7/21/2019 AP Statistics 13.1. http://slidepdf.com/reader/full/ap-statistics-131 1/22 AP Statistics 13.1 To Conduct a 2 Sample t Test
7/21/2019 AP Statistics 13.1.
http://slidepdf.com/reader/full/ap-statistics-131 1/22
AP Statistics 13.1
To Conduct a 2 Sample t Test
7/21/2019 AP Statistics 13.1.
http://slidepdf.com/reader/full/ap-statistics-131 2/22
Two Sample Problems
• The goal of this inference is tocompare the responses to 2treatments or to compare thecharacteristics of 2 populations.
• We have a separate sample fromeach treatment or each population.
• The responses of each group areindependent of those in the othergroup.
• P. !" 13.3
7/21/2019 AP Statistics 13.1.
http://slidepdf.com/reader/full/ap-statistics-131 3/22
2 sample tests
• 2 Sample tests often are e#perimentswith 2 independent groups $ notmatched pairs.
• Sample si%es do not have to bee&ual'
• T(pical 2 sample problems P. !1e#ample 13.1
7/21/2019 AP Statistics 13.1.
http://slidepdf.com/reader/full/ap-statistics-131 4/22
7/21/2019 AP Statistics 13.1.
http://slidepdf.com/reader/full/ap-statistics-131 5/22
)otation
Parameter Statistics
Population
*ariable
+ean Standard,eviation
SampleSi%e
+ean Standard,eviation
1 #1 -1 1 n1 s1
2 #2 -2 2 n2 s2
1 x
2 x
7/21/2019 AP Statistics 13.1.
http://slidepdf.com/reader/full/ap-statistics-131 6/22
)ull and Alternative/(pothesis• 0n our h(pothesis tests we are
comparing the means of the twogroups
• /o -1 -2 or -1 4 -2 56 there is no
di7erence in their means
• /a -
1 ≠ -
2 or -
1 4 -
2≠ 5
-1 8 -2 or -1 4 -2 8 5
-1 9 -2 or -1 4 -2 9 5 .
7/21/2019 AP Statistics 13.1.
http://slidepdf.com/reader/full/ap-statistics-131 7/22
,oes Calcium :educe ;lood
Pressure<
• A stud( on 21 blac= men wasconducted. 15 of these men wererandoml( assigned to get a calcium
pill and 11 were randoml( assignedto get a placebo.
• The variable of interest was the
decrease in s(stolic blood pressure.• A > value meant a decrease in ;P a
$ value meant an increase in ;P.
7/21/2019 AP Statistics 13.1.
http://slidepdf.com/reader/full/ap-statistics-131 8/22
Test :esults
• ?roup 1 calcium 4@1!1434"1151142
• ?roup 2 placebo 4112414334""24114143
• Summar( Stats
• 0s this outcome good evidence that calcium
decreases ;P in the entire population ofhealth( blac= men more than a placebo does<
?roup Treatment
n s
1 Calcium 15 ".55 !.@3
2 Placebo 11 45.23 ".51
x
7/21/2019 AP Statistics 13.1.
http://slidepdf.com/reader/full/ap-statistics-131 9/22
Bnce Again
• State the null and alternativeh(pothesis
• Chec= the conditions
• S:S
• 0ndependence
•
)ormalit( of two samples how is thishandled<D +ore about this later.
7/21/2019 AP Statistics 13.1.
http://slidepdf.com/reader/full/ap-statistics-131 10/22
7/21/2019 AP Statistics 13.1.
http://slidepdf.com/reader/full/ap-statistics-131 11/22
7/21/2019 AP Statistics 13.1.
http://slidepdf.com/reader/full/ap-statistics-131 12/22
/old on a minute
• ;efore we calculate t here are somefacts about the sampling distribution ofthe di7erence
• The mean of is -1 $ -2 theunbiased estimator.
• The variance of the di7erence of is
• 0f the 2 populations are normal thenis also normal
21 x x −
21 x x −
21 x x −
2
2
2
1
2
1
nn
σ σ +
21 x x −
7/21/2019 AP Statistics 13.1.
http://slidepdf.com/reader/full/ap-statistics-131 13/22
7/21/2019 AP Statistics 13.1.
http://slidepdf.com/reader/full/ap-statistics-131 14/22
;ut do we ever reall( =nowσ
7/21/2019 AP Statistics 13.1.
http://slidepdf.com/reader/full/ap-statistics-131 15/22
B+? how do we calculate ,
of E<
7/21/2019 AP Statistics 13.1.
http://slidepdf.com/reader/full/ap-statistics-131 16/22
:eall( how do we calculate, of E<• Fse the smaller of n1 $ 1D or n2 $ 1D or
• Bh Gust use the calculator'
• :un calculator :un' Eind p value t value
d of f ,:AW A P0CTF:H• Conclusion in conte#t of the problem'
• Create a t interval for the data using a 5I
CJ and interpret in conte#t of the problem• As a h(pothesis test are the results the
same as before< Wh( or wh( not<
7/21/2019 AP Statistics 13.1.
http://slidepdf.com/reader/full/ap-statistics-131 17/22
7/21/2019 AP Statistics 13.1.
http://slidepdf.com/reader/full/ap-statistics-131 18/22
7/21/2019 AP Statistics 13.1.
http://slidepdf.com/reader/full/ap-statistics-131 19/22
:obustness
• Sample si%e strongl( inKuences p valuesof test.
• When the si%es of the 2 samples are
e&ual and the 2 populations beingcompared have distributions with similarshapes then p values are accurate with
n1 n2 ". Small samples• 0n planning a stud( tr( to ma=e n1 n2 .
7/21/2019 AP Statistics 13.1.
http://slidepdf.com/reader/full/ap-statistics-131 20/22
+ore on conditions for tprocedure• Fse t procedures if
• n1 > n2 8 35 no problems
•35 L n1 > n2 L 1" loo= foroutliers and severe s=ewness
• 1" L n1 > n2 be ver( war( of
s=ewness and outliers'
7/21/2019 AP Statistics 13.1.
http://slidepdf.com/reader/full/ap-statistics-131 21/22
7/21/2019 AP Statistics 13.1.
http://slidepdf.com/reader/full/ap-statistics-131 22/22
Hver(one out of the pool
• Briginall( when two populations hadthe same variance (ou could MpoolN oraverage the 2 variables. Advantage
was the degrees of freedom n1 > n2 42 and it made calculations easier.
• ,onOt pool unless (ou are as=ed to doso. 0f (ou pool (ou will have to Gustif(
(our pooling this can be done b(stating the variances appear e&ualD.