Download 31 Questions with Answers - Statistics for Engineering Applications | STAT 3704 and more Exams Statistics in PDF only on Docsity! Se G may, oer
STAT 3704 Exam 3
180 points possible
The best of your 3 exam scores will be rescaled to 300 points
UNDERGRADUATE HONOR PLEDGE. J have neither given nor received unauthorized
assistance on this exam.
My STAT 3704 ID uumber is: _
MATERIALS PERMITTED. You are allowed to use a calculator. You will find necessary
probability tables attached to the exam.
POINT VALUES. All questions are five points unless otherwise noted.
INSTRUCTIONS. Read the test carefully and raise your hand if you have any questions.
Good luck!
1. Influential points are unique in which direction(s)?
az bey c. and y d. neither
2. Least squares estimation assumes that the residuals follow what distribution?
a, any b. chi-squared c F d. Normal
3. Why do we randomize an experiment?
a. to estimate bias c. to reduce bias
b. to estimate experimental error d. to reduce experimental error
4, Why do we replicate an experiment?
'
a. to estimate bias ¢. to reduce bias
[BDte estimate experimental error d. to reduce experimental error
5. If we have many factors but cannot afford to run a 2* experiment, what type of experiment could
we run instead?
a. a factoria) experiment c. a 3* experiment
b. a fractional factorial experiment d. a response surface experiment
6. Does a design have to have center runs?
a. U0
b. yes
©
10.
ll.
12.
13
14.
15.
. Suppose you have done a factorial experiment. You assumcd that if is safe to use linear interpolation
between the low levels and high levels of all of the factors. What is one thing you can do to find
out whether the interpolation is really adequate?
a. use center points c. randomize the experiment
b. replicate the experiment d. none of the above
. High-order interactions are less important than Jower-order ones. All interactions are less important,
than main effects. Name this principle.
oR
a. effect heredity c. effect sparsity
§ main effects d! none of the above
If we don't know which factors in our experiment are significant, the experiment is likely to include
some insignificant factors. Name this principle.
a. effect heredity c. effect sparsity
b. main effects d. none of the above
If we are fairly certain that some of the factors are insignificant, is replication likely to be absolutely
necessary in order to test the remaining factors?
a. yes b. no
If G, <0 then we say that the response and the regressor are
a. not: associated b. negatively associated c. positively associated
What assumption about the random errors in a regression model can be evaluated by doing a
histogram of the residuals?
a. constant variance b. independence c. normality
Which of the following plots is most useful in residual analysis for checking the assumption of
constant variance in the random errors?
a. histogram of the residuals
b. normal probability plot of the residuals
c. plotting residuals against the order of the observations
@ plotting residuals against the fitted (predicted) values
What. distribution is used to test whether or not all of the individual model coefficients in a
regression are equal to zero?
b. Normal ct d. chi-squared
Parallel lines on an interaction plot from a factorial design indicate that:
a. an interaction is present
b. an interaction is not. present
c. none of the above