %8 le!el3 5,
-.2hy is it not surprising that - of the KK tests were significant at the %8 le!el3
4ell 5, of 77 t$ial) i) a&out #"5" -o we )o$t of e5pe(t %akin* a %i)take a&out # ti%e) the)e 2 te)t) (ould ea)il/ &e 0u)t
te)t) whe$e we %ade a T/pe = e$$o$ and $eall/ do not indi(ate that -(hioB) a$e diffe$ent wit $e*a$d to the %ea)u$ed
va$ia&le)"
M*S--. Proucer1sD)onsumer1s Ris(
&n .uality control applications of hypothesis testing' the null an alternati!e hypotheses are fre.uently specifie as
6$he prouction process is wor(ing6 an 6$he prouction process is not wor(ing6 respecti!ely. *ecause of this'
alpha is sometimes referre to as the proucer1s ris( an beta is referre to as the consumer1s ris(. An inHection
moler prouces plastic golf tees. $he process is esigne to prouce tees with mean weight of .-% ounce. $o
in!estigate whether the inHection moler is operating satisfactorily' ,0 tees were ranomly sample from the last
hour1s prouction. $heir weights "in ounces# are liste in the attache ataset.
1.Fo the ata pro!ie sufficient e!ience to conclue that the process is not operating satisfactorily3 =se alpha : .01.
'o, p<"000! )o we will $e0e(t the h/pothe)i) that the$e i) no p$o&le% with the %anufa(tu$in* p$o(e))" &n the conte0t of
this problem' e0plain why it ma(es sense to call alpha the proucer1s ris( an beta the consumer1s ris(.
4e %i*ht &e $e0e(tin* thi) without *ood $ea)on &ut (han(e) fo$ that would &e le)) than "0!" -o the$e i) a $i)k that the
p$odu(e$ will *o th$ou*h the t$ou&le of (he(kin* the p$o(e)) fo$ %i)take) &ut the$e i) $eall/ none" That i) wh/ we (all thi)
t/pe of e$$o$ when we $e0e(t H0 &ut H0 i) t$ue the G p$odu(e$B) $i)kG" The othe$ e$$o$ would &e failin* to $e(o*nie the
p$o&le% althou*h the$e i) one, i"e" failin* to $e0e(t H0 when it i) in fa(t fal)e the G(on)u%e$) $i)kG in thi) (a)e",
MR*-4. 2hy are Large Samples *etter
Statisticians prefer large samples. Fescribe briefly the effect of increasing the si9e of a sample "or the number of
subHects in an e0periment# on each of the following5
1.$he margin of error of a 4%8 confience inter!al
the P-!alue of a test' when the null hypothesis "Ho# is false an all facts about the population remain unchange as
n increases. de($ea)e)
7ptional $he power of a fi0e le!el a test' when a' the alternati!e hypothesis' an all facts about the population
remain unchange. in($ea)e)
Confidence intervals connection
2EB-K. Highway Statistics
&n 1440 the a!erage passenger !ehicle was ri!en 10.; thousan miles' as reporte by the =.S. <eeral Highway
Aministration in Highway Statistics. A ranom sample of %00 passenger !ehicles ha a mean of 10.1 thousan
miles ri!en for last year. Let m enote last year1s mean istance for all !ehicles.
1.Perform the hypothesis test
Ho5 m:10.; thousan miles
Ha5 m M10.; thousan miles
at the %8 significance le!el. Assume last year1s stanar e!iation of istances ri!en for all passenger !ehicles is
>.0 thousan miles.
'=500, %ean=!0"#, )=A the p-value i) the a$ea to the left of !0"! and to the $i*ht of !0"5 whi(h i) 2D"22.="459" -o the$e
i) no $ea)on to $e0e(t H0"D=ntp$
-.<in a 4%8 confience inter!al for m.
6."57A, !0"A298
;.Foes the !alue of 10.; thousan miles' hypothesi9e for the mean m' in the null hypothesis of .uestion 1' lie within
your confience inter!al from .uestion -3 :e)"
,. Repeat .uestions 1-; if the %00 passenger !ehicles sample were ri!en a mean of 10.4 thousan miles last year.
He0e(t Ho" 2onf" int:6!0"#7A,!!"4298, no"
%.*ase on your obser!ations in the first four .uestions' complete the following statements concerning the
relationship between a two-taile hypothesis test'
H05 m : m0
Ha5 m N m0'
at the significance le!el a an a "1- a#-le!el confience inter!al of m5
i. &f m N lies within the "1- a#-le!el confience for m' then the null hypothesis "will' will not # be reHecte.
ii. &f m N lies outsie the "1- a#-le!el confience for m' then the null hypothesis "will' will not # be reHecte.
MRA-K. SA$M
Let us suppose that scores on the mathematics part of the Scholastic Assessment $est "SA$M# in the absence of
coaching !ary normally with mean m : ,K% an s : 100. Suppose also that coaching may change m but oes not
change s. An increase in the SA$M score from ,K% to ,KJ is of no importance in see(ing amission to college' but
this unimportant change can be statistically !ery significant if it occurs in a large enough sample.
$o see this' calculate the p-!alue for the test of
Ho5 m : ,K%
Ha5 m greater than ,K%
in each of the following situations5
1.A coaching ser!ice coaches 100 stuents. $heir SA$M scores a!erage y-bar : ,KJ.
@="#9
-.*y the ne0t year' the ser!ice has coache 1000 stuents. $heir SA$M scores a!erage y-bar : ,KJ.
@="!7!
;.An a!ertising campaign brings the number of stuents coache to 10'000. $heir a!erage score is still y-bar : ,KJ.
@="00! 6)3$ootn=! in thi) (a)e )o we a$e # -I awa/ fo$% the %ean8
MRA-J. SA$M )onfience &nter!al
Gi!e a 448 confience inter!al for the mean SA$M score m after coaching in each part of the pre!ious e0ercise.
<or large samples' the confience inter!al tells us' 6Ies' the mean score is higher than ,K% after coaching' but only
by a small amount.6
1.A coaching ser!ice coaches 100 stuents. $heir SA$M scores a!erage is ,KJ.
6455,4.58
-.*y the ne0t year' the ser!ice has coache 1000 stuents. $heir SA$M scores a!erage is ,KJ.
64A., 49!8
;.An a!ertising campaign brings the number of stuents coache to 10'000. $heir a!erage score is still ,KJ.
6474,4778
Hypothesis tests for proportions
M*S-K. $he Pepsi )hallenge
6$a(e the Pepsi )hallenge6 was a mar(eting campaign use recently by the Pepsi-)ola )ompany. )oca-)ola
rin(ers participate in a blin taste test where they were as(e to taste unmar(e cups of Pepsi an )o(e an
were as(e to select their fa!orite. &n one Pepsi tele!ision commercial' an announcer states that 6in recent blin
taste tests' more than half the Fiet )o(e rin(ers sur!eye sai they preferre the taste of Fiet Pepsi6. Suppose 100
Fiet )o(e rin(ers too( the Pepsi )hallenge an %> preferre the taste of Fiet Pepsi. $est the hypothesis that more
than half of all Fiet )o(e rin(ers will select Fiet Pepsi in the blin taste test. =se alpha : .0%. 2hat are the
conse.uences of the results from )oca-)ola1s perspecti!e3
Ho: p="5
Ha: p>"5
2" phat="5A u)in* the tool the @6phat>"5A8="4#
-o we fail to $e0e(t the h/pothe)i) that 50, of diet (oke d$inke$) will (hoo)e diet (oke" -o we $eall/ (annot (lai% that diet
(oke d$inke$) favo$ Iiet @ep)i"
#" the $e)ult on the othe$ end )how) a) if (oke d$inke$) (ould not di)tin*ui)h the ta)te) )o it look) like thei$ p$efe$en(e i) fo$
the &$and and not the ta)te"
M)S-%. S(in &mpro!ements
Pon1s Age-Fefying )omple0' a cream with alpha-hyro0y aci' a!ertises that it can reuce wrin(les an impro!e
the s(in. &n a stuy publishe in Archi!es of Fermatology "?une 144>#' ;; women o!er age ,0 use a cream with
alpha-hyro0y aci for twenty-two wee(s. At the en of the stuy perio' -; of the women e0hibite s(in
impro!ement "as Huge by a ermatologist#.
&s this e!ience that the cream will impro!e the s(in of more than >08 of women o!er age ,03 $est using alpha :.0%.
!"H0: p="A
Ha: p>"A
2" n=##, phat=2#3##="7
#" -et den)it/ toll p="A, n=##
,.<in an interpret the p-!alue of the test. @="!2 )o at 55 level the$e i) not enou*h eviden(e to &elieve that p>"A" 4e
fail to $e0e(t H0"
The p-value tell) u) what i) the (han(e that in a )a%ple of ## wo%en 2# o$ %o$e will )ee i%p$ove%ent if in the population
A0, of the wo%en ended up with i%p$oved )kin"
M)S-K. Political caniates
$o get their names on the ballot of a local election' political caniates often must obtain petitions bearing the
signatures of a minimum number of registere !oters. &n Pinellas )ounty' <loria' a certain political caniate
obtaine petitions with 1J'-00 signatures "St. Petersburg $imes' Apr. K' 144-#. $o !erify that the names on the
petitions were signe by actual registere !oters' election officials ranomly sample 100 of the names an chec(e
each for authenticity. 7nly two were in!ali signatures.
&s 4J out of 100 !erifie signatures sufficient to belie!e that more than 1K'000 of the total 1J'-00 signatures are
!ali3
1. H0: p=17,000/18200=.93
Ha: p>.93
2. N=100
phat=.98
3. Chances to see 98% in the sample if p=.93 e!al to the p"#al!e $hich is p=.02% so the&e is onl' a 2.% % chance to see
this if p(=.93 that is if $e ha#e 17,000 o& less #ali) si*nat!&es. +he&efo&e the&e is a *oo) &eason to ,elie#e that $e ha#e
eno!*h e#i)ence fo& mo&e than 17.00 #ali) si*nat!&es ,ase) on the sample.
Repeat part "1# if only 1>'000 !ali signatures are re.uire. Thi) p-value hen
H0:p=!A0003!9200="99 i) %u(h )%alle$ )o ou$ eviden(e i) even %o$e )i*nifi(ant )uppo$tin* the (lai% that at lea)t !A000
)i*natu$e) a$e valid ove$all"
M)S-J. Earth.ua(es
Earth.ua(es are not uncommon in )alifornia. An article in the Annals of the Association of American Geographers
"?une 144-# in!estigate many factors that )alifornia resients consier when purchasing earth.ua(e insurance.
$he sur!ey re!eale that only 1;; of ;;K ranomly selecte resiences in Los Angeles )ounty were protecte by
earth.ua(e insurance.
2hat are the appropriate null an alternati!e hypotheses to test the research hypothesis that less than ,08 of the
resients of Los Angeles )ounty were protecte by earth.ua(e insurance3
!"H0J p="4
Ha: p<"4
Fo the ata pro!ie sufficient e!ience to support the research hypothesis3 =se m :.10.
-. phat=!##3!#7="#.4A
n=##7
@6phat<"#.4A8=0"4! )o the$e i) a *ood (han(e to )ee a )a%ple p$opo$tion a) low a) #."4A, o$ lowe$ even if the t$ue
population p$opo$tion i)"4" 'ot enou*h eviden(e to $e0e(t H0 at an/ level not even at !0, level"

You might also like