Trinomial Distribution
Trinomial Distribution
edu/stat414)
Home>TheTrinomialDistribution
TheTrinomialDistribution
YoumightrecallthatthebinomialdistributiondescribesthebehaviorofadiscreterandomvariableX,whereXisthe
numberofsuccessesinntries,wheneachtryresultsinoneofonlytwopossibleoutcomes.Whathappensifthere
aren'ttwo,butratherthree,possibleoutcomes?That'swhatwe'llexplorehereonthispage,endingupnotwiththe
binomialdistribution,butratherthetrinomialdistribution.Aratherfittingname,Imightsay!
Example
Supposen=20studentsareselectedatrandom:
LetAbetheeventthatarandomlyselectedstudentwenttothefootball
gameonSaturday.Also,letP(A)=0.20=p1,say.
LetBbetheeventthatarandomlyselectedstudentwatchedthe
footballgameonTVonSaturday.LetP(B)=0.50=p2,say.
LetCbetheeventthatarandomlyselectedstudentcompletelyignored
thefootballgameonSaturday.LetP(C)=0.30=1p1p2.
Onepossibleoutcome,then,ofselectingthe20studentsatrandomis:
BBCABBAACABBBCCBCBCB
Thatis,thefirsttwostudentswatchedthegameonTV,thethirdstudentignoredthegame,thefourthstudentwentto
thegame,andsoon.Now,ifweletXdenotethenumberinthesamplewhowenttothefootballgameonSaturday,
letYdenotethenumberinthesamplewhowatchedthefootballgameonTVonSaturday,andletZdenotethe
numberinthesamplewhocompletelyignoredthefootballgame,theninthiscase:
X=4(becausethereare4As)
Y=10(becausethereare10Bs)
Z=20XY=6(andyes,indeed,thereare6Cs)
WhatisthejointprobabilitymassfunctionofXandY?
Solution.
Thisexamplelendsitselftothefollowingformaldefinition.
Definition.Supposewerepeatanexperimentnindependenttimes,witheachexperimentendinginoneofthreemutually
exclusiveandexhaustiveways(success,firstkindoffailure,secondkindoffailure).IfweletXdenotethenumberof
timestheexperimentresultsinasuccess,letYdenotethenumberoftimestheexperimentresultsinafailureofthefirst
kind,andletZdenotethenumberoftimestheexperimentresultsinafailureofthesecondkind,thenthejointprobability
massfunctionofXandYis:
n!
f (x, y) = P (X = x, Y = y) =
nxy
p p (1 p1 p2 )
x!y!(n x y)!
with:
x=0,1,,n
y=0,1,,n
x+yn
Example(continued)
WhatarethemarginalprobabilitymassfunctionsofXandY?AreXandYindependent?ordependent?
Solution.Wecaneasilyjustlumpthetwokindsoffailuresbacktogether,therebygettingthatX,thenumber
ofsuccesses,isabinomialrandomvariablewithparametersnandp1.Thatis:
n!
f (x) =
x!(n x)!
nx
p (1 p 1 )
1
withx=0,1,,n.Similarly,wecanlumpthesuccessesinwiththefailuresofthesecondkind,thereby
gettingthatY,thenumberoffailuresofthefirstkind,isabinomialrandomvariablewithparametersnandp2.
Thatis:
n!
f (y) =
y!(n y)!
ny
p (1 p 2 )
2
withy=0,1,,n.Therefore,XandYmustbedependent,becauseifwemultiplythep.m.f.sofXandY
together,wedon'tgetthetrinomialp.m.f.Thatis,f(x,y)f(x)f(y):
[
n!
x!y!(n x y)!
nxy
p p (1 p 1 p 2 )
] [
n!
x!(n x)!
nx
p (1 p 1 )
1
] [
n!
y!(n y)!
Bytheway,there'salsoanotherwayofarguingthatXandYmustbedependent...becausethejointsupport
ofXandYistriangular!
SourceURL:https://onlinecourses.science.psu.edu/stat414/node/106
ny
p (1 p 2 )