STATISTICS 642 - ASSIGNMENT 7

DUE DATE: 8am Central, FRIDAY, July 23, 2021

Name (Typed)

Email Address (Typed)

1
 • Due 8am Central, FRIDAY, July 23, 2021
• Read Handout 9
• Supplemental Reading: Chapters 17 & 18 - Design & ANOVA Book

• Hand in the following Problems:

Problem 1. ( 45 points) A traffic engineering study was designed to evaluate the effects of three types of traffic
signals: pre-timed signals, semi-actuated signals, and fully actuated signals; on traffic delay at intersections. Also,
two methods of measuring traffic delays: point-sample and path-trace were used to estimate stopped time per
vehicle at an intersection. Two intersections were randomly assigned to each of the three types of traffic signals.
Data was collected at each of the six intersections during a rush hour period and a nonrush hour period and
using both methods of measuring traffic delays. The measured traffic delays (in seconds) per vehicle are given
here:
Point-Sample Path-Trace
Signal Intersection Rush Traffic NonRush Traffic Rush Traffic NonRush Traffic
Pretimed 1 61.7 57.4 53.1 36.5
2 35.8 18.5 35.5 15.9
Semi-actuated 3 20.0 24.6 17.0 21.0
4 2.7 3.1 1.5 1.1
Fully Actuated 5 35.7 26.8 35.4 20.7
6 24.3 25.9 27.5 23.3

Use the SAS code ASSIGN7-PROBI-SUMMER2021.sas to assist you in answering the following
questions:

1. Write a model for this study.

2. Provide an AOV table for this study.

3. Provide the Expected Mean Squares for all sources of Variation in the AOV table.

4. What can you conclude at the α = .05 levels about the effect of Type of Traffic Signal, Measuring Method,
and Level of Traffic on the average traffic delay?

5. Provide estimates of all the variance components and their proportions of the total variance in traffic
delay measurements.

Problem II. ( 30 points) An experiment was conducted with three Factors: A at 4 random levels, B at 5 fixed
levels, and C at 3 random levels nested within factor B. There were 6 experimental units randomly assigned to
each of the treatments. The following model was fit to the 360 responses obtained in the experiment:
yijkl = µ + ai + βj + ck(j) + (aβ)ij + (ac)ik(j) + eijkl ,
with i = 1, 2, 3, 4; j = 1, 2, 3, 4, 5; k = 1, 2, 3; l = 1, 2, 3, 4, 5, 6;
where µ and βj are population parameters with β5 = 0; and ai , ck(j) , (aβ)ij , (ac)ik(j) and
2 2 2 2
eijkl are independent rv’s with N (0, σA ), N (0, σC(B) ), N (0, σAB ) , N (0, σAC(B) ), and N (0, σe2 )

distributions, respectively.

2
 Source DF MS Expected Mean Squares .

A 24.5

B 19.7

A×B 8.9

C(B) 7.5

A × C(B) 6.8

Error 5.8

1. Complete the above AOV table for the experiment by filling in the degrees of freedom and Expected Mean
Squares.

2. Test for a significant AB interaction (α = 0.05). Note that the AOV table is providing the MS, not SS
for each source of variation.

3. Test for a significant B main effect (α = 0.05).

4. Compute the variance of the difference in the estimated treatment means for levels 1 and 2 of Factor B:
ȳ.1.. − ȳ.2.. .
• Provide an estimate of this variance and the degrees of freedom of the estimate.

5. Compute the value of Tukey-Kramer HSD with α = .05 that would be used to determine which pairs of
means across the levels of Factor B are different.

Problem III. ( 25 points) For each of the following experiments provide an AOV table with
Source of Variation, DF, and Expected Mean Squares.
1. Cholesterol was measured in the serum samples of five randomly selected patients from a large pool of
patients. Two independent replicate tubes were prepared for each patient for each of four runs on a
spectrophotometer. The objective of the study was to determine whether the relative cholesterol mea-
surements for patients were consistent from run to run in the clinic. The data are mg/dl of cholesterol in
the the replicate samples from each patient on each run.

2. An experiment was run with four factors A, B, C, and D with B nested within A, C nested within B, and
D nested within C. All four factors have randomly selected levels thus producing the model:

yijkl = µ + ai + bj(i) + ck(i,j) + dl(i,j,k) with i = 1, 2, 3, 4; j = 1, 2, 3; k = 1, 2; l = 1, 2, 3

3. An experiment was conducted with four factors, A with 3 fixed levels; B with 2 fixed levels; C nested
within A and B with 6 random levels at each of the 6 levels of A and B; and D with 5 fixed levels. The
experiment was a CRD with 6 replications. The model is given by

yijkl = µ + αi + βj + (αβ)ij + ck(i,j) + δl + (αδ)il + (βδ)jl + (αβδ)ijl + (cδ)lk(i,j) + em(i,j,k,l)

with i = 1, 2, 3; j = 1, 2; k = 1, 2, 3, 4, 5, 6; l = 1, 2, 3, 4, 5; m = 1, 2, 3, 4, 5, 6