ex. kilo runs a track meaiuring 400m.

His
running tines anenor ally distr i byted with o 40s

Find his average running time if 33°% of

his vuns took less +hn l20.6 sec.

o: 10sey P(242) = 0.3300

2: -0.44
→-0.44 =J20.4-A
→120.o -= -4.4
u 120.6+ 4-4 125 sec.

EMPIRICAL RULE

- Very uce ful rule of thumb for normally distributtd sets of data values.

1approxC8 ot eobservathions
lie in [M-o, utoJ.
2) approx 457 of the observationslit in [a-20, Mt20.
3 pp rox 94.7%, of the obce rva tion ie in u-3o. M + 3o].

M-3o -lo M to t26 u+36

L 45
q9-7°%

APINGbiSTRIBUTIONS
WHAT ARE SAMPLE DISTRIBITIONS?

the probabiity distributon of a cample statisic that's formed when

samples of sie n aru peatdly taten from a populaton.nahyods&
Poputation
sample
Sample 3
Danp
2

sample
cample sample
4
 If he samplestatiìcic s thesamplemean,thenthedistributionis the
sampling distr ibution of the cample means (%X, X,x4,Xç.X,).
SAMPLING DISTRIBVTION MEAN R VARIANCE

mean of sampling distribution of X is cqual to the mean of the distribution

X.

- the variane of X is cmaler than Hhat ot x, that is o< o whun

sampling is done with neplaeneat or from an inhinie po pulaton:

7631

the standard deviation of the sampling distribution ot he sample means is

called the ctand ard error of the mman.
whun sampling is done withoyt placment or from a nie populahion of
Siu N, thevar icne is

but can be igno rtd if n < 0os N,.

ex. Tuwotrom the population valus {s 0,IS, 20y art
randomly celecttd wih eplacenunt.
Find the mean, variane b standard deviation ot

the population.
ne 12.S

o st)+ o( 4)+Is(4)+ o()-12s:31.2

List all Samples of sie n2 & calculat the mean-

Sample Samplemean Somple Sample mean

S IS. S

IS, 10 |2-5
IS, IS

20

0,S 12-S

(0, 10 10 20, 10

,IS 20,1S

I0, 0 0.0
fi
 reat a probability distribution & nd the
ean & standard error of the sample uan.

| Probabiity
0.002X Io(0.1878)+ ILs (0100) +is(01)
2 0.1200 + s(0-10) 20(0.0L2)2s
3 0.I8S Or

4 0. 2s00
-, 12.r
IS 3 0-185
2 0.2SD +Is'(0.250o)+iro.1895)
0.0628 + ns(o.ro) +0(0.0bs)-ns
= Is-o24

or

Oz- Jisus 3.4528

THE CCNTRAL LIMIT THĘDREM
- If a sample cie n, that is suthiciently large, is toaken
from a population with any type of distribution that

has a man u & standard dwiabon -o, the

sample mans have a normal distiouthon.
- If thepopulation itselt is normally distributed, the
Sample mans will have a no rmal distri bution for X

any sam ple size n.

- for eithercace
Ox

ex. the heights of magnolia loushes have a me an

hight of 8Et & standard cdw. of 0.t.38

buskes arc randonly selected 2 the mean of

cach sanple is determid.

Find th mean standard trror.

fi
 Find the probability that the mean height of

the 3s bushes is less than .8 tt

be cause 4he sample Siu is
Z: :-.82
Suf cienty large, th sapling P(R<1.9): P(c82)
dlistnbion can be approxiatd = 0.0344

by he normal distri buhon

AON
INFERENTIAL STATISTICS

- neing sample data to draw conclusions about th popula tion data.

yhypothesis tesing
conarnd witth det ining the vaidity of th assertions made
on the Paramettrs of the population or on he distriution of

the variable of intevest.

L estmation
• an arta of inirenthal statistis wher sample masures are used

to determun the true values of uknown populahon measures.

ESTIMATLON OF THE POPVLATION MEAN

point estimation
y compuRSfor a singleapproximatevalue fom a rando sanple to cpesent
an unknown populaion maJur.

L te vale calcwlated is called the point estimae.

Ly the vale/ formula used to get it is cailed te point estimator.

•mean? X=, Xi
v• arianu;s' ,(K;-*)
X
* probablity:p: n whr x no. ofsuccesses
- interval estmahon
L constructs an interval of possible values from a vandom smple to
esima an unknown pavameter of interest.

L theinervaliscalledthe intrvalestinatk. 3 s el
L the ale /formu la used to get it is called th interval estimatr.
fifi
fi
 L Whn a degr ot conhdenceis attached to an inteval estinate, the
ioterva is called a conhiene interval estimate.

dlegre/evel of con dnce deseribes how conhdent can

we be that th interval contains the tue value.

→ (l-a) xI00°%.
• Signi cane level is how willing you art to be wrong &

denoted by a.

• the margın of eror e dnotes the vange.

→ (Ā-e, Kte), where

o forknowno, e Zan
oforunkown o, e: tae , whre df n-1

ex. In a then park a random cample of 49

visitors g ave a man of P300 & a standard

deviation of PG00.
Find te pointestimationof mony spent per
isi tor.
X=P3400
Constrnct a 9s% conhdene inerval for it-

6 unknown
d.= I- 0.45- 0.08, df = 44-l= 4g

Based
onanother
table -
= 2.004 . |12.2
→ (3100 -172- 2, 310oo+ 112.2)

- (3s27.8, 2872.2)
* o iS the s. d. of a populahion & s is the s.d. of sam ple si.

ESTÍMATLON OF TH6 POPULATION PROPDRT ION

- proportion is a traction of a population/sample sie that has a certain
characterisies.
pointestimathon e
• haspointestimator p , xno. successes
of B n samplesie.
L jnteral es
timahon savo J a 4 Jn 3
• (ệ-e,p te) wun
fi
fi
 ex. 830 eenages out of a vandom sample of SDD

have rtgular acess to e iternet

Find the point es himate of true pporhons

of teenags who have access.

310.

Constmct a 9s% conidene inurval forit.

d |-0.9S 0.0S

e= Zo-os, D. (0.39)
SYpposedly -196 but juct

9e+ th posiie
→ (0-0b-0.D4IS, 0-b + D. 0415)
- (0.618s, 0.7015)
SAMPLE SIZE DETERMINATION
San ple sie dteminahion for the mean
n: (Z«s , het e= emr ofmargin
ex. How large a cample shod you tae to be a

Con6idnt that the eshimated tue man of

intes of sleep ill ditfur by no mnorHhan
0.30, pl4s the stondard dw. is 1.9J ain

<- -0.9S-0.0r
Z.as/2 -1-9b
Rovnd up for --→)n :(1.46 ) 62.3)0
San ple sie
- sample sie detemination for theprrportidn
n: Zenp-)
ex. How large a sample sie should you take to
be 1% confdent hat G6, of 4eens have

vegular access to the intent, plus the eshnatd

tue proportion will ditfer by 0.03?
= -0.4J=0.0
Zo.0572 -|.46
 -iAL) o- (0.342: qr
n .03*

- Sanple siu detwminahion without info about a prvious shudy

(Zay)
4e ex. How large a san ple to be 9s%

Con Gdent with 21, ewor of margin.

d |-.4 0.0S
Zoos/a : -1.46

n= (-196) 240l
A(0.02 =24D|
- sample sie determinahon witth knounpopulathonsize & sampk itardavd deiahon

n: a/z)

tx. Populaton aiu is 2000 with s.d. 0.shous.

How large a sample sie should you take to

be 45. Con dent to not hae the

esinate ditfer over 0-10.

d: |- 095 =0.0
Zo.o/2 -|96
n: (C1.96)(0.r) 42
0-l0+ (-1947(or)
2000

HYFOTHEGIS ESTING
TEST OF HYPDTHESIS

- Statisticalhypothesis
L→ Null hypothesis (Ho)

• either a statement of equality no diteren a, no chanu,

oelathionshipI or o effeet.
•Hoi 8-6o,0> .,, 040,
y Alternahehypothesis(Ha
contrasting statement believed to be tue whenever Hh null
hy pothsis is rejected.

•Ha:0 Đ,, 8>Oa, 20a

fi
 Types of tests of hypothesic

One-tailed test
3.sts a nul hy
pothesisaguinst a diecional alternahve
bypotesis,

Ha:0>,, .
,Two-tailedtest
• tess a hull hypothesis against a ondirutional alternaive
hypothess.

One-taijed es (eEt) Tuo taied test One-taied st Cnght)

H, M:Po Ho u:,
Ha: M>Mo

eX. Themanager of a bank clains that teir

non-ATM ustomers, onaveragt, wait for at

moSt 40 win. b efor they art seved

Ho* 4 10
|Contradicts Ho

ex. Th average dauly sales of a store is 12 units

The store owhr believed that it inceased

afwr thy adopd a new stratgy

Ho ui 42
Ha us12
TYPES OF ERROR

- Tupc I eror
L an eror of rjecting the null hypothecis that is true.
(Reject Hol Ho is true)

- Type I ewor
y an evor of taing to njet (AKA aceping) a null hypothesis that is
false.
L (Fait to viect Hol Ho is false)
 ex. Ho: The detendant is ot guilty
Ha: The detendant is guilhy

Type I error Thecourtconcludes hat

an innoOent efendant is

guilty-

Tpe I ewori The ort concluderthat

a guity detendant is

innoent.

- uness population data is available, you can't determine thetuth or. falsity of the
null & alturnative hy potheses. Howerr, the probabilihes ot committing errors can be

evaluaed.

4 level of signifcane
• measurts the visk of rejecting a tvue null hypothesis

• P(TypeI ror): o
L level of con hdee
• probability of mnaking th corret decision of accephing a true
null hypothesis.

Probability of coming a Typei evor measurts the nisk of failing to vejet

('acuping') a t alse nul hypothesis.

•P(TypeI wnr) P

-« &B are inverely related. Both can be eduud by increasing tthe sanple sie.
TEST HYPOTHESTS ABOT PARAMETEKS OF ONG Po PULATTON

test hypothesisabout th populahon men whn is known

y teststahishc: Z-X-

Ha Decisionae rject H, if

Z2-Z oj2 or Z > Zalz

Z > Za
 ex. A manutachirer of sports equipmunt developed

a nw fsling lin clained to have a mean

breaking strength ot Bk with s.d. of 0.stg.

Tes thehypothesisthat M:8kg 4gainst Mt Sky if

a random sampll of SD ines are tested found

to hare a brenting stngt nean 18kg. UseD.0

level of signi can

H, 8rg H,: 8kg

d = 0.0)

Z3s/so : -2.83
UsingHa RL, --- -→ -2a/2 .o1/2 =
-2o.01 2.5S
Sine - 2.93<-2.s 5
Conclusion? at de°%, thue is enough

endna to reject te clain that

themean brtaking strngth of the

nw Asling lint is equal to 8kg-

- fest hupothesis about th populathionmean when o is unknown?

Ly test staishe
t: kS
Ha Decision aeiiect Ho if

t<-t, n4or t ta
t>tajn
tl-ta,n1
x. It is rcommuhded that cild 43yrs dd have
a 1070kcal of enrgy intake. A semple of 78

children reportod an avera gt enugy intake of

only 4o kal vith s.d. of190kcal Can you

conctudu that the mean energy intale of the

sample is lower th an the reomnunded Uses

lul ofsigni cance.

Ho M= lorokal Ha:|010Kcal
x440kcal sU9Dkeal d0-0s
fi
fi
 t qu0-1070
140//76
= - 29.2g

UsingHa u<H, --+-ta,n toos, n 4.664

Sine -29.25 <-1. bu4
Concusion at d*Sl%, thr isehough
idene to reject that themean

enrgy ntate of 4his

sample is not cqual to

1070kca.
- 4est hypothsis about the population proporhon

test stahshe 2: fPo

Ha Decision rulei reject Mo if

p Po Z-lan or 2 > Zalz

P> P Z > Za
Z<-ZA
ex An ageny claims that overseas Filipinoscomprnse
of less than ho tourists in the country- A sanple

of Goo tourists who visited the Boracay beaches was

selected, 30 of which wte overseas Filipinos. poes

the data Sugport the agenys claim? Use S%lew

of Sigi cance.

Hqi p<to n G0

po 10 p:300 d=0.oS
Z = 4.08
Vio(ovoo
Using Hai p<p. ZaZo.os - I-G4
Sina -4.08 - I-4S
Concsion at d sl, thurt is enough
emdence to clain thatoverscas
FilipinosS compnse of less than

Vo tourists.
fi
 TESTING HYPOTHESTS ABOUTPARANETERSDF Tub POPULATLONS

case 43 inde pendent samples

L, tesing hy pothis aboutthe ditlerihe oftuwo

populahon
means -e,

•Z (3,-*)-d

Ha Deision nae ; rjeet Ho if

2-an or 2> Zn

|Kik do 2 -Z4
L teeting hypothusisabout h ditference ot twopopulahonmeans A-hz, &

6, unknown lout assumed to be equali

.t= (K-x)-do

poole d sample variane

nn2 -2
• for tw0 populations, df n,+ n,- 2
Ha Decision rues rejcct Ho if

t<-t(o,en,-2)
or tt,(hn,)
do
-t< do
t>ta,(n,na2)
t<-ta.cn,tne)
K. A vandom sample of 100wOmen trom nual urban
artas cach have these staishcs regarding the age

thy rst maried o

residene mean s tandard dev.

21 4-S
rural
urban 23 S.I
Does ts provide evidene that ural womnmany
earlier han urban women?Usesl%level of sigkan.

|Sina were trying toprove -+ HoM,-M,0

Hhat mral marry younger, d = 0.0s X, =2| K,=23 $,=s.)
Ha states that k - Mz (i00-)4-s+(o0-)S 2.3. 13
(00+100-2
(21-23) -0
|ic less +han o
t DB()-2.94
fi
 Uking4,-k, <d 4-ta,(mang-t)to.0s, nt -1.6
Sina - 2.44 <-1.lb
Conclusion; at q:s, thure is enoughevidence
toSupport te clain that ural
womn marny earlier than urban

women.

* Another way to chck a claim is through p-value. The nul hypothusis is rejeced it
the p value is less than the si gni can level (use an onine p-value calu).

ex. Sam ques thon as previous Cxample.

p-value 0.001830
Sine o.oo18 36( 0.0S
Concusion; at af % thue isenonghevidene
to support the c laim that rural

Wonen mary earlir than uban

VOmn.
*example for two populations Mi -as

0
with Ehown
o.
oors are assignd to est cach way. Does hu
data incate wax is mort effecthve than

wax2 at sigiicane level s%?

Wax sample on in months populahon s. d.

3 0.33
2 2.4 0.30

d:0.05 ,:3 : 2.4G,=0.33

,-03%
Z - (3-2.9)- o
0.330-3
+

La Zo.os -l-olS
Sin
Concdusion ; at os% tht is not nough
ewidnce to sufport te claim that
wax 1 is mor efhiáent than

Wax 2.
fl
fi
 cAse 2: rlaed or paired samples
Lyindepen dent samples taken from a normal populahon with sie n;

•td-d, A: averagt of odifferences, Sa S.d. of ditferenes

Ha Decision Rule:Reject H, it;

Bi-Be tdo te-tg.cn- or t>tę,(n-)

t>ta,(n)
t<-te, ch-)
tx. A ntritionist claims that indinduals tend to eport

deureasing dietayy itake the more thy art

interviewed Using l0% level of signihcane, es+
this hypothesis.

Student 4 2 3
s 18
day 1 19052237 |(831241 22l 428 (G02

day 2 |1G53)41 10026 1999| 1097 243|u4

|dayl- day2 |244|5s 63 15 331 -22I-79
Hai MA>0 d= 0.]0
143.250
t 48343 1.12
ta,cn) 2 to.l, = |4|S

Since l.12 X I.4 IS

OR

p-value= 0-i44832
Sine o.(44992 o.
Conciusion at a 10% thue is notenaugh
eviden to support the claim
that migher dietary intae is
observed on day 1.

testing hypothesis about diterene ot twO populahon prop0rions -Pz

SFAG+4)

p: n, + n2 I
pooledestimattot po
fi
 Ha Decisionute ject Ho if
2<-Za2 or Z>Z2
Z > Za
PPe <do

ex. A random sanple of ns hypertensi & IsD normal sen ios

werasked it thu readnutrihonlabels.At s

lerel of siqni cance, is hre enough evideha to

say ther is signiicant atterence of hyper tensivel

nomal seniors who nad nutnion labels?

reads nutition labels
hea lth staMS to tal

hypertensivt 35 q0 128

normal 38 112

total 13 02 25
Ho P, -Pz 0 Ha pi-fe t0 d > 0.DS

p 0.2s 0.253 0.20s

7:02g- 0.263

Zap Zo.o1s *4.4b

Sine 0.DS .I4%
Concusion at o So thee is ot enough
wideha to Support the claim that

thrt is sigvi tant diterence of

hyperensive & normal seror s who

vead nutiion labels

iORRELATONCORHVENTG 1810FPORHO)
CORRELATLON COEFFICLENT (P)
made of "o (together) & relation.
- correlahon is positive it values inurea se togethr.

- correlathion is negaive when one deuases white the othe incrases.

fi
fi