Business Statistics Communicating with
Numbers 5th Edition Advanced Prep: Master
Statistical Analysis, Probability, and
Inference Practice Questions
Subject: Business Statistics (Chapters 1–20)
Question 1: In the context of descriptive statistics, how does the presence of significant
positive skewness specifically influence the relationship between the mean, median, and
mode, and what are the implications for a business analyst choosing a measure of central
tendency?
A) Mean < Median < Mode; the mean is the most robust measure for skewed financial data.
B) Mode < Median < Mean; the mean is pulled toward the tail, making the median a more
representative measure for typical values.
C) Mode < Mean < Median; the median is heavily influenced by outliers, necessitating the use of
the mean.
D) Mean = Median = Mode; skewness does not affect the choice of central tendency.
Correct Answer: B) Mode < Median < Mean; the mean is pulled toward the tail, making the
median a more representative measure for typical values.
Explanation: In a positively skewed distribution, the long tail extends toward higher values.
Because the mean is calculated using all data points, it is sensitive to these extreme values,
causing it to shift in the direction of the tail. The median, being a positional measure, remains
more centered on the bulk of the data, providing a better 'typical' value for business decision-
making.
Question 2: A market researcher calculates a sample standard deviation ($s$) for a dataset
of customer acquisition costs. If every cost in the dataset is increased by a constant value
$c$, how does the standard deviation change?
A) It increases by $c$.
B) It remains unchanged.
C) It increases by $c^2$.
D) It decreases by $c$.
,Correct Answer: B) It remains unchanged.
Explanation: The standard deviation measures the dispersion or spread of data points around
the mean. Adding a constant to every value in a dataset shifts the entire distribution along the
number line but does not change the distance between individual data points, thus the variance
and standard deviation remain constant.
Question 3: When performing a hypothesis test for the difference between two population
means ($\mu_1 - \mu_2$), what is the primary consequence of failing to assume equal
variances when the populations actually do have equal variances?
A) The test becomes biased and invalid.
B) The test loses power (Type II error rate increases) because the calculation of the pooled
variance is not utilized.
C) The test will incorrectly reject the null hypothesis (Type I error rate increases).
D) There is no consequence; the t-test is perfectly robust to all variance conditions.
Correct Answer: B) The test loses power (Type II error rate increases) because the
calculation of the pooled variance is not utilized.
Explanation: If variances are truly equal, using a pooled variance estimate (standard t-test)
provides a more precise estimate of the standard error than using separate variance estimates
(Welch's t-test). By not pooling when they are equal, you are using a less efficient estimator,
which reduces the test's statistical power.
Question 4: Which of the following best describes the fundamental difference between the
Sampling Distribution of the Sample Mean and the Distribution of the Population?
A) The sampling distribution is always more skewed than the population distribution.
B) The sampling distribution represents the distribution of all possible sample means, and its
standard deviation (standard error) decreases as the sample size $n$ increases.
C) The sampling distribution has the same standard deviation as the population distribution,
regardless of $n$.
D) The population distribution must be normal for the sampling distribution to be approximately
normal.
Correct Answer: B) The sampling distribution represents the distribution of all possible
sample means, and its standard deviation (standard error) decreases as the sample size $n$
increases.
,Explanation: The Central Limit Theorem dictates that as the sample size increases, the standard
error ($\sigma/\sqrt{n}$) decreases, meaning sample means cluster more tightly around the
population mean. This is distinct from the population distribution, which reflects the variability
of individual observations.
Question 5: In a simple linear regression model $Y = \beta_0 + \beta_1X + \epsilon$, what
does the coefficient of determination ($R^2$) actually represent in terms of variance?
A) The ratio of the unexplained variance to the total variance.
B) The proportion of the total variation in the dependent variable $Y$ that is explained by the
independent variable $X$.
C) The correlation between the error term $\epsilon$ and $X$.
D) The percentage of observations that fall exactly on the regression line.
Correct Answer: B) The proportion of the total variation in the dependent variable $Y$ that
is explained by the independent variable $X$.
Explanation: Mathematically, $R^2 = 1 - (SSE/SST)$, where SSE is the sum of squared errors
(unexplained variation) and SST is the total sum of squares. It measures the 'goodness of fit,'
specifically how much of the dependent variable's movement is accounted for by the regression
model.
Question 6: When analyzing time series data, why is the 'Durbin-Watson' statistic used,
and what does a value near 2.0 indicate?
A) It measures heteroscedasticity; 2.0 indicates non-constant variance.
B) It measures autocorrelation in the residuals; 2.0 indicates no evidence of first-order
autocorrelation.
C) It measures the significance of the slope; 2.0 indicates the slope is significant.
D) It measures multicollinearity; 2.0 indicates no collinearity.
Correct Answer: B) It measures autocorrelation in the residuals; 2.0 indicates no evidence of
first-order autocorrelation.
Explanation: In time series regression, residuals are often correlated with their preceding
values. The Durbin-Watson test checks for this. A value of 2.0 suggests no autocorrelation (the
ideal scenario), while values toward 0 indicate positive autocorrelation and values toward 4
indicate negative autocorrelation.
, Question 7: A researcher is testing a hypothesis at the 5% significance level
($\alpha=0.05$). The resulting p-value is 0.048. What is the correct statistical
interpretation?
A) There is strong evidence to reject the null hypothesis; the result is highly significant.
B) We fail to reject the null hypothesis because the p-value is very close to 0.05.
C) We reject the null hypothesis at the 5% level, but we recognize the result is only marginally
significant.
D) The probability that the null hypothesis is true is 4.8%.
Correct Answer: C) We reject the null hypothesis at the 5% level, but we recognize the
result is only marginally significant.
Explanation: Statistical decision-making at a fixed $\alpha$ level is binary (Reject/Fail to
Reject). Since $0.048 < 0.05$, we reject the null. However, because $0.048$ is very close to
$0.05$, a sophisticated analyst notes the marginal nature of the evidence, which is distinct from
"strong" evidence ($p < 0.001$).
Question 8: In multiple regression, what is the specific danger of 'Multicollinearity' and
how does it manifest in the model output?
A) It causes the residuals to be non-normal; it manifests as a low $R^2$.
B) It inflates the standard errors of the coefficient estimates, leading to unstable coefficients and
insignificant t-tests for variables that may actually be important.
C) It causes the model to be biased; it manifests as a very high p-value for the overall F-test.
D) It creates heteroscedasticity; it manifests as a curved pattern in the residual plot.
Correct Answer: B) It inflates the standard errors of the coefficient estimates, leading to
unstable coefficients and insignificant t-tests for variables that may actually be important.
Explanation: Multicollinearity occurs when independent variables are highly correlated. This
makes it mathematically difficult for the regression model to isolate the individual effect of each
variable, leading to high standard errors and wide confidence intervals, even when the model as
a whole has high predictive power.
Question 9: Which of the following conditions is required for the use of the One-Way
ANOVA?
A) The independent variable must be quantitative, and the dependent variable must be
categorical.
Numbers 5th Edition Advanced Prep: Master
Statistical Analysis, Probability, and
Inference Practice Questions
Subject: Business Statistics (Chapters 1–20)
Question 1: In the context of descriptive statistics, how does the presence of significant
positive skewness specifically influence the relationship between the mean, median, and
mode, and what are the implications for a business analyst choosing a measure of central
tendency?
A) Mean < Median < Mode; the mean is the most robust measure for skewed financial data.
B) Mode < Median < Mean; the mean is pulled toward the tail, making the median a more
representative measure for typical values.
C) Mode < Mean < Median; the median is heavily influenced by outliers, necessitating the use of
the mean.
D) Mean = Median = Mode; skewness does not affect the choice of central tendency.
Correct Answer: B) Mode < Median < Mean; the mean is pulled toward the tail, making the
median a more representative measure for typical values.
Explanation: In a positively skewed distribution, the long tail extends toward higher values.
Because the mean is calculated using all data points, it is sensitive to these extreme values,
causing it to shift in the direction of the tail. The median, being a positional measure, remains
more centered on the bulk of the data, providing a better 'typical' value for business decision-
making.
Question 2: A market researcher calculates a sample standard deviation ($s$) for a dataset
of customer acquisition costs. If every cost in the dataset is increased by a constant value
$c$, how does the standard deviation change?
A) It increases by $c$.
B) It remains unchanged.
C) It increases by $c^2$.
D) It decreases by $c$.
,Correct Answer: B) It remains unchanged.
Explanation: The standard deviation measures the dispersion or spread of data points around
the mean. Adding a constant to every value in a dataset shifts the entire distribution along the
number line but does not change the distance between individual data points, thus the variance
and standard deviation remain constant.
Question 3: When performing a hypothesis test for the difference between two population
means ($\mu_1 - \mu_2$), what is the primary consequence of failing to assume equal
variances when the populations actually do have equal variances?
A) The test becomes biased and invalid.
B) The test loses power (Type II error rate increases) because the calculation of the pooled
variance is not utilized.
C) The test will incorrectly reject the null hypothesis (Type I error rate increases).
D) There is no consequence; the t-test is perfectly robust to all variance conditions.
Correct Answer: B) The test loses power (Type II error rate increases) because the
calculation of the pooled variance is not utilized.
Explanation: If variances are truly equal, using a pooled variance estimate (standard t-test)
provides a more precise estimate of the standard error than using separate variance estimates
(Welch's t-test). By not pooling when they are equal, you are using a less efficient estimator,
which reduces the test's statistical power.
Question 4: Which of the following best describes the fundamental difference between the
Sampling Distribution of the Sample Mean and the Distribution of the Population?
A) The sampling distribution is always more skewed than the population distribution.
B) The sampling distribution represents the distribution of all possible sample means, and its
standard deviation (standard error) decreases as the sample size $n$ increases.
C) The sampling distribution has the same standard deviation as the population distribution,
regardless of $n$.
D) The population distribution must be normal for the sampling distribution to be approximately
normal.
Correct Answer: B) The sampling distribution represents the distribution of all possible
sample means, and its standard deviation (standard error) decreases as the sample size $n$
increases.
,Explanation: The Central Limit Theorem dictates that as the sample size increases, the standard
error ($\sigma/\sqrt{n}$) decreases, meaning sample means cluster more tightly around the
population mean. This is distinct from the population distribution, which reflects the variability
of individual observations.
Question 5: In a simple linear regression model $Y = \beta_0 + \beta_1X + \epsilon$, what
does the coefficient of determination ($R^2$) actually represent in terms of variance?
A) The ratio of the unexplained variance to the total variance.
B) The proportion of the total variation in the dependent variable $Y$ that is explained by the
independent variable $X$.
C) The correlation between the error term $\epsilon$ and $X$.
D) The percentage of observations that fall exactly on the regression line.
Correct Answer: B) The proportion of the total variation in the dependent variable $Y$ that
is explained by the independent variable $X$.
Explanation: Mathematically, $R^2 = 1 - (SSE/SST)$, where SSE is the sum of squared errors
(unexplained variation) and SST is the total sum of squares. It measures the 'goodness of fit,'
specifically how much of the dependent variable's movement is accounted for by the regression
model.
Question 6: When analyzing time series data, why is the 'Durbin-Watson' statistic used,
and what does a value near 2.0 indicate?
A) It measures heteroscedasticity; 2.0 indicates non-constant variance.
B) It measures autocorrelation in the residuals; 2.0 indicates no evidence of first-order
autocorrelation.
C) It measures the significance of the slope; 2.0 indicates the slope is significant.
D) It measures multicollinearity; 2.0 indicates no collinearity.
Correct Answer: B) It measures autocorrelation in the residuals; 2.0 indicates no evidence of
first-order autocorrelation.
Explanation: In time series regression, residuals are often correlated with their preceding
values. The Durbin-Watson test checks for this. A value of 2.0 suggests no autocorrelation (the
ideal scenario), while values toward 0 indicate positive autocorrelation and values toward 4
indicate negative autocorrelation.
, Question 7: A researcher is testing a hypothesis at the 5% significance level
($\alpha=0.05$). The resulting p-value is 0.048. What is the correct statistical
interpretation?
A) There is strong evidence to reject the null hypothesis; the result is highly significant.
B) We fail to reject the null hypothesis because the p-value is very close to 0.05.
C) We reject the null hypothesis at the 5% level, but we recognize the result is only marginally
significant.
D) The probability that the null hypothesis is true is 4.8%.
Correct Answer: C) We reject the null hypothesis at the 5% level, but we recognize the
result is only marginally significant.
Explanation: Statistical decision-making at a fixed $\alpha$ level is binary (Reject/Fail to
Reject). Since $0.048 < 0.05$, we reject the null. However, because $0.048$ is very close to
$0.05$, a sophisticated analyst notes the marginal nature of the evidence, which is distinct from
"strong" evidence ($p < 0.001$).
Question 8: In multiple regression, what is the specific danger of 'Multicollinearity' and
how does it manifest in the model output?
A) It causes the residuals to be non-normal; it manifests as a low $R^2$.
B) It inflates the standard errors of the coefficient estimates, leading to unstable coefficients and
insignificant t-tests for variables that may actually be important.
C) It causes the model to be biased; it manifests as a very high p-value for the overall F-test.
D) It creates heteroscedasticity; it manifests as a curved pattern in the residual plot.
Correct Answer: B) It inflates the standard errors of the coefficient estimates, leading to
unstable coefficients and insignificant t-tests for variables that may actually be important.
Explanation: Multicollinearity occurs when independent variables are highly correlated. This
makes it mathematically difficult for the regression model to isolate the individual effect of each
variable, leading to high standard errors and wide confidence intervals, even when the model as
a whole has high predictive power.
Question 9: Which of the following conditions is required for the use of the One-Way
ANOVA?
A) The independent variable must be quantitative, and the dependent variable must be
categorical.
