Download Midterm Exam - Statistics for Engineering | STAT 4706 and more Exams Statistics in PDF only on Docsity! 3? - gro
MIDTERM EXAM STATISTICS 4706 April 9, 2002
Each problem is given equal weight. Show work for partial credit.
1. & 2. a.) Complete the following ANOVA table based on the regression equation:
—— Y= fot frait Pore + Bst3 + Bats and 20 observations:
/ 5 Source ss-/ DP, MS
Ms, \E
Model 350 | KT MONE BBS 38.25
Error BO | nll) isso) Y/
7 7 '
Corrected Total 500
Answer the following questions: ~
b.) Is the overall regression significant?
c.} Suppose you computed the following sums of squares due to regression:
R(A1, 2, Bs) = 300
R(Bi, 82, Bs) = 250
(Bo, Bs, Bs) = 325
R(A1, Bs, Ba) = 340
Fill in the following “computer output” a
Fe “
Source _DF. Type IM S$_ _F-value_ salue PR >F
ay ' ). 1245
to 3841
3 4 AS .0042
a4 at 0401
d.) If you were using the Backward bh A on Technique, would one of the variables
be removed from the regression equation at this stage? If so, which one and why?
Use a = .25.
Suppose you computed the following sums of squares due to regression:
F(A, Pr) = 100
R(B1, Ba} 150
RBs, Bs) = 330
e.) Using the original model based on 21, £2, #3, and 2, test the hypothesis:
4
IL
Ao: PB. = 0 and By =0 vs.
A: Ho is not true.
(in the presence of z3 and x4)
f.) Use a partial F test to compare, at a = .05, the two models:
Bot Bre, + Bory + Psas + €
Bo + Brt1 + Box, + €
ee
|
3. Consider a one way ANOVA experiment with K = 4 treatments. The observed data
are summarized as follows:
Treatment
a 1 2 3 4
Yi. 10 12 15 di
6 —
SO -Y2)" 60 58 42 60
Y,, = 12, Mm = Ng =Ng3 = Ng = 6.
The corresponding ANOVA table follows:
Source SS” DF MS F.
Model 84 3 24 2.18
Error 220 20 li
Corrected Total 304
a.) Test Hp : are = pty = 0.
b.) Compute $8, for the contrast w = are — Hl.
c.) Do all pairwise comparisons among the treatment means using Duncan’s multiple
range test at a = .05? What can you say about the experimentwise crror rate for
this procedure?