Vous êtes sur la page 1sur 6

Ca/l/lof Co/j c /u c /e _

su l'e c h J
S <G/
+ -
9f l!f!J. f ~ f ~.s fb.5 f L /
sf- > (,I.
l(/e_ c / o 1 1 o f / ''V Q ef I/ ., . t { J e _
fhc ,;f lhe Q///1fox1/J ///C/'eo5e3
/ : srSfc . /J e e fo f,,f e. / 01 5o r; ,
Subject 1 2 3 4 5 6 7 8 9 10 11 12
Difference -0.2 1.3 2.0 2.7 -1.6 0.7 -0.1 -0.5 0.9 0.9 -1.2 0.6
- J
9 I/
I J._
- l o s -(
- 3
G . S - G .5 -?
~
~
/fl e c lt a /J z: 0
~
'
/lJl(,c l/ Q /J > 0
L. (10 points) A biomedical researcher wants to test the effectiveness of a synthetic antitoxin.
Twelve randomly selected subjects are tested for resistance to a particular poison. They are
retested after receiving the antitoxin. Below are the differences in before and after readings
for the subjects. (A positive value indicates that the subject's resistance to the poison has
increased.) Use the Wilcoxon signed-rank test (with a= .05) to determine if the antitoxin
tends to increase the subjects' resistance to the poison. Be sure to state the appropriate
hypotheses and to give your conclusion in words.
Make sure your exam has 6 different pages.
You have two hours to complete the exam.
You may use two formula sheets, a calculator and the statistical tables I posted at the
website. (If a required degrees of freedom is not in the table, use the next smaller
degrees of freedom that is in the table.)
To get full credit show as much of your work as possible.
Good luck and have a great summer!
Instructions:
Final Exam
Name STAT 212-502
May 10, 2010
can t Jof c o4 c _ lu ole //[c ;f
f' . 1 0 Ve c iJJ;_ re/ l/ o/cyies.
l1i 1 1 . J- F ? - - LtJ 7,
c / o ! Jo / F ejecf /{ d J e
0, 71/f
179 ~~0/./7 0~-j - /
e r :, :J o , 0 5 z: Lj, I 7)
)
;i.7. 5~J59-179. ~s-o~7
The data set included n . =35 cases, the SSE for model 1 was 179.45047, and the SSE
for model 2 was 183.58259. Use this information (with a= .05) to test the hypothesis
that , 83 =0 in model 1.
(b) (10 points) Another possible model (model 2) for the heights data is
Y = f3o + f31x1 + f32x2 + E .
(a) (8 points) Of interest is testing the hypothesis Ho : (33 = 0 in model 1. In words, what
does this hypothesis imply about the male and female models?
lle . /Y/G le
f fe . same _
Y = f3o + f31.T1 + f32x2 + (33X1 X2 + E ,
where Y is a student's height, x1 is 1 for female students and 0 for male students, and x2
is the height of a student's same-sex parent. This model implies that the females have the
regression model Y = (30+(31 +(f32+, 83)x2+E , while the males have model Y = f3o +f32x2+t.
2. Consider the following model (call it model 1 ) for some heights data that were collected
in a previous 21 2 class:
2
(a) (8 points) Name the type of design used in this experiment and justify your answer.
Aa1 1 c lo / 1 1 / i'ec / ~lo c k < A~j/ J, / :auf 5r /!ye cf
rS a /J/oci. . ar;c / /1e _ /le afme /Jfs t u e/'e _
('u1 7r/ o m/ y ass51 1 ec / ( (J jl; .re51 1 ec :f lo
or ~e _ I' ) / ~ e oc J 6 loc i : . .
Subject
Noise 1 2 3 4 5 6 7 Totals
None 10 7 14 10 12 1 5 8 76
Pachelbel 10 8 13 8 11 12 9 71
Smashing Pumpkins 14 10 17 12 14 18 6 91
SST= 208.6667
3. A psychologist is studying the effect of background noise on reaction time. E ach subject
is exposed to three levels of background noise: no noise, classical music (Pachelbel's Can o n ),
and rock music (Smashing Pumpkins, Siamese Dream). At each level, the subject has to
perform a variety of tasks, and an overall estimate of his or her reaction time is determined.
The particular ordering of the three noise levels is randomized for each subject. The results
for seven randomly selected subjects follow.
/Jof a /'a /Jc /o /l'l
/c /e /l~f;e i 6! " '-
/S
No . /Je c /Qss
5a7fe fro m
;o Juk/ ;0'7,
(c) (8 points) Using statistical methods we can supposedly make inferences about large
populations based on relatively small samples. Would it be appropriate to draw a
conclusion about the relationship, in general, between heights of children and their
same-sex parents based on the data obtained from this STAT 212 class? Why or why
not?
3
S ;J
(c) (10 points) Use Tukey's procedure with et= 0.05 to compare average reaction times for
the three types of noise. State vour conclusions in words _, --.L IM
( } " 11/c / / 'I. /rt e a?
'\" ' 05 3 /J =-
5. 7
7 o n e. / o , 8"
> > fJac } . 10. I~
IV -= 3. 77 ff = J. NI ' 5P 1 ? >
(b) (10 points) Complete the following analysis of variance table.
Source of Degrees of Sum of Mean
variation freedom squares square F
Noise . ~ 30, 95 c l 38 /~Jl'f 6. 5.lf ~
Subject 0149.3333 ----3> J" f,'l<f /t O.~J...
E rror /<A . J.?,3SIOJ.. ). 3GS-
Total J-0 so E '. 666 7
551/ 7 [(7; - ;.;:)~(J- - Jff /r(ct.f -;;;: /)
: : : : : 3o, 95'J 3~
4
-
where E is a random variable having the normal distribution with mean 0 and variance 14.
(a) (10 points) A park ranger is asked by his boss to give an educated guess for the number
of Chao bo rus larvae that would be found in a sample taken at a location where the lake
is 10 meters Jeep and has a dissolved oxygen reading of 4 milligrams per liter. Help the
ranger out by using the above equation to obtain an interval within which you're 95%
surethelarvalcountis. FA c / ),- 5 a /le / 97,! ) /e _ _ /'Ce /l/;/e s.
Y " - " AJ(/ 5r;,.z (1 0 )- J , L f(? t ) / - 1 ) z : ;t;(/ 7.7",1 1 ')
P( y :> y) =PC l > y~) z z : ' ' l? !J ='>
1- I l/t - =- ! . 90 ~ y z : ).If, 7 3 < ' 3 y 5 y;n tn efo r ' /k _
vfii ),5 j Je /'ce /JPk ;s /0. 0 7
(b) ( 10 points) At a certain spot in the lake the water depth is 5 meters and the dissolved
oxygen count is 2.5 milligrams per liter. What is the probability that the larval count
will be no more than 11 for a sample taken at this point?
~ = /!J-/-/.)(5 )- ~. f~ 5) =:. /~
P( y ~ I () z : P( l ::!:: ' ' ;}! )
P(l- < - - / 07)
Y = 15 + l.2x1 - 2.4x2 + E ,
4. The phantom midge, genus Chao bo rus, resembles the mosquito in appearance but not in
bloodsucking habits. Let Y be the number of Chao bo rus larvae collected in a grab sample
of sediment from the bottom of a lake. Y is related to .T1, the lake depth (in meters) at
the sampling point, and x2, the dissolved oxygen (milligrams per liter) in the water sampled
from the lake bottom, in the following way:
5
/e .l'val 15
where e = the unknown probability that the coin comes up heads on any one toss.
Would your methodology in answering (a) be any different in this case? Why or why
not? (Note: You are not being asked to do any calculations; just describe how your
analysis would differ, if any.)
L I 4 /ou /c / be drffve /ff~ Q wo vlc / have .
lo ;e _ esf/mafec/ f1 o ff! l~e c /afo Q /!q ff e.
;r'" - d ;:;re_ e_ - s o f J;,e.ec lo / 1 7 ! r/ o u/ c / 6 e. B
/fl6fe oc / of L /.
. P(X =k) =8(1 -e )k- 1 , k =1, 2, ... ,
Co / lc / uc / e
e x ; e.r> #Je/ 1 T
I\ a ;;:::_ /3. ) 7 7
1 0/)t_ ;
5f uc / v1 hd tc / 1 1 o f Ca/'/''/ / ;re.
Co //sc/e/J /1ou s ly,
(b) ( 6 points) Suppose all students used the same coin 111 th ell' experiments. To allow for the
possibility that the coin isn't fair, the probability distribution of X (if the experiment
were done correctly) would now be
~ /L /, /3
(a) (10 points) Does it appear that the students carried out the experiment in a conscien-
tious fashion? Test the ap~ropriate hypothesis with ex = .01. ( ), )._
X
~ =- ( ~-'8'7. fi) > t : (60 -4 ~. 75 /_ -;- _ 1- d . /, f 75
' 1 7 . f; ,,Y3, 75' -;)_ /. ' if7 !)
~ ~
- 1 - ( _ ;)-10 .. 9~Z s: L -f {9-!o. 93 75") _ _
/o .'137~ /o.9375
'i 7., /f8. 7G" Ji .8 75' /0 .9375' /0. 9575'
E xpected no. of experiments
Number of experiments
1 2 3 4 2:5
85 6 0 9 12 9
Number of tosses
The results of 175 students' experiments are summarized below.
P(X=k)=.5k, k=l, 2, ....
5. In the 211 class I taught in the 1992 fall semester I asked each of my students to conduct
the following experiment. Toss a fair coin until heads appears, and then stop. Record the
number of tosses that were performed. If X is one student's number of tosses, then X
should have a geometric distribution with parameter .5. In other words, X should have the
probability mass function
6

Vous aimerez peut-être aussi