ISYE 6414 - MIDTERM 2 COMPREHENSIVE
EXAM FULL QUESTIONS AND CORRECT
ANSWERS ALREADY PASSED 2026
◍ True or False: It is possible to have a model that fits the data well
but does not explain much of the variability in the response variable.
Answer: True
◍ True or False: ANOVA is the same as a linear regression model
with a qualitative variable as the predictor. Answer: True
◍ True or False: To determine whether a predictor's coefficient (beta)
is statistically significantly different from 0, we can check the p-value
against a significance level. Answer: True
◍ True or False: There are different ways to calculate the residuals
when we run a logistic regression on data that has replications.
Answer: True
◍ True or False: When comparing pairs of means using ANOVA, we
correct for simultaneous inference, resulting in wider confidence
intervals. Answer: True
◍ What are the model assumptions of ANOVA? Answer: Constant
variance, independence, normality
, ◍ True or False: A low p-value for the Deviance Test for a Poisson
model indicates that the model is a good fit. Answer: False. When the
p-value of a Deviance Test for a Poisson model is low, that means the
residuals are large, which indicates a poor fit.
◍ True or False: Simple Linear Regression and ANOVA models have
the same model assumptions. Answer: False. Simple Linear
Regression requires there to be a linear relationship ("Linearity/Mean
Zero Assumption") between the predictor and the outcome variable
(M1T1L5). This assumption is not a part of ANOVA (M1T2L11).
◍ True or False: The essential difference between Standard Linear
Regression with log-transformation and Poisson Regression is that the
former applies the log-transformation on the data whereas the latter
applies the log-transformation to the expectation. Answer: True. For
SLR, we take the expectation of the log (of Y) and the variance of the
log (of Y); for PR, we take the log of the expectation and the log of
the variance.
◍ True or False: Poisson Regression model uses Maximum
Likelihood Estimation to find the exact unbiased estimates. Answer:
False. PR uses a numerical algorithm iteratively, which is why it's
nice to have a computer estimate a PR model for you.
◍ True or False: When we use a 'link function,' we are modifying the
response variable to create a linear relationship between the modified
response variable and the independent variables. Answer: We call any
transformative function applied to the response variable a 'link'
EXAM FULL QUESTIONS AND CORRECT
ANSWERS ALREADY PASSED 2026
◍ True or False: It is possible to have a model that fits the data well
but does not explain much of the variability in the response variable.
Answer: True
◍ True or False: ANOVA is the same as a linear regression model
with a qualitative variable as the predictor. Answer: True
◍ True or False: To determine whether a predictor's coefficient (beta)
is statistically significantly different from 0, we can check the p-value
against a significance level. Answer: True
◍ True or False: There are different ways to calculate the residuals
when we run a logistic regression on data that has replications.
Answer: True
◍ True or False: When comparing pairs of means using ANOVA, we
correct for simultaneous inference, resulting in wider confidence
intervals. Answer: True
◍ What are the model assumptions of ANOVA? Answer: Constant
variance, independence, normality
, ◍ True or False: A low p-value for the Deviance Test for a Poisson
model indicates that the model is a good fit. Answer: False. When the
p-value of a Deviance Test for a Poisson model is low, that means the
residuals are large, which indicates a poor fit.
◍ True or False: Simple Linear Regression and ANOVA models have
the same model assumptions. Answer: False. Simple Linear
Regression requires there to be a linear relationship ("Linearity/Mean
Zero Assumption") between the predictor and the outcome variable
(M1T1L5). This assumption is not a part of ANOVA (M1T2L11).
◍ True or False: The essential difference between Standard Linear
Regression with log-transformation and Poisson Regression is that the
former applies the log-transformation on the data whereas the latter
applies the log-transformation to the expectation. Answer: True. For
SLR, we take the expectation of the log (of Y) and the variance of the
log (of Y); for PR, we take the log of the expectation and the log of
the variance.
◍ True or False: Poisson Regression model uses Maximum
Likelihood Estimation to find the exact unbiased estimates. Answer:
False. PR uses a numerical algorithm iteratively, which is why it's
nice to have a computer estimate a PR model for you.
◍ True or False: When we use a 'link function,' we are modifying the
response variable to create a linear relationship between the modified
response variable and the independent variables. Answer: We call any
transformative function applied to the response variable a 'link'
