SlideShare una empresa de Scribd logo
1 de 26
H YP O TH E S IS TE S TIN G
In th is s e s s ion … .

 Wh at is h yp oth e s is te s ting?
 Inte rp re ting and s e le cting s ignificance le ve l
  P R O BABILITY
 Typ e I and Typ e II e rrors
 O ne taile d and twoTIO NsS
  D IS TR IBU taile d te ts
 H yp oth e s is te s ts for p op u lation m e an
 H yp oth e s is te s ts for p op u lation p rop ortion
 H yp oth e s is te s ts for p op u lation s tand ard d e viation
Wh at is H yp oth e s is Te s ting?


H yp oth e s is te s ting re fe rs to
• M aking an as s u m p tion, calle d h yp oth e s is , ab ou t a
                              - the B-school
   p op u lation p aram e te r.
• C olle cting s am p le d ata.
• C alcu lating a s am p le s tatis tic.
• U s ing th e s am p le s tatis tic to e valu ate th e h yp oth e s is (h ow
   like ly is it th at ou r h yp oth e s ize d p aram e te r is corre ct. To
   te s t th e valid ity of ou r as s u m p tion we d e te rm ine th e
   d iffe re nce b e twe e n th e h yp oth e s ize d p aram e te r valu e and
   th e s am p le valu e .)
H YP O T H E
                                         S IS
                                     T E S T IN G
    N u ll h yp oth e s is , H 0             Alte rnative
                                             h yp oth e s is ,H A
S tate th e h yp oth e s ize d valu e of   All p os s ib le alte rnative s oth e r
th e p aram e te r b e fore s am p ling.    th an th e nu ll h yp oth e s is .
Th e as s u m p tion we wis h to te s t    E .g µ ≠ 20
(or th e as s u m p tion we are trying
to re j ct)
       e                                    µ > 20
E .g p op u lation m e an µ = 20           µ < 20
 Th e re is no d iffe re nce b e twe e n   Th e re is a d iffe re nce b e twe e n
coke and d ie t coke                        coke and d ie t coke
N u ll H yp oth e s is


Th e nu ll h yp oth e s is H 0 re p re s e nts a th e ory th at h as b e e n

                            - the B-school
p u t forward e ith e r b e cau s e it is b e lie ve d to b e tru e or
b e cau s e it is u s e d as a b as is for an argu m e nt and h as
not b e e n p rove n. F or e xam p le , in a clinical trial of a ne w
d ru g, th e nu ll h yp oth e s is m igh t b e th at th e ne w d ru g is no
b e tte r, on ave rage , th an th e cu rre nt d ru g. We wou ld write
 H 0: th e re is no d iffe re nce b e twe e n th e two d ru gs on an
ave rage .
Alte rnative H yp oth e s is

Th e alte rnative h yp oth e s is , H A, is a s tate m e nt of wh at a
s tatis tical h yp oth e s is te s t is s e t u p to e s tab lis h . F or e xam p le ,
                                - the B-school
in th e clinical trial of a ne w d ru g, th e alte rnative h yp oth e s is
m igh t b e th at th e ne w d ru g h as a d iffe re nt e ffe ct, on ave rage ,
com p are d to th at of th e cu rre nt d ru g. We wou ld write
 H A: th e two d ru gs h ave d iffe re nt e ffe cts , on ave rage .
or
 H A: th e ne w d ru g is b e tte r th an th e cu rre nt d ru g, on ave rage .

Th e re s u lt of a h yp oth e s is te s t:
‘R e j ct H 0 in favou r of H A’ O R ‘D o not re j ct H 0’
     e                                           e
S e le cting and inte rp re ting s ignificance
   le ve l
• D e cid ing on a crite rion for acce p ting or re j cting th e nu ll
                                                             e
   h yp oth e s is .
                               - the B-school
• S ignificance le ve l re fe rs to th e p e rce ntage of s am p le m e ans
   th at is ou ts id e ce rtain p re s crib e d lim its . E .g te s ting a
   h yp oth e s is at 5% le ve l of s ignificance m e ans
 th at we re j ct th e nu ll h yp oth e s is if it falls in th e two re gions
                  e
   of are a 0.025.
 D o not re j ct th e nu ll h yp oth e s is if it falls with in th e re gion of
                 e
   are a 0.95.
5. Th e h igh e r th e le ve l of s ignificance , th e h igh e r is th e
   p rob ab ility of re j cting th e nu ll h yp oth e s is wh e n it is tru e .
                        e
   (acce p tance re gion narrows )
Typ e I and Typ e II E rrors

• Typ e I e rror re fe rs to th e s itu ation wh e n we re j ct th e nu ll
                                                                e
   h yp oth e s is wh e n it is tru e (H 0 is wrongly re j cte d ).
                                                            e
                                   - the B-school
   e .g H 0: th e re is no d iffe re nce b e twe e n th e two d ru gs on
   ave rage .
   Typ e I e rror will occu r if we conclu d e th at th e two d ru gs
   p rod u ce d iffe re nt e ffe cts wh e n actu ally th e re is n’t a d iffe re nce .
   P rob (Typ e I e rror) = s ignificance le ve l = α
2. Typ e II e rror re fe rs to th e s itu ation wh e n we acce p t th e nu ll
   h yp oth e s is wh e n it is fals e .
   H 0: th e re is no d iffe re nce b e twe e n th e two d ru gs on ave rage .
   Typ e II e rror will occu r if we conclu d e th at th e two d ru gs
   p rod u ce th e s am e e ffe ct wh e n actu ally th e re is a d iffe re nce .
   P rob (Typ e II e rror) = ß
Typ e I and Typ e II E rrors – E xam p le

   You r nu ll h yp oth e s is is th at th e b atte ry for a h e art
   p ace m ake r h as an ave rage life of 300 d ays , with th e
                            - the B-school
   alte rnative h yp oth e s is th at th e ave rage life is m ore th an
   300 d ays . You are th e qu ality control m anage r for th e
   b atte ry m anu factu re r.
(b)Wou ld you rath e r m ake a Typ e I e rror or a Typ e II e rror?
(c)Bas e d on you r ans we r to p art (a), s h ou ld you u s e a h igh
   or low s ignificance le ve l?
Typ e I and Typ e II E rrors – E xam p le

   G ive n H 0 : ave rage life of p ace m ake r = 300 d ays , and
   H A: Ave rage life of p ace m ake r > 300 d ays
                             - the eB-school is fals e i.e
(b)It is b e tte r to m ake a Typ e II rror (wh e re H   0
   ave rage life is actu ally m ore th an 300 d ays b u t we acce p t
   H 0 and as s u m e th at th e ave rage life is e qu al to 300 d ays )

(c)As we incre as e th e s ignificance le ve l (α) we incre as e th e
   ch ance s of m aking a typ e I e rror. S ince h e re it is b e tte r to
   m ake a typ e II e rror we s h all ch oos e a low α.
Two Tail Te s t

Two taile d te s t will re j ct th e nu ll h yp oth e s is if th e s am p le
                             e
m e an is s ignificantly h igh e r or lowe r th an th e h yp oth e s ize d
                             - the B-school
m e an. Ap p rop riate wh e n H 0 : µ = µ0 and H A: µ ≠ µ0
e .g Th e m anu factu re r of ligh t b u lb s wants to p rod u ce ligh t
b u lb s with a m e an life of 1 000 h ou rs . If th e life tim e is s h orte r
h e will los e cu s tom e rs to th e com p e tition and if it is longe r
th e n h e will incu r a h igh cos t of p rod u ction. H e d oe s not want
to d e viate s ignificantly from 1 000 h ou rs in e ith e r d ire ction.
Th u s h e s e le cts th e h yp oth e s e s as
 H 0 : µ = 1 000 h ou rs and H A: µ ≠ 1 000 h ou rs
and u s e s a two tail te s t.
O ne Tail Te s t

A one -s id e d te s t is a s tatis tical h yp oth e s is te s t in wh ich th e
valu e s for wh ich we can re j ct th e nu ll h yp oth e s is , H 0 are
                                     e

                             - the B-school
locate d e ntire ly in one tail of th e p rob ab ility d is trib u tion.
Lowe r taile d te s t will re j ct th e nu ll h yp oth e s is if th e s am p le
                               e
m e an is s ignificantly lowe r th an th e h yp oth e s ize d m e an.
Ap p rop riate wh e n H 0 : µ = µ0 and H A: µ < µ0
e .g A wh ole s ale r b u ys ligh t b u lb s from th e m anu factu re r in
large lots and d e cid e s not to acce p t a lot u nle s s th e m e an life
is at le as t 1 000 h ou rs .
 H 0 : µ = 1 000 h ou rs and H A: µ < 1 000 h ou rs
and u s e s a lowe r tail te s t.
i.e h e re j cts H 0 only if th e m e an life of s am p le d b u lb s is
           e
s ignificantly b e low 1 000 h ou rs . (h e acce p ts H A and re j cts th e
                                                                     e
lot)
O ne Tail Te s t

U p p e r taile d te s t will re j ct th e nu ll h yp oth e s is if th e s am p le
                                 e
m e an is s ignificantly h igh e r th an th e h yp oth e s ize d m e an.
                              - the B-school
Ap p rop riate wh e n H 0 : µ = µ0 and H A: µ > µ0
e .g A h igh way s afe ty e ngine e r d e cid e s to te s t th e load
b e aring cap acity of a 20 ye ar old b rid ge . Th e m inim u m load -
b e aring cap acity of th e b rid ge m u s t b e at le as t 1 0 tons .
H 0 : µ = 1 0 tons and H A: µ > 1 0 tons
and u s e s an u p p e r tail te s t.
i.e h e re j cts H 0 only if th e m e an load b e aring cap acity of th e
           e
b rid ge is s ignificantly h igh e r th an 1 0 tons .
H yp oth e s is te s t for p op u lation m e an

                                              n ( x − µ0 )
H 0 : µ = µ0 and Te s t s tatis tic =
                                  ∆
                                                   s

                     e        - the B-school
F or H A: µ > µ0, re j ct H 0 if > t n −1,α
                               ∆

F or H A: µ < µ0, re j ct H 0 if < −t n −1,α
                     e         ∆

                               ∆ > t n −1,α 2
F or H A: µ ≠ µ0, re j ct H 0 if
                     e
                         t n −1,α by zα
F or n ≥ 30, re p lace
H yp oth e s is te s t for p op u lation m e an


A we igh t re d u cing p rogram th at inclu d e s a s trict d ie t and

                             - the B-school
e xe rcis e claim s on its online ad ve rtis e m e nt th at it can h e lp an
ave rage ove rwe igh t p e rs on los e 1 0 p ou nd s in th re e m onth s .
F ollowing th e p rogram ’s m e th od a grou p of twe lve ove rwe igh t
p e rs ons h ave los t 8.1 5.7 1 1 .6 1 2.9 3.8 5.9 7.8 9.1
7.0 8.2 9.3 and 8.0 p ou nd s in th re e m onth s . Te s t at 5%
le ve l of s ignificance wh e th e r th e p rogram ’s ad ve rtis e m e nt is
ove rs tating th e re ality.
H yp oth e s is te s t for p op u lation m e an


S olu tion:
H 0: µ = 1 0 (µ0) H A: µ < 1 0 (µ0)
n = 1 2, x(b ar) = 8.027, s = 2.536, α = 0.05
     12(8.075 − 10) 3.46 × −1.925
∆=                 =              = −2.62
        2.536           2.536
 C ritical t-valu e = -tn-1 ,α= - t1 1 ,0.05 = -2. 201 (TIN V)

 S ince ∆ < -tn-1 ,α we re j ct H 0 and conclu d e th at th e
                           e
 p rogram is ove rs tating th e re ality.
 (Wh at h ap p e ns if we take α = 0.01 ? Is th e p rogram
 ove rs tating th e re ality at 1 % s ignificance le ve l?)
H yp oth e s is te s t for p op u lation p rop ortion

                                              n ( p − p0 )
                                                  ˆ
H 0 : p = p 0 and Te s t s tatis tic =
                                   ∆
                                              p0 (1 − p0 )
F or H A: p > p 0   re j ct H
                       e
                              -if∆ > z B-school
                                 the   α
                            0


F or H A: p < p 0 re j ct H 0 if∆ < −zα
                     e
                                   ∆ > zα 2
F or H A: p ≠ p 0 re j ct H 0 if
                     e
H yp oth e s is te s t for p op u lation p rop ortion


A ke tch u p m anu factu re r is in th e p roce s s of d e cid ing wh e th e r
                             - the B-school
to p rod u ce an e xtra s p icy b rand . Th e com p any’s m arke ting
re s e arch d e p artm e nt u s e d a national te le p h one s u rve y of
6000 h ou s e h old s and fou nd th e e xtra s p icy ke tch u p wou ld b e
p u rch as e d b y 335 of th e m . A m u ch m ore e xte ns ive s tu d y
m ad e two ye ars ago s h owe d th at 5% of th e h ou s e h old s
wou ld p u rch as e th e b rand th e n. At a 2% s ignificance le ve l,
s h ou ld th e com p any conclu d e th at th e re is an incre as e d
inte re s t in th e e xtra-s p icy flavor?
H yp oth e s is te s t for p op u lation p rop ortion

                   335
n = 6000,      p=
               ˆ       = 0.05583
                  6000
                               - the B-school
H0 : p = 0.05( p0 ) H A : p > 0.05
       n ( p − p0 )
           ˆ              6000 × 0.00583
∆=                    =
       p0 (1 − p0 )         0.05 × 0.95
     77.459 × 0.00583
  =                     = 2.072
           0.218
α = 0.02
Zα (the critical value of Z ) = 2.05 (N O R M S IN V
Q ∆ > Z we reject H i.e th e cu rre nt inte re s t is s ignificantly gre ate r
         α                 0
                                     )
                               th an th e inte re s t of two ye ars ago.
H yp oth e s is te s t for p op u lation s tand ard
  d e viation
                                         (n − 1)s 2
H 0 : σ = σ 0 and Te s t s tatis tic ∆ =
                                           σ 02
                             - the B-school
F or H A: σ > σ0 re j ct H 0 if
                    e           ∆ > χ(2(−1),α
                                      n
                                        R)




F or H A: σ < σ0 re j ct H 0 if
                    e          ∆ < χ(2(−1),1−α
                                     n
                                       R)



                               ∆ < χ(2(−1),1−α 2
                                       R)
                                                    ∆ > χ(2(−1),α 2
                                                            R)
F or H A: σ ≠ σ0 re j ct H 0 if
                    e                n
                                                   or     n
H yp oth e s is te s t for com p aring two p op u lation
m e ans
C ons id e r two p op u lations with m e ans µ1 , µ2 and s tand ard d e viations σ1
and= σ12 .and µ x = µ2
µx µ
  1             2

                                       σx
                                            am σ x
p op u lation2 re s p e ctive ly.  - the B-school
              are th e m e ans of th e sand p ling d is trib u tions of p op u lation1 and
                                      1         2


                                                   d e note th e s tand ard e rrors of th e2
                                                                                         σ1 σ 22
µam xp2 ling d is trib u tions of th e m e ans .
sx−                                                                           σ X −X =       +
   1                                                                          1   2
                                                                                          n1 n2
          is th e m e an of th e d iffe re nce b e twe e n s am p le m e ans and
                                            ( X1 − X ) − ( µ − µ2 )H
              is th e corre s p ond ing = tand ard 2e rror.1
                                       ∆ s                        0

                                                       σ x −x
H 0 : µ1 = µ2 and te s t s tatis tic,                1   2


                                                  H e re ∆ d e note s th e s tand ard ize d
F or H A: µ1 > µ2 re j ct H 0 if ∆ > Z α
                          e                       d iffe re nce of s am p le m e ans
F or H A: µ1 < µ2 re j ct H 0 if∆ > - α 2
                     e          ∆ < ZZ α
F or H A: µ1 ≠ µ2 re j ct H 0 if
                     e
(d e cis ion m ake rs m ay b e conce rne d with p aram e te rs of two p op u lations
e .g d o fe m ale e m p loye e s re ce ive lowe r s alary th an th e ir m ale
H yp oth e s is te s t for com p aring p op u lation
m e ans

A s am p le of 32 m one y m arke t m u tu al fu nd s was ch os e n on
                          - the B-school
Janu ary 1 , 1 996 and th e ave rage annu al rate of re tu rn ove r
th e p as t 30 d ays was fou nd to b e 3.23% and th e s am p le
s tand ard d e viation was 0.51 % . A ye ar e arlie r a s am p le of 38
m one y-m arke t fu nd s s h owe d an ave rage rate of re tu rn of
4.36% and th e s am p le s tand ard d e viation was 0.84% . Is it
re as onab le to conclu d e (at α = 0.05) th at m one y-m arke t
inte re s t rate s d e cline d d u ring 1 995?
H yp oth e s is te s t for com p aring p op u lation
m e ans
n1 = 32, x1 = 3.23, σ 1 = 0.51          n2 = 38, x2 = 4.36, σ 2 = 0.84
H0 : µ1 = µ2            H A : µ1 < µ2

σ X1 − X 2   =
               σ 12 σ 22
                   +     =
                           0.26 0.71
                               +
                                        - the B-school
                                     = 0.026 = 0.163
               n1    n2     32   38
       ( x1 − x2 ) − ( µ1 − µ2 )H0     −1.13 − 0
∆=                                   =           = −6.92
                σ X1 − X 2              0.163
α = 0.05
Critical value of Z = −Zα = −1.64
Q ∆ < −Zα we reject H0 and conclude that there has
been a decline.
H yp oth e s is te s t for com p aring p op u lation
p rop ortions
                                                     p1
C ons id e r two s am p le s of s ize s n 1 and n 2 with           p2
                                                                  and      as th e re s p e ctive
p rop ortions of s u cce s s e s . Th e n

 p=
 ˆ
       n1p1 + n2 p2
         n1 + n2
                                        - the B-school
                         is th e e s tim ate d ove rall p rop ortion of s u cce s s e s in th e
                         two p op u lations .
               ˆˆ ˆˆ
               pq pq is th e e s tim ate d s tand ard e rror of th e d iffe re nce
 σ p1 − p2 =
  ˆ              +
               n1 n2 b e twe e n th e two p rop ortions .
                                             ( p1 − p2 ) − ( p1 − p2 )H0
H 0 : p 1 = p 2 and te s t s tatis tic,∆ =             σ x1 −x2
                                                        ˆ
F or H A: p 1 > p 2 re j ct H 0 if ∆ > Z α
                       e                            A training d ire ctor m ay wis h to
F or H A: p 1 < p 2 re j ct H 0 if ∆ < - Z α
                       e                            d e te rm ine if th e p rop ortion of
                                     ∆ > Zα 2       p rom otab le e m p loye e s at one office
F or H A: p 1 ≠ p 2 re j ct H 0 if
                       e                            is d iffe re nt from th at of anoth e r.
H yp oth e s is te s t for com p aring p op u lation
 p rop ortions

A large h ote l ch ain is trying to d e cid e wh e th e r to conve rt

                             - the B-school
m ore of its room s into non-s m oking room s . In a rand om
s am p le of 400 gu e s ts las t ye ar, 1 66 h ad re qu e s te d non-
s m oking room s . Th is ye ar 205 gu e s ts in a s am p le of 380
p re fe rre d th e non-s m oking room s . Wou ld you re com m e nd
th at th e h ote l ch ain conve rt m ore room s to non-s m oking?
S u p p ort you r re com m e nd ation b y te s ting th e ap p rop riate
h yp oth e s e s at a 0.01 le ve l of s ignificance .
H yp oth e s is te s t for com p aring p op u lation
    p rop ortions
                 166                                 205
n1 = 400, p1 =        = 0.415,      n2 = 380, p2 =       = 0.5395
                 400                                 380
H0 : p1 = p2       H A : p1 < p2

p=
ˆ              =
                                    - the B-school
   n1p1 + n2 p2 400 × 0.415 + 380 × 0.5395          (P rop ortion of s u cce s s
                                           = 0.4757 in th e two p op u lations )
     n1 + n2            400 + 380
                   1 1                            1      1 
σ p1 − p2 = pq  +  = 0.4757 × 0.5243 
 ˆ             ˆˆ                                       +       = 0.0358
                    n1 n2                         400 380 
α = 0.01                                                 Th e h ote l ch ain s h ou ld
Critical value of Z = −Zα = −2.32                        conve rt m ore room s to
        ( p1 − p2 ) − ( p1 − p2 )H0 −0.1245 − 0          non-s m oking room s as
∆=                                 =            = −3.48 th e re h as b e e n a
                  σ p1 − p2
                   ˆˆ ˆ               0.0358             s ignificant incre as e in th e
Q ∆ < −Zα we reject H0                                   nu m b e r of gu e s ts s e e king
                                                         non-s m oking room s .

Más contenido relacionado

Destacado

Tools of the Trade
Tools of the TradeTools of the Trade
Tools of the TradeLovemarks
 
The Attraction of Lovemarks
The Attraction of LovemarksThe Attraction of Lovemarks
The Attraction of LovemarksLovemarks
 
Start With Respect
Start With RespectStart With Respect
Start With RespectLovemarks
 
Priceless Value
Priceless ValuePriceless Value
Priceless ValueLovemarks
 
The Power of Love: Evidence of the Power of Emotional Advertising
The Power of Love: Evidence of the Power of Emotional AdvertisingThe Power of Love: Evidence of the Power of Emotional Advertising
The Power of Love: Evidence of the Power of Emotional AdvertisingLovemarks
 
All You Need Is Love: Quotes on Brand Love and Lovemarks.
All You Need Is Love: Quotes on Brand Love and Lovemarks.All You Need Is Love: Quotes on Brand Love and Lovemarks.
All You Need Is Love: Quotes on Brand Love and Lovemarks.Lovemarks
 
Perencanaan pembangunan desa
Perencanaan pembangunan desaPerencanaan pembangunan desa
Perencanaan pembangunan desaRooy Salamony
 
2011 03 07 Convaincre et fidéliser la clientèle en optimisant son site web - ...
2011 03 07 Convaincre et fidéliser la clientèle en optimisant son site web - ...2011 03 07 Convaincre et fidéliser la clientèle en optimisant son site web - ...
2011 03 07 Convaincre et fidéliser la clientèle en optimisant son site web - ...COMPETITIC
 
Batir sa strategie editoriale pour seduire ses clients et google - CCI Bordea...
Batir sa strategie editoriale pour seduire ses clients et google - CCI Bordea...Batir sa strategie editoriale pour seduire ses clients et google - CCI Bordea...
Batir sa strategie editoriale pour seduire ses clients et google - CCI Bordea...echangeurba
 
Competitic - visibilité sur le web - numerique en entreprise
Competitic - visibilité sur le web - numerique en entrepriseCompetitic - visibilité sur le web - numerique en entreprise
Competitic - visibilité sur le web - numerique en entrepriseCOMPETITIC
 
Obligations site internet
Obligations site internetObligations site internet
Obligations site internetCOMPETITIC
 
Guide de Trading Options Binaires
Guide de Trading Options BinairesGuide de Trading Options Binaires
Guide de Trading Options Binairesanyoption
 
Conférence "Optimiser son référencement Internet"
Conférence "Optimiser son référencement Internet"Conférence "Optimiser son référencement Internet"
Conférence "Optimiser son référencement Internet"Sylvie de Meeûs
 
2011 06 23 aspects juridiques ecommerce by competitic
2011 06 23 aspects juridiques ecommerce by competitic2011 06 23 aspects juridiques ecommerce by competitic
2011 06 23 aspects juridiques ecommerce by competiticCOMPETITIC
 
Le web marketing ou comment bien communiquer sur la toile
Le web marketing ou comment bien communiquer sur la toileLe web marketing ou comment bien communiquer sur la toile
Le web marketing ou comment bien communiquer sur la toileSylvie de Meeûs
 
Ressources numériques en bibliothèque
Ressources numériques en bibliothèqueRessources numériques en bibliothèque
Ressources numériques en bibliothèqueRenaud Chauvet
 

Destacado (20)

Tools of the Trade
Tools of the TradeTools of the Trade
Tools of the Trade
 
The Attraction of Lovemarks
The Attraction of LovemarksThe Attraction of Lovemarks
The Attraction of Lovemarks
 
Wish statements
Wish statementsWish statements
Wish statements
 
Start With Respect
Start With RespectStart With Respect
Start With Respect
 
Priceless Value
Priceless ValuePriceless Value
Priceless Value
 
The Power of Love: Evidence of the Power of Emotional Advertising
The Power of Love: Evidence of the Power of Emotional AdvertisingThe Power of Love: Evidence of the Power of Emotional Advertising
The Power of Love: Evidence of the Power of Emotional Advertising
 
Conclusiones
ConclusionesConclusiones
Conclusiones
 
All You Need Is Love: Quotes on Brand Love and Lovemarks.
All You Need Is Love: Quotes on Brand Love and Lovemarks.All You Need Is Love: Quotes on Brand Love and Lovemarks.
All You Need Is Love: Quotes on Brand Love and Lovemarks.
 
Perdes
PerdesPerdes
Perdes
 
Administrasi desa
Administrasi desaAdministrasi desa
Administrasi desa
 
Perencanaan pembangunan desa
Perencanaan pembangunan desaPerencanaan pembangunan desa
Perencanaan pembangunan desa
 
2011 03 07 Convaincre et fidéliser la clientèle en optimisant son site web - ...
2011 03 07 Convaincre et fidéliser la clientèle en optimisant son site web - ...2011 03 07 Convaincre et fidéliser la clientèle en optimisant son site web - ...
2011 03 07 Convaincre et fidéliser la clientèle en optimisant son site web - ...
 
Batir sa strategie editoriale pour seduire ses clients et google - CCI Bordea...
Batir sa strategie editoriale pour seduire ses clients et google - CCI Bordea...Batir sa strategie editoriale pour seduire ses clients et google - CCI Bordea...
Batir sa strategie editoriale pour seduire ses clients et google - CCI Bordea...
 
Competitic - visibilité sur le web - numerique en entreprise
Competitic - visibilité sur le web - numerique en entrepriseCompetitic - visibilité sur le web - numerique en entreprise
Competitic - visibilité sur le web - numerique en entreprise
 
Obligations site internet
Obligations site internetObligations site internet
Obligations site internet
 
Guide de Trading Options Binaires
Guide de Trading Options BinairesGuide de Trading Options Binaires
Guide de Trading Options Binaires
 
Conférence "Optimiser son référencement Internet"
Conférence "Optimiser son référencement Internet"Conférence "Optimiser son référencement Internet"
Conférence "Optimiser son référencement Internet"
 
2011 06 23 aspects juridiques ecommerce by competitic
2011 06 23 aspects juridiques ecommerce by competitic2011 06 23 aspects juridiques ecommerce by competitic
2011 06 23 aspects juridiques ecommerce by competitic
 
Le web marketing ou comment bien communiquer sur la toile
Le web marketing ou comment bien communiquer sur la toileLe web marketing ou comment bien communiquer sur la toile
Le web marketing ou comment bien communiquer sur la toile
 
Ressources numériques en bibliothèque
Ressources numériques en bibliothèqueRessources numériques en bibliothèque
Ressources numériques en bibliothèque
 

Similar a Hypothesis testing

Ceh v8 labs module 12 hacking webservers
Ceh v8 labs module 12 hacking webserversCeh v8 labs module 12 hacking webservers
Ceh v8 labs module 12 hacking webserversMehrdad Jingoism
 
A03 history of hungary
A03 history of hungaryA03 history of hungary
A03 history of hungaryjavierrm
 
Survey analysis
Survey analysisSurvey analysis
Survey analysisAlexSexton
 
Startup Weekend Education Delhi April 2012 Facilitator's Deck
Startup Weekend Education Delhi April 2012 Facilitator's DeckStartup Weekend Education Delhi April 2012 Facilitator's Deck
Startup Weekend Education Delhi April 2012 Facilitator's DeckNikhil Wason
 
All guidance live.pdf. try it >>> https://bit.ly/3HEXGsi
All guidance live.pdf.    try it >>>  https://bit.ly/3HEXGsi All guidance live.pdf.    try it >>>  https://bit.ly/3HEXGsi
All guidance live.pdf. try it >>> https://bit.ly/3HEXGsi Bossmancyfer
 
All guidance live.pdf
All guidance live.pdf All guidance live.pdf
All guidance live.pdf Bossmancyfer
 
Lesson outline the 21 demands
Lesson outline the 21 demandsLesson outline the 21 demands
Lesson outline the 21 demandsRyan Campbell
 
TAGS: Transparent Armored Gunshields
TAGS: Transparent Armored GunshieldsTAGS: Transparent Armored Gunshields
TAGS: Transparent Armored Gunshields1st_TSG_Airborne
 
Tulsi Gabbard FEC complaint Mufi Hannemann
Tulsi Gabbard FEC complaint Mufi HannemannTulsi Gabbard FEC complaint Mufi Hannemann
Tulsi Gabbard FEC complaint Mufi HannemannHonolulu Civil Beat
 
Mso excel 2003 tips & tricks
Mso excel 2003   tips & tricksMso excel 2003   tips & tricks
Mso excel 2003 tips & tricksSkender Beu
 
The1101 experiment handbook 2020
The1101 experiment handbook 2020The1101 experiment handbook 2020
The1101 experiment handbook 2020Paul MacFarlane
 
Efaw presentation slideshare version
Efaw presentation slideshare versionEfaw presentation slideshare version
Efaw presentation slideshare versionAidTrain
 

Similar a Hypothesis testing (20)

Ceh v8 labs module 12 hacking webservers
Ceh v8 labs module 12 hacking webserversCeh v8 labs module 12 hacking webservers
Ceh v8 labs module 12 hacking webservers
 
A03 history of hungary
A03 history of hungaryA03 history of hungary
A03 history of hungary
 
Over view of disabilities
Over view of disabilitiesOver view of disabilities
Over view of disabilities
 
Legalism
LegalismLegalism
Legalism
 
HLABC Forum: January 2008
HLABC Forum: January 2008HLABC Forum: January 2008
HLABC Forum: January 2008
 
Cs9700 software upgrade notice 3 07
Cs9700 software upgrade notice 3 07Cs9700 software upgrade notice 3 07
Cs9700 software upgrade notice 3 07
 
Survey analysis
Survey analysisSurvey analysis
Survey analysis
 
Be stylishsalon
Be stylishsalonBe stylishsalon
Be stylishsalon
 
Startup Weekend Education Delhi April 2012 Facilitator's Deck
Startup Weekend Education Delhi April 2012 Facilitator's DeckStartup Weekend Education Delhi April 2012 Facilitator's Deck
Startup Weekend Education Delhi April 2012 Facilitator's Deck
 
All guidance live.pdf. try it >>> https://bit.ly/3HEXGsi
All guidance live.pdf.    try it >>>  https://bit.ly/3HEXGsi All guidance live.pdf.    try it >>>  https://bit.ly/3HEXGsi
All guidance live.pdf. try it >>> https://bit.ly/3HEXGsi
 
All guidance live.pdf
All guidance live.pdf All guidance live.pdf
All guidance live.pdf
 
Lesson outline the 21 demands
Lesson outline the 21 demandsLesson outline the 21 demands
Lesson outline the 21 demands
 
Film plot
Film plotFilm plot
Film plot
 
TAGS: Transparent Armored Gunshields
TAGS: Transparent Armored GunshieldsTAGS: Transparent Armored Gunshields
TAGS: Transparent Armored Gunshields
 
Diabetes
DiabetesDiabetes
Diabetes
 
Tulsi Gabbard FEC complaint Mufi Hannemann
Tulsi Gabbard FEC complaint Mufi HannemannTulsi Gabbard FEC complaint Mufi Hannemann
Tulsi Gabbard FEC complaint Mufi Hannemann
 
Aboriton
AboritonAboriton
Aboriton
 
Mso excel 2003 tips & tricks
Mso excel 2003   tips & tricksMso excel 2003   tips & tricks
Mso excel 2003 tips & tricks
 
The1101 experiment handbook 2020
The1101 experiment handbook 2020The1101 experiment handbook 2020
The1101 experiment handbook 2020
 
Efaw presentation slideshare version
Efaw presentation slideshare versionEfaw presentation slideshare version
Efaw presentation slideshare version
 

Último

16042024_First India Newspaper Jaipur.pdf
16042024_First India Newspaper Jaipur.pdf16042024_First India Newspaper Jaipur.pdf
16042024_First India Newspaper Jaipur.pdfFIRST INDIA
 
57 Bidens Annihilation Nation Policy.pdf
57 Bidens Annihilation Nation Policy.pdf57 Bidens Annihilation Nation Policy.pdf
57 Bidens Annihilation Nation Policy.pdfGerald Furnkranz
 
15042024_First India Newspaper Jaipur.pdf
15042024_First India Newspaper Jaipur.pdf15042024_First India Newspaper Jaipur.pdf
15042024_First India Newspaper Jaipur.pdfFIRST INDIA
 
Rohan Jaitley: Central Gov't Standing Counsel for Justice
Rohan Jaitley: Central Gov't Standing Counsel for JusticeRohan Jaitley: Central Gov't Standing Counsel for Justice
Rohan Jaitley: Central Gov't Standing Counsel for JusticeAbdulGhani778830
 
Experience the Future of the Web3 Gaming Trend
Experience the Future of the Web3 Gaming TrendExperience the Future of the Web3 Gaming Trend
Experience the Future of the Web3 Gaming TrendFabwelt
 
complaint-ECI-PM-media-1-Chandru.pdfra;;prfk
complaint-ECI-PM-media-1-Chandru.pdfra;;prfkcomplaint-ECI-PM-media-1-Chandru.pdfra;;prfk
complaint-ECI-PM-media-1-Chandru.pdfra;;prfkbhavenpr
 
Global Terrorism and its types and prevention ppt.
Global Terrorism and its types and prevention ppt.Global Terrorism and its types and prevention ppt.
Global Terrorism and its types and prevention ppt.NaveedKhaskheli1
 
IndiaWest: Your Trusted Source for Today's Global News
IndiaWest: Your Trusted Source for Today's Global NewsIndiaWest: Your Trusted Source for Today's Global News
IndiaWest: Your Trusted Source for Today's Global NewsIndiaWest2
 

Último (8)

16042024_First India Newspaper Jaipur.pdf
16042024_First India Newspaper Jaipur.pdf16042024_First India Newspaper Jaipur.pdf
16042024_First India Newspaper Jaipur.pdf
 
57 Bidens Annihilation Nation Policy.pdf
57 Bidens Annihilation Nation Policy.pdf57 Bidens Annihilation Nation Policy.pdf
57 Bidens Annihilation Nation Policy.pdf
 
15042024_First India Newspaper Jaipur.pdf
15042024_First India Newspaper Jaipur.pdf15042024_First India Newspaper Jaipur.pdf
15042024_First India Newspaper Jaipur.pdf
 
Rohan Jaitley: Central Gov't Standing Counsel for Justice
Rohan Jaitley: Central Gov't Standing Counsel for JusticeRohan Jaitley: Central Gov't Standing Counsel for Justice
Rohan Jaitley: Central Gov't Standing Counsel for Justice
 
Experience the Future of the Web3 Gaming Trend
Experience the Future of the Web3 Gaming TrendExperience the Future of the Web3 Gaming Trend
Experience the Future of the Web3 Gaming Trend
 
complaint-ECI-PM-media-1-Chandru.pdfra;;prfk
complaint-ECI-PM-media-1-Chandru.pdfra;;prfkcomplaint-ECI-PM-media-1-Chandru.pdfra;;prfk
complaint-ECI-PM-media-1-Chandru.pdfra;;prfk
 
Global Terrorism and its types and prevention ppt.
Global Terrorism and its types and prevention ppt.Global Terrorism and its types and prevention ppt.
Global Terrorism and its types and prevention ppt.
 
IndiaWest: Your Trusted Source for Today's Global News
IndiaWest: Your Trusted Source for Today's Global NewsIndiaWest: Your Trusted Source for Today's Global News
IndiaWest: Your Trusted Source for Today's Global News
 

Hypothesis testing

  • 1. H YP O TH E S IS TE S TIN G
  • 2. In th is s e s s ion … .  Wh at is h yp oth e s is te s ting?  Inte rp re ting and s e le cting s ignificance le ve l P R O BABILITY  Typ e I and Typ e II e rrors  O ne taile d and twoTIO NsS D IS TR IBU taile d te ts  H yp oth e s is te s ts for p op u lation m e an  H yp oth e s is te s ts for p op u lation p rop ortion  H yp oth e s is te s ts for p op u lation s tand ard d e viation
  • 3. Wh at is H yp oth e s is Te s ting? H yp oth e s is te s ting re fe rs to • M aking an as s u m p tion, calle d h yp oth e s is , ab ou t a - the B-school p op u lation p aram e te r. • C olle cting s am p le d ata. • C alcu lating a s am p le s tatis tic. • U s ing th e s am p le s tatis tic to e valu ate th e h yp oth e s is (h ow like ly is it th at ou r h yp oth e s ize d p aram e te r is corre ct. To te s t th e valid ity of ou r as s u m p tion we d e te rm ine th e d iffe re nce b e twe e n th e h yp oth e s ize d p aram e te r valu e and th e s am p le valu e .)
  • 4. H YP O T H E S IS T E S T IN G N u ll h yp oth e s is , H 0 Alte rnative h yp oth e s is ,H A S tate th e h yp oth e s ize d valu e of All p os s ib le alte rnative s oth e r th e p aram e te r b e fore s am p ling. th an th e nu ll h yp oth e s is . Th e as s u m p tion we wis h to te s t E .g µ ≠ 20 (or th e as s u m p tion we are trying to re j ct) e µ > 20 E .g p op u lation m e an µ = 20 µ < 20  Th e re is no d iffe re nce b e twe e n Th e re is a d iffe re nce b e twe e n coke and d ie t coke coke and d ie t coke
  • 5. N u ll H yp oth e s is Th e nu ll h yp oth e s is H 0 re p re s e nts a th e ory th at h as b e e n - the B-school p u t forward e ith e r b e cau s e it is b e lie ve d to b e tru e or b e cau s e it is u s e d as a b as is for an argu m e nt and h as not b e e n p rove n. F or e xam p le , in a clinical trial of a ne w d ru g, th e nu ll h yp oth e s is m igh t b e th at th e ne w d ru g is no b e tte r, on ave rage , th an th e cu rre nt d ru g. We wou ld write H 0: th e re is no d iffe re nce b e twe e n th e two d ru gs on an ave rage .
  • 6. Alte rnative H yp oth e s is Th e alte rnative h yp oth e s is , H A, is a s tate m e nt of wh at a s tatis tical h yp oth e s is te s t is s e t u p to e s tab lis h . F or e xam p le , - the B-school in th e clinical trial of a ne w d ru g, th e alte rnative h yp oth e s is m igh t b e th at th e ne w d ru g h as a d iffe re nt e ffe ct, on ave rage , com p are d to th at of th e cu rre nt d ru g. We wou ld write H A: th e two d ru gs h ave d iffe re nt e ffe cts , on ave rage . or H A: th e ne w d ru g is b e tte r th an th e cu rre nt d ru g, on ave rage . Th e re s u lt of a h yp oth e s is te s t: ‘R e j ct H 0 in favou r of H A’ O R ‘D o not re j ct H 0’ e e
  • 7. S e le cting and inte rp re ting s ignificance le ve l • D e cid ing on a crite rion for acce p ting or re j cting th e nu ll e h yp oth e s is . - the B-school • S ignificance le ve l re fe rs to th e p e rce ntage of s am p le m e ans th at is ou ts id e ce rtain p re s crib e d lim its . E .g te s ting a h yp oth e s is at 5% le ve l of s ignificance m e ans  th at we re j ct th e nu ll h yp oth e s is if it falls in th e two re gions e of are a 0.025.  D o not re j ct th e nu ll h yp oth e s is if it falls with in th e re gion of e are a 0.95. 5. Th e h igh e r th e le ve l of s ignificance , th e h igh e r is th e p rob ab ility of re j cting th e nu ll h yp oth e s is wh e n it is tru e . e (acce p tance re gion narrows )
  • 8. Typ e I and Typ e II E rrors • Typ e I e rror re fe rs to th e s itu ation wh e n we re j ct th e nu ll e h yp oth e s is wh e n it is tru e (H 0 is wrongly re j cte d ). e - the B-school e .g H 0: th e re is no d iffe re nce b e twe e n th e two d ru gs on ave rage . Typ e I e rror will occu r if we conclu d e th at th e two d ru gs p rod u ce d iffe re nt e ffe cts wh e n actu ally th e re is n’t a d iffe re nce . P rob (Typ e I e rror) = s ignificance le ve l = α 2. Typ e II e rror re fe rs to th e s itu ation wh e n we acce p t th e nu ll h yp oth e s is wh e n it is fals e . H 0: th e re is no d iffe re nce b e twe e n th e two d ru gs on ave rage . Typ e II e rror will occu r if we conclu d e th at th e two d ru gs p rod u ce th e s am e e ffe ct wh e n actu ally th e re is a d iffe re nce . P rob (Typ e II e rror) = ß
  • 9. Typ e I and Typ e II E rrors – E xam p le You r nu ll h yp oth e s is is th at th e b atte ry for a h e art p ace m ake r h as an ave rage life of 300 d ays , with th e - the B-school alte rnative h yp oth e s is th at th e ave rage life is m ore th an 300 d ays . You are th e qu ality control m anage r for th e b atte ry m anu factu re r. (b)Wou ld you rath e r m ake a Typ e I e rror or a Typ e II e rror? (c)Bas e d on you r ans we r to p art (a), s h ou ld you u s e a h igh or low s ignificance le ve l?
  • 10. Typ e I and Typ e II E rrors – E xam p le G ive n H 0 : ave rage life of p ace m ake r = 300 d ays , and H A: Ave rage life of p ace m ake r > 300 d ays - the eB-school is fals e i.e (b)It is b e tte r to m ake a Typ e II rror (wh e re H 0 ave rage life is actu ally m ore th an 300 d ays b u t we acce p t H 0 and as s u m e th at th e ave rage life is e qu al to 300 d ays ) (c)As we incre as e th e s ignificance le ve l (α) we incre as e th e ch ance s of m aking a typ e I e rror. S ince h e re it is b e tte r to m ake a typ e II e rror we s h all ch oos e a low α.
  • 11. Two Tail Te s t Two taile d te s t will re j ct th e nu ll h yp oth e s is if th e s am p le e m e an is s ignificantly h igh e r or lowe r th an th e h yp oth e s ize d - the B-school m e an. Ap p rop riate wh e n H 0 : µ = µ0 and H A: µ ≠ µ0 e .g Th e m anu factu re r of ligh t b u lb s wants to p rod u ce ligh t b u lb s with a m e an life of 1 000 h ou rs . If th e life tim e is s h orte r h e will los e cu s tom e rs to th e com p e tition and if it is longe r th e n h e will incu r a h igh cos t of p rod u ction. H e d oe s not want to d e viate s ignificantly from 1 000 h ou rs in e ith e r d ire ction. Th u s h e s e le cts th e h yp oth e s e s as H 0 : µ = 1 000 h ou rs and H A: µ ≠ 1 000 h ou rs and u s e s a two tail te s t.
  • 12. O ne Tail Te s t A one -s id e d te s t is a s tatis tical h yp oth e s is te s t in wh ich th e valu e s for wh ich we can re j ct th e nu ll h yp oth e s is , H 0 are e - the B-school locate d e ntire ly in one tail of th e p rob ab ility d is trib u tion. Lowe r taile d te s t will re j ct th e nu ll h yp oth e s is if th e s am p le e m e an is s ignificantly lowe r th an th e h yp oth e s ize d m e an. Ap p rop riate wh e n H 0 : µ = µ0 and H A: µ < µ0 e .g A wh ole s ale r b u ys ligh t b u lb s from th e m anu factu re r in large lots and d e cid e s not to acce p t a lot u nle s s th e m e an life is at le as t 1 000 h ou rs . H 0 : µ = 1 000 h ou rs and H A: µ < 1 000 h ou rs and u s e s a lowe r tail te s t. i.e h e re j cts H 0 only if th e m e an life of s am p le d b u lb s is e s ignificantly b e low 1 000 h ou rs . (h e acce p ts H A and re j cts th e e lot)
  • 13. O ne Tail Te s t U p p e r taile d te s t will re j ct th e nu ll h yp oth e s is if th e s am p le e m e an is s ignificantly h igh e r th an th e h yp oth e s ize d m e an. - the B-school Ap p rop riate wh e n H 0 : µ = µ0 and H A: µ > µ0 e .g A h igh way s afe ty e ngine e r d e cid e s to te s t th e load b e aring cap acity of a 20 ye ar old b rid ge . Th e m inim u m load - b e aring cap acity of th e b rid ge m u s t b e at le as t 1 0 tons . H 0 : µ = 1 0 tons and H A: µ > 1 0 tons and u s e s an u p p e r tail te s t. i.e h e re j cts H 0 only if th e m e an load b e aring cap acity of th e e b rid ge is s ignificantly h igh e r th an 1 0 tons .
  • 14. H yp oth e s is te s t for p op u lation m e an n ( x − µ0 ) H 0 : µ = µ0 and Te s t s tatis tic = ∆ s e - the B-school F or H A: µ > µ0, re j ct H 0 if > t n −1,α ∆ F or H A: µ < µ0, re j ct H 0 if < −t n −1,α e ∆ ∆ > t n −1,α 2 F or H A: µ ≠ µ0, re j ct H 0 if e t n −1,α by zα F or n ≥ 30, re p lace
  • 15. H yp oth e s is te s t for p op u lation m e an A we igh t re d u cing p rogram th at inclu d e s a s trict d ie t and - the B-school e xe rcis e claim s on its online ad ve rtis e m e nt th at it can h e lp an ave rage ove rwe igh t p e rs on los e 1 0 p ou nd s in th re e m onth s . F ollowing th e p rogram ’s m e th od a grou p of twe lve ove rwe igh t p e rs ons h ave los t 8.1 5.7 1 1 .6 1 2.9 3.8 5.9 7.8 9.1 7.0 8.2 9.3 and 8.0 p ou nd s in th re e m onth s . Te s t at 5% le ve l of s ignificance wh e th e r th e p rogram ’s ad ve rtis e m e nt is ove rs tating th e re ality.
  • 16. H yp oth e s is te s t for p op u lation m e an S olu tion: H 0: µ = 1 0 (µ0) H A: µ < 1 0 (µ0) n = 1 2, x(b ar) = 8.027, s = 2.536, α = 0.05 12(8.075 − 10) 3.46 × −1.925 ∆= = = −2.62 2.536 2.536 C ritical t-valu e = -tn-1 ,α= - t1 1 ,0.05 = -2. 201 (TIN V) S ince ∆ < -tn-1 ,α we re j ct H 0 and conclu d e th at th e e p rogram is ove rs tating th e re ality. (Wh at h ap p e ns if we take α = 0.01 ? Is th e p rogram ove rs tating th e re ality at 1 % s ignificance le ve l?)
  • 17. H yp oth e s is te s t for p op u lation p rop ortion n ( p − p0 ) ˆ H 0 : p = p 0 and Te s t s tatis tic = ∆ p0 (1 − p0 ) F or H A: p > p 0 re j ct H e -if∆ > z B-school the α 0 F or H A: p < p 0 re j ct H 0 if∆ < −zα e ∆ > zα 2 F or H A: p ≠ p 0 re j ct H 0 if e
  • 18. H yp oth e s is te s t for p op u lation p rop ortion A ke tch u p m anu factu re r is in th e p roce s s of d e cid ing wh e th e r - the B-school to p rod u ce an e xtra s p icy b rand . Th e com p any’s m arke ting re s e arch d e p artm e nt u s e d a national te le p h one s u rve y of 6000 h ou s e h old s and fou nd th e e xtra s p icy ke tch u p wou ld b e p u rch as e d b y 335 of th e m . A m u ch m ore e xte ns ive s tu d y m ad e two ye ars ago s h owe d th at 5% of th e h ou s e h old s wou ld p u rch as e th e b rand th e n. At a 2% s ignificance le ve l, s h ou ld th e com p any conclu d e th at th e re is an incre as e d inte re s t in th e e xtra-s p icy flavor?
  • 19. H yp oth e s is te s t for p op u lation p rop ortion 335 n = 6000, p= ˆ = 0.05583 6000 - the B-school H0 : p = 0.05( p0 ) H A : p > 0.05 n ( p − p0 ) ˆ 6000 × 0.00583 ∆= = p0 (1 − p0 ) 0.05 × 0.95 77.459 × 0.00583 = = 2.072 0.218 α = 0.02 Zα (the critical value of Z ) = 2.05 (N O R M S IN V Q ∆ > Z we reject H i.e th e cu rre nt inte re s t is s ignificantly gre ate r α 0 ) th an th e inte re s t of two ye ars ago.
  • 20. H yp oth e s is te s t for p op u lation s tand ard d e viation (n − 1)s 2 H 0 : σ = σ 0 and Te s t s tatis tic ∆ = σ 02 - the B-school F or H A: σ > σ0 re j ct H 0 if e ∆ > χ(2(−1),α n R) F or H A: σ < σ0 re j ct H 0 if e ∆ < χ(2(−1),1−α n R) ∆ < χ(2(−1),1−α 2 R) ∆ > χ(2(−1),α 2 R) F or H A: σ ≠ σ0 re j ct H 0 if e n or n
  • 21. H yp oth e s is te s t for com p aring two p op u lation m e ans C ons id e r two p op u lations with m e ans µ1 , µ2 and s tand ard d e viations σ1 and= σ12 .and µ x = µ2 µx µ 1 2 σx am σ x p op u lation2 re s p e ctive ly. - the B-school are th e m e ans of th e sand p ling d is trib u tions of p op u lation1 and 1 2 d e note th e s tand ard e rrors of th e2 σ1 σ 22 µam xp2 ling d is trib u tions of th e m e ans . sx− σ X −X = + 1 1 2 n1 n2 is th e m e an of th e d iffe re nce b e twe e n s am p le m e ans and ( X1 − X ) − ( µ − µ2 )H is th e corre s p ond ing = tand ard 2e rror.1 ∆ s 0 σ x −x H 0 : µ1 = µ2 and te s t s tatis tic, 1 2 H e re ∆ d e note s th e s tand ard ize d F or H A: µ1 > µ2 re j ct H 0 if ∆ > Z α e d iffe re nce of s am p le m e ans F or H A: µ1 < µ2 re j ct H 0 if∆ > - α 2 e ∆ < ZZ α F or H A: µ1 ≠ µ2 re j ct H 0 if e (d e cis ion m ake rs m ay b e conce rne d with p aram e te rs of two p op u lations e .g d o fe m ale e m p loye e s re ce ive lowe r s alary th an th e ir m ale
  • 22. H yp oth e s is te s t for com p aring p op u lation m e ans A s am p le of 32 m one y m arke t m u tu al fu nd s was ch os e n on - the B-school Janu ary 1 , 1 996 and th e ave rage annu al rate of re tu rn ove r th e p as t 30 d ays was fou nd to b e 3.23% and th e s am p le s tand ard d e viation was 0.51 % . A ye ar e arlie r a s am p le of 38 m one y-m arke t fu nd s s h owe d an ave rage rate of re tu rn of 4.36% and th e s am p le s tand ard d e viation was 0.84% . Is it re as onab le to conclu d e (at α = 0.05) th at m one y-m arke t inte re s t rate s d e cline d d u ring 1 995?
  • 23. H yp oth e s is te s t for com p aring p op u lation m e ans n1 = 32, x1 = 3.23, σ 1 = 0.51 n2 = 38, x2 = 4.36, σ 2 = 0.84 H0 : µ1 = µ2 H A : µ1 < µ2 σ X1 − X 2 = σ 12 σ 22 + = 0.26 0.71 + - the B-school = 0.026 = 0.163 n1 n2 32 38 ( x1 − x2 ) − ( µ1 − µ2 )H0 −1.13 − 0 ∆= = = −6.92 σ X1 − X 2 0.163 α = 0.05 Critical value of Z = −Zα = −1.64 Q ∆ < −Zα we reject H0 and conclude that there has been a decline.
  • 24. H yp oth e s is te s t for com p aring p op u lation p rop ortions p1 C ons id e r two s am p le s of s ize s n 1 and n 2 with p2 and as th e re s p e ctive p rop ortions of s u cce s s e s . Th e n p= ˆ n1p1 + n2 p2 n1 + n2 - the B-school is th e e s tim ate d ove rall p rop ortion of s u cce s s e s in th e two p op u lations . ˆˆ ˆˆ pq pq is th e e s tim ate d s tand ard e rror of th e d iffe re nce σ p1 − p2 = ˆ + n1 n2 b e twe e n th e two p rop ortions . ( p1 − p2 ) − ( p1 − p2 )H0 H 0 : p 1 = p 2 and te s t s tatis tic,∆ = σ x1 −x2 ˆ F or H A: p 1 > p 2 re j ct H 0 if ∆ > Z α e A training d ire ctor m ay wis h to F or H A: p 1 < p 2 re j ct H 0 if ∆ < - Z α e d e te rm ine if th e p rop ortion of ∆ > Zα 2 p rom otab le e m p loye e s at one office F or H A: p 1 ≠ p 2 re j ct H 0 if e is d iffe re nt from th at of anoth e r.
  • 25. H yp oth e s is te s t for com p aring p op u lation p rop ortions A large h ote l ch ain is trying to d e cid e wh e th e r to conve rt - the B-school m ore of its room s into non-s m oking room s . In a rand om s am p le of 400 gu e s ts las t ye ar, 1 66 h ad re qu e s te d non- s m oking room s . Th is ye ar 205 gu e s ts in a s am p le of 380 p re fe rre d th e non-s m oking room s . Wou ld you re com m e nd th at th e h ote l ch ain conve rt m ore room s to non-s m oking? S u p p ort you r re com m e nd ation b y te s ting th e ap p rop riate h yp oth e s e s at a 0.01 le ve l of s ignificance .
  • 26. H yp oth e s is te s t for com p aring p op u lation p rop ortions 166 205 n1 = 400, p1 = = 0.415, n2 = 380, p2 = = 0.5395 400 380 H0 : p1 = p2 H A : p1 < p2 p= ˆ = - the B-school n1p1 + n2 p2 400 × 0.415 + 380 × 0.5395 (P rop ortion of s u cce s s = 0.4757 in th e two p op u lations ) n1 + n2 400 + 380 1 1  1 1  σ p1 − p2 = pq  +  = 0.4757 × 0.5243  ˆ ˆˆ +  = 0.0358  n1 n2   400 380  α = 0.01 Th e h ote l ch ain s h ou ld Critical value of Z = −Zα = −2.32 conve rt m ore room s to ( p1 − p2 ) − ( p1 − p2 )H0 −0.1245 − 0 non-s m oking room s as ∆= = = −3.48 th e re h as b e e n a σ p1 − p2 ˆˆ ˆ 0.0358 s ignificant incre as e in th e Q ∆ < −Zα we reject H0 nu m b e r of gu e s ts s e e king non-s m oking room s .